Skip to main content

Full text of "Land-Use Modelling in Planning Practice [electronic resource]"

See other formats


The GeoJournal Library 101 




Springer 



Land-Use Modelling in Planning Practice 



The Geojournal Library 



Volume 101 



Managing Editor: 

Daniel Z. Sui, College Station, USA 

Founding Series Editor: 

Wolf Tietze, Helmstedt, Germany 

Editorial Board: Paul Claval, France 

Yehuda Gradus, Israel 

Sam Ock Park, South Korea 

Herman van der Wusten, The Netherlands 



For further volumes: 
http://www.springer.com/series/6007 



Eric Koomen • Judith Borsboom-van Beurden 
Editors 



Land-Use Modelling 
in Planning Practice 



Springer 



Editors 

Eric Koomen Judith Borsboom-van Beurden 

VU University Amsterdam TNO Behavioural and Societal Sciences 

Spatial Economics/SPINlab PO Box 49 , 2600 AA Delft 

FEWEB/RE The Netherlands 

De Boelelaan 1 105 judith.borsboom@tno.nl 

1081 HV Amsterdam 

The Netherlands 

e.koomen@vu.nl 



ISSN 0924-5499 

ISBN 978-94-007-1821-0 e-ISBN 978-94-007-1822-7 

DOI 10.1007/978-94-007-1822-7 

Springer Dordrecht Heidelberg London New York 

Library of Congress Control Number: 201 1935563 

© Springer Science+Business Media B.V. 201 1 

No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by 
any means, electronic, mechanical, photocopying, microfilming, recording or otherwise , without written 
permission from the Publisher, with the exception of any material supplied specifically for the purpose 
of being entered and executed on a computer system, for exclusive use by the purchaser of the work. 

Printed on acid-free paper 



Springer is part of Springer Science+Business Media (www.springer.com) 



Foreword 



A generation or more ago when land use transport models were first being 
developed, the focus was on how different models compared with one another in 
technical and theoretical terms. There was a concern for dynamics, for interaction 
and for calibration and validation, but less so for how such models might be 
interfaced with wider planning processes and the stakeholders that operate such 
systems and are indeed influenced by the plans that emerged from them. The state 
of the art then consisted of comparative studies of cross-sectional comprehensive 
spatial interaction model applications catalogued, for example, in the ISGLUTI 
Project - the International Study Group on Land Use Transport Interaction - and 
reported in the book by Webster, Bly and Paulley (1988). The dominant focus was 
very much in terms of the technical performance of models rather than their use in 
planning or policy-making. 

As our experience of these models grew and evolved, this focus began to shift 
to the context in which models were best used. Onto the agenda came ideas about 
the various tools that had been developed to inform how we might best make good 
plans, and how these could be stitched together into coherent planning methods. 
Planning support systems in analogy to decision support in management were first 
formally suggested over 20 years ago by Britton Harris (1989) in his seminal article 
Beyond geographic information systems: computers and the planning professional 
as a way of bridging the development of computer models and tools with the 
activities of plan-making. Since then, a series of contributions to ways of building 
this bridge have been forged, the most recent being reported by Brail's (2008) in the 
collection of papers in his book Planning Support Systems for Cities and Regions. 
Many of these sketch the wider context and illustrate how a diversity of models and 
methods are coming together to define appropriate forums for dialogues between 
model builders, planners and the wider set of stakeholders involved in policy and its 
implementation. 

So far we do not have a detailed blow by blow account of building and applying 
models as part of planning support systems. Until now that is, because this book 
represents the first such chronology of how a suite of land-use modelling tools 
called LUMOS - Land Use Modelling System - which is centred on the Land 
Use Scanner model with another model Environment Explorer being sometimes 



V 



vi 



Foreword 



used in parallel, is being fashioned to examine a wide array of different planning 
issues ranging from climate change to ways of reducing energy use in transport. 
This book should convince sceptics of the need to use formal tools in a sensitive 
and appropriate manner to explore different urban and regional futures that can best 
address the various grand challenges involving the environment that will dominate 
the next 20 years and beyond. All of the authors writing here provide a splendid 
picture of planning support systems in action, in fact of several variants of a generic 
planning support system fashioned around the various tools and models that have 
been developed by many groups in the Netherlands which are now maintained by 
PBL Netherlands Environmental Assessment Agency. 

This book reflects experience of using these tools over a 15 year period from the 
time when the Land Use Scanner was first developed, through its development to 
finer scale levels of spatial resolution and through its development from an analytical 
tool to one with a direct optimisation capability. The first three chapters in the book 
set the context by describing these models and setting them in the wider context of 
spatial modelling more generally. Koomen, Hilferink and Borsboom-van Beurden 
provide a comprehensive and technically useful description in the first chapter where 
they define the basic structure and purpose of the model as a 'specification of 
regional demand for land, a definition of local suitability, an allocation module', 
and resulting depictions of future land use. 

This introduction is followed by setting the LUMOS-models in an international 
perspective based on a report by Timmermans, Batty, Couclelis and Wegener who 
were involved in developing a critique of the experience in 2007. It might seem a 
little odd that one of these reviewers is writing this foreword but as a group, we had a 
privileged role in learning about the project, and thus I can communicate our feelings 
that this entire effort should be brought to the attention of the wider world of land- 
use and urban modellers as well as planners engaged in the search for good practice 
in the kinds of planning that LUMOS has been used to support. In fact, the LUMOS 
toolbox is unusual in that the models generally operate across several scales from the 
countrywide Netherlands itself down to quite small urban and rural regions. What 
indeed is impressive is the range of applications that are reported here. These pick 
up on significant questions about sprawl, environment, city compaction, climate 
change, and energy reduction in the context of sustainability. The toolbox does not 
quite extend to dealing with demographic factors per se but there are plenty of hooks 
to suggest how these other sectoral models can be plugged into any planning support 
system fashioned on LUMOS principles. 

Before the various contributions move onto applications and extensions of the 
models, van Schrojenstein Lantman, Verburg, Bregt, and Geertman provide an 
interesting and informative review of land-use models ranging from land cover to 
cellular automata and thence to agent based models. They review six generic types 
in more detail including GEOMOD2 which is land cover based, SLEUTH which is 
a cellular automata land development model, UrbanSim which is probably the best 
example of a contemporary land use transport model based on discrete choice theory 
but also embodying fine scale spatial grain with an agent-based focus, IMAGE an 
ecological-environmental framework to explore the long-term dynamics of global 



Foreword 



vii 



change, CORMAS a multi-agent framework simulating natural resources allocation, 
and ILUMASS a micro-simulation model of urban land use. A brief review of the 
modelling process involving calibration and validation is present and this sets the 
scene for many applications in practice . 

Sustainability issues follow focusing on climate change - flooding risks and 
water damage, shortages and salt-water intrusion - biodiversity, accessibility and 
environmental impacts, quality of life, global business issues, and landscape quality. 
Transport modelling and its relation to land-use change are then explored using the 
example of a well-established model in which many features of the land market 
appear. This model which is called Tigris XL is linked to Land Use Scanner through 
the housing market which lies at the heart of linking different kinds of urban 
model. Applications then focus on the potential for resource allocation across the 
Netherlands and its region where the various simulation models are used to look at 
bio energy production and more general regional spatial strategy planning. Many 
important lessons for the use of models in planning support are gleaned from these 
analyses. These lessons have been noted many times but here, they are based on a 
wealth of experience which is only possible when you have had the sort of sustained 
modelling effort that has been characteristic of planning in the Netherlands for the 
last 20-30 years at least. 

Future developments are then charted and it is here that we see how the critical 
mass built up from this experience provides an important guide to how these models 
might be extended and improved. Dekkers and Rietveld begin this process by 
developing a land market basis for Land Use Scanner while Kuijpers-Linde provides 
the wider context of planning support. Last but not least a new market-based land- 
use model is proposed by Borsboom-van Beurden and Zondag. This builds on the 
Tigris XL schema and from this and other contributions below, it clear that the 
general consensus is that all these tools need a stronger economic underpinning for 
the processes of land allocation that they simulate and forecast as well as optimise. 

This book is a timely and important contribution to ways in which we might 
use models in planning, models in practice, and how we might best use them to 
inform the dialogue between professionals and decision-makers. Case studies are 
essential in this but all too often, we do not have enough detail to know how effective 
the models and tools applied have actually been. This book redresses this balance 
for it contains a wealth of experience that is not available anywhere else. What is 
unusual and impressive is the way this experience is being used to improve planning 
support, to reconcile a changing balance between experts, professionals, informed 
lay interests and the public-at-large. Joshua Epstein (2008) in a fascinating essay 
entitled Why Model? makes the point 'The choice, then, is not whether to build 
models; it's whether to build explicit ones. In explicit models, assumptions are laid 
out in detail, so we can study exactly what they entail'. The contributions in this 
book provide this explicitness that Epstein calls for in a way that provides us with 
clear rules of engagement for the use of models in planning. 



London, UK 



Michael Batty 



viii 



Foreword 



References 

Brail, R. K. (Ed.). (2008). Planning support systems for cities and regions. Cambridge, MA: 

Lincoln Institute of Land Policy. 
Epstein, J. M. (2008). Why model? Journal of Artificial Societies and Social Simulation, 11(4), 12. 

Available at http://jasss.soc.surrey.ac.Uk/l l/4/12.html. Retrieved March 201 1 . 
Hams, B. (1989). Beyond geographic information systems: Computers and the planning 

professional. Journal of the American Planning Association, 55, 85-90 
Webster, F. V., Bly, P. H., & Paulley, N. J. (1988). Urban land use and transport interaction: 

Policies and models. Farnborough, Hants: Gower. 



Preface 



Land-use modelling has been firmly established in Dutch planning practice 
in the past 10 years. The Land Use Modelling System (LUMOS) toolbox, 
managed by PBL Netherlands Environmental Assessment Agency, has made this 
development possible. The development of this toolbox started in 1996 and 
builds on the cooperation of public research organisations such as PBL and 
the Agricultural Economics Research Institute (LEI), academic institutes (VU 
University Amsterdam) and commercial IT companies (Geodan and Object Vision). 
The ongoing model development process is unique in the sense that it reflects almost 
15 years of continuous interaction between planners, researchers and IT specialists. 

This book gives an overview of the wealth of recent applications and 
developments of the LUMOS toolbox. It contains contributions of the many partners 
that are active in applying and developing the toolbox and focuses specifically on 
the Land Use Scanner model that was applied in a wide range of policy-related 
studies in the past years. In addition to being employed for trend extrapolation, 
scenario studies and optimisation at the national level, the model has also been 
frequently used at the lower, regional scale level as is demonstrated in the various 
regional cases that are included in the book. Besides these applications, the book 
also considers some of the more theoretical aspects of land-use models and discusses 
various studies preparing the further development of the model. As such, this book 
is a continuation of the previous Dutch Ruimtescanner book published in 2001 that 
described the development and initial applications of Land Use Scanner. 

The current book is aimed at planners and researchers worldwide that are 
interested in the current state of the art of land-use modelling in planning practice. It 
shows which types of applications are possible with current operational instruments 
and discusses possible pathways for further development. The book allows scholars 
and practitioners around the globe to learn from the extensive experience of Dutch 
planners and modellers. This may be particularly interesting since the Netherlands 
have a longstanding experience in this field, which is exemplified by the fact that 
the well-known and often-used CLUE and MOLAND-based models also originate 
here. 

Obviously, the book is only a snapshot of work in progress. It does, for example, 
not document recent work related to climate adaptation that is carried out within the 



IX 



X 



Preface 



Climate changes Spatial Planning and Knowledge for Climate research programs. 
Neither does it pay attention to the many recent land-use models - based on 
Land Use Scanner - that were built in international projects related to, amongst 
others, the catchment areas of the Rivers Elbe, Rhine and Meuse, the region of 
Flanders, Surinam and Honduras. It does also not highlight the pan-European 
EU-ClueScanner model commissioned by the European Commission that is built 
upon the Geo-DMS model framework underlying Land Use Scanner. This new 
model follows the specification of the Dyna-CLUE model and uses a dynamic 
version of the algorithm that is also underlying the new discrete version of Land Use 
Scanner. Publications on these and other new model developments can be found on 
the websites: www.lumos.info and www.feweb.vu.nl/gis/research/lucas. 

The first part of the book discusses the scientific and theoretical aspects of 
applying land-use models. After a concise introduction of the Land Use Scanner 
model in Chapter 1, the evaluation of the two land-use models that comprise the 
LUMOS toolbox (Land Use Scanner and Environment Explorer) by an international 
audit committee in 2007 and their recommendations for improvement of the current 
models are summarised in Chapter 2. Following, Chapter 3 explores the theoretical 
foundation of current land-use models and examines the pros and cons of various 
concepts and methods in land-use modelling. 

Then, Part II discusses a number of applications of Land Use Scanner for 
a wide range of research and policy questions in environment, agriculture and 
spatial planning, and at various scale levels. The Chapters 4 and 5 highlight the 
comprehensive application of Land Use Scanner at the national level for the Second 
Sustainability Outlook on the future of the Netherlands. Chapter 4 introduces this 
study, while Chapter 5 discusses the link to the Tigris XL transport model that 
was realised in order to be able to analyse the joint impact of spatial planning and 
transportation measures. Chapter 6 explores the potential for bio-mass production 
in a regional case-study in the Province of Friesland and its agro-economic benefits. 
Subsequently, the role of different optimisations of land-use patterns and their 
environmental impact in a regional spatial planning process in the Province of 
Overijssel is evaluated in Chapter 7 . Then, Chapter 8 zooms in at the methodological 
aspects of a number of recent applications at the regional scale level and their 
similarities and dissimilarities. 

The final part of the book reports recent research initiatives working towards 
the development of a new land-use model. Chapter 9 describes how information 
on actual land prices can be used to develop a new method for modelling land-use 
transitions in Land Use Scanner. Subsequently, in Chapter 10 the information needs 
of spatial planning, in particular on land-use changes, and the requirements to a 
new model from the perspective of actual policy questions, are considered. Lastly, 
the way forward to a model meeting those requirements, and the various options to 
realise such a model in a cost-efficient way, are outlined in Chapter 1 1 . 

This book would not have been here without the joint efforts of many individuals 
and organisations. We are particularly grateful to the authors who contributed to 
this book and the many people at PBL (notably Bas van Bemmel, Filip de Blois, 
Bart Rijken and Annemieke Righart) who helped with the logistics of production 



Preface 



xi 



including the revision of text and graphics. In addition, we want to thank the Dutch 
National research programme 'Climate Changes Spatial Planning' for sponsoring 
part of the extensive work involved in editing the book. 

We hope that this book provides inspiration to planners worldwide to use 
a modelling approach to better understand the spatial context of their planning 
problems and to suggest potential solutions. A demonstration version of the model 
has therefore been made available on a separate website (www.feweb.vu.nl/gis/ 
landusescanner.htm) to familiarise users with the potential of this kind of tools. 



Amsterdam, The Netherlands 
Delft, The Netherlands 



Eric Koomen 
Judith Borsboom-van Beurden 



Contents 



Part I Introduction 

1 Introducing Land Use Scanner 3 

Eric Koomen, Maarten Hilferink, and Judith Borsboom-van Beurden 

2 Lumos Models from an International Perspective 23 

Harry Timmermans, Michael Batty, Helen Couclelis, 
and Michael Wegener 

3 Core Principles and Concepts in Land-Use Modelling: 

A Literature Review 35 

Jonas van Schrojenstein Lantman, Peter H. Verburg, 
Arnold Bregt, and Stan Geertman 

Part II Practice 

4 A Sustainable Outlook on the Future of The Netherlands 61 

Rienk Kuiper, Marianne Kuijpers-Linde, and Arno Bouwman 

5 Coupling a Detailed Land-Use Model and a Land-Use 

and Transport Interaction Model 79 

Barry Zondag and Karst Geurs 

6 Biomass on Peat Soils? 97 

Tom Kuhlman, Rene Verburg, Janneke van Dijk, 
and Nga Phan-Drost 

7 Simulation of Future Land Use for Developing a Regional 

Spatial Strategy 117 

Arjen Koekoek, Eric Koomen, Willem Loonen, and Egbert Dijk 

8 Lessons Learned from Using Land-Use Simulation 

in Regional Planning 131 

Chris Jacobs, Arno Bouwman, Eric Koomen, 
and Arjen van der Burg 



xiii 



xiv Contents 

Part III Future Developments 

9 Explaining Land-Use Transition in a Segmented Land Market 153 

Jasper Dekkers and Piet Rietveld 

10 A Policy Perspective of the Development of Dutch 

Land-Use Models 177 

Marianne Kuijpers-Linde 

11 Developing a New, Market-Based Land-Use Model 191 

Judith Borsboom-van Beurden and Barry Zondag 

Index 211 



Contributors 



Michael Batty Centre for Advanced Spatial Analysis (CAS A), University College 
London, 1-19 Torrington Place, London WC1E 6BT, UK, m.batty@ucl.ac.uk 

Judith Borsboom-van Beurden TNO Behavioural and Societal Sciences, 
PO Box 49, 2600 AA Delft, The Netherlands, judith.borsboom@tno.nl 

Arno Bouwman PBL Netherlands Environmental Assessment Agency, 
PO Box 303, 3720 AH Bilthoven, The Netherlands, arno.bouwman@pbl.nl 

Arnold Bregt Laboratory of Geo-Information Science and Remote Sensing, 
Wageningen University, PO Box 47, 6700 AA Wageningen, The Netherlands, 
arnold .bregt @ wur.nl 

Arjen van der Burg Ministry of Infrastructure and Environment, PO Box 20901 , 
2500 EX The Hague, The Netherlands, arjen.vanderburg@minvrom.nl 

Helen Couclelis Department of Geography, University of California, Santa 
Barbara, CA 93106, USA, cook@geog.ucsb.edu 

Jasper Dekkers Department of Spatial Economics/SPINlab, VU University 
Amsterdam, De Boelelaan 1105, 1081 HV Amsterdam, The Netherlands, 
j. dekkers @vu .nl 

Janneke van Dijk Agricultural Economics Research Institute (LEI), 

PO Box 29703, 7502 LS The Hague, The Netherlands, janneke.vandijk@ded.de 

Egbert Dijk Province of Overijssel, PO Box 10078, 8000 GB Zwolle, 
The Netherlands, E.Dijk@overijssel.nl 

Stan Geertman Faculty of Geosciences, Utrecht University, PO Box 801 15, 3508 
TC Utrecht, The Netherlands, s.geertman@geo.uu.nl 

Karst Geurs Centre for Transport Studies, University of Twente, PO Box 217, 
7500 AE Enschede, The Netherlands, k.t.geurs@utwente.nl 

Maarten Hilferink Object Vision, c/o VU University Amsterdam, De Boelelaan 
1087, 1081 HV Amsterdam, The Netherlands, mhilferink@objectvision.nl 



XV 



xvi 



Contributors 



Chris Jacobs Department of Spatial Economics/SPINlab, VU University 
Amsterdam, De Boelelaan 1105, 1081 HV Amsterdam, The Netherlands, 
c.g.w.jacobs@vu.nl 

Arjen Koekoek Geodan, President Kennedylaan 1, 1079 MB Amsterdam, 
The Netherlands, arjen.koekoek@geodan.nl 

Eric Koomen Department of Spatial Economics/SPINlab, VU University 
Amsterdam, De Boelelaan 1105, 1081 HV Amsterdam, The Netherlands; Geodan, 
President Kennedylaan 1, 1079 MB Amsterdam, The Netherlands, 
e.koomen@vu.nl 

Tom Kuhlman Agricultural Economics Research Institute (LEI), PO Box 29703, 
7502 LS The Hague, The Netherlands, tom.kuhlman@wur.nl 

Marianne Kuijpers-Linde TNO Urban Development, PO Box 49, 2600 AA 
Delft, The Netherlands, marianne.kuijpers@tno.nl 

Rienk Kuiper PBL Netherlands Environmental Assessment Agency, 

PO Box 30314, 2500 GH The Hague, The Netherlands, rienk.kuiper@pbl.nl 

Jonas van Schrojenstein Lantman Nelen & Schuurmans, PO Box 1219, 3500 
BE Utrecht, The Netherlands, jonas.vanschrojenstein@nelen-schuurmans.nl 

Willem Loonen ProRail, PO Box 2038, Fl .09, 3500 GA Utrecht, 
The Netherlands, willem.loonen@prorail.nl 

Nga Phan-Drost Department of Spatial Economics/SPINlab, VU University 
Amsterdam, Amsterdam, The Netherlands, phandrost@gmail.com 

Piet Rietveld Department of Spatial Economics, VU University Amsterdam, 
De Boelelaan 1105, 1081 HV Amsterdam, The Netherlands, p.rietveld@vu.nl 

Harry Timmermans Urban Planning Group/EIRASS, Eindhoven University 
of Technology, PO Box 513, 5600 MB Eindhoven, The Netherlands, 
h j .p .timmermans @ b wk .tue .nl 

Peter H. Verburg Institute for Environmental Studies, VU University 
Amsterdam, De Boelelaan 1087, 1081 HV Amsterdam, The Netherlands, 
peter, verburg @ ivm. vu .nl 

Rene Verburg Agricultural Economics Research Institute (LEI), PO Box 29703, 
7502 LS The Hague, The Netherlands, rene.verburg@wur.nl 

Michael Wegener Spiekermann & Wegener (S&W) Urban and Regional 
Research, Lindemannstrasse 10, D-441 37 Dortmund, Germany, 
mw @ spiekermann-wegener.de 

Barry Zondag PBL Netherlands Environmental Assessment Agency, PO Box 
30314, 2500 GH The Hague, The Netherlands, barry.zondag@pbl.nl 



Parti 
Introduction 



Chapter 1 

Introducing Land Use Scanner 

Eric Koomen, Maarten Hilferink, and Judith Borsboom-van Beurden 



1.1 Introduction 

The PBL Netherlands Environmental Assessment Agency has a long tradition in 
land-use modelling. Indeed, the PBL has been putting spatially explicit models 
of land -use change into practice for almost 15 years The agency manages the 
Land Use Modelling System (LUMOS) toolbox, which currently consists of 
two well-known models for simulating land-use change: Land Use Scanner and 
Environment Explorer; as well as a set of tools for pre- and post-processing 
of the modelling results, of the latter of which the Map Comparison Kit is an 
example. 

Dealing with urban, natural and agricultural land functions all together, Land 
Use Scanner offers an integrated view of spatial changes in all types of land use. 
Since the development of its first version in 1997, it has been applied in a large 
number of policy -related research projects. These include the simulation of future 
land use following various scenarios (Borsboom-van Beurden, Bakema & Tijbosch, 
2007; Dekkers and Koomen, 2007; Schotten and Heunks, 2001); the evaluation of 
alternatives for a new national airport (Scholten, Van de Velde, Rietveld & Hilferink, 
1999); the preparation of the Fifth National Spatial Strategy (Schotten, Goetgeluk, 
Hilferink, Rietveld & Scholten, 2001); an outlook for the prospects of agricultural 
land use in the Netherlands (Koomen, Kuhlman, Groen & Bouwman, 2005); and 
the potential impact of climate change on land-use patterns (Koomen, Loonen & 
Hilferink, 2008). In addition to these Dutch applications, Land Use Scanner has 
also been used in several European countries (Hartje et al., 2008; Hoymann, 2010; 
Schotten et al., 2001; Wagtendonk, Juliao & Schotten, 2001). For a full account 
of the methodological and technical details of the original model see Hilferink and 
Rietveld (1999). For an extensive overview of all publications related to Land Use 
Scanner, see www.lumos.info and www.feweb.vu.nl/gis. 



E. Koomen (0) 

Department of Spatial Economics/SPINlab, VU University Amsterdam, De Boelelaan 1 105, 
1081 HV Amsterdam, The Netherlands; Geodan, President Kennedylaan 1 , 1079 MB Amsterdam, 
The Netherlands 
e-mail: e.koomen@vu.nl 



E. Koomen, J. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 
Practice, The GeoJournal Library 101, DOI 10.1007/978-94-007-1822-7_l , 
© Springer Science+Business Media B.V. 201 1 



3 



4 



E. Koomen et al. 



A brief overview of the various ways in which land-use models at the PBL 
have contributed to the information needed in the preparation of environmental 
and spatial planning policies is given in Section 1.2. The structure of the 
Land Use Scanner model is briefly discussed in Section 1.3. Finally, several 
implementation issues related to using Land Use Scanner in practice are highlighted 
in Section 1.4. 



1.2 Land Use Scanner in the Context of Dutch Spatial 
Planning and Environmental Policy 

The objective of most Dutch spatial planning-related Land Use Scanner applications 
is to provide probable spatial patterns of land-use change related to predefined 
conditions on demographic and economic scenario assumptions or specific policy 
interventions. Basically, three approaches can be followed: 

1 . elaboration of diverging scenarios; 

2. extrapolation of trends; 

3 . optimisation of land use . 

Depending upon the approach chosen, research or policy questions are translated 
into a regional demand for land, rules for allocation and a GIS-database, as is 
demonstrated in the descriptions of the following three applications. 



1.2.1 Elaboration of Diverging Scenarios 

To date, the majority of the applications of Land Use Scanner on a national level 
followed a scenario-based approach to deal with the uncertainties around future 
spatial developments. These uncertainties are largely determined by demographic 
and economic changes: population and GDP growth, ageing, decreasing household 
size, economic sector shifts and changes in productivity. Another major source of 
uncertainty is government intervention in the spatial domain. By describing a set 
of diverging views on the future - as is common in, for example, the reports of 
the Intergovernmental Panel on Climate Change (IPCC, 2001) - a broad range of 
spatial developments can be simulated, offering an overview of potential land-use 
changes. A scenario will, as such, not necessarily contain the most likely prospects, 
but, taken together, the simulations provide the bandwidth of possible land-use 
changes. In such a study, the individual scenarios should, in fact, not strive to 
be as probable as possible, but should stir the imagination and broaden the view 
on the future. Important elements are: plausible unexpectedness and informational 
vividness (Xiang & Clarke, 2003). An example of such a scenario-based simulation 
of land-use change is the Spatial Impressions project by the PBL (Borsboom-van 
Beurden et al., 2005; 2007). This analysis was performed to evaluate the possible 



1 Introducing Land Use Scanner 



5 



impact on nature and landscape of economic and demographic changes in the 
future, as described in the first Sustainability Outlook study (MNP, 2004). The 
qualitative storylines of the original scenario framework were translated in spatially 
explicit assumptions, regarding the location preferences and future demand of a 
large number of land -use types, by means of expert-workshops and sector specific 
regional models. The results of the study were subsequently used to inform the 
National Parliament. The general public was also informed through, for example, 
publicity in the national media (Schreuder, 2005). The study pointed out that 
increased land use for housing, employment and leisure will contribute to significant 
further urbanisation, especially in the already heavily urbanised western part of 
the Netherlands. This will result in deterioration of nature areas and valuable 
landscapes, the extent depending upon the degree of government protection assumed 
in a scenario (Fig. 1.1). 

Scenario-based, exploratory approaches as taken in the Spatial Impressions 
project follow a what-if approach: they indicate what may happen (/ certain 
conditions occur. This implies that the main task of the land-use model in these 
applications is not so much to create the most probable future land-use pattern, 
but rather to produce outcomes that investigate the range of possible land-use 



Global Market Global Solidarity 




~J\ Very low Situation 2000 0 50 100 km 

| | Low □ Water 1 1 1 

Medium to high Built-up area 

I Very high I Valuable Landscape 



Fig. 1.1 Land use simulated according to the Global Market (left) and Global Solidarity (right) 
scenarios: the intensity of red areas indicates a possible increase in urban pressure; the green areas 
inside the grey contours signify valuable landscapes (Borsboom-van Beurden et al., 2005) 



6 



E. Koomen et al. 



changes. The implementation of policy-specific scenarios that are used to sketch 
potential land-use patterns resulting from anticipated policies is a special type of 
scenario-based land-use simulation. This type of application is especially useful in 
impact assessment studies (see, for example, Chapter 7). 

In practice, the provision of a range of possible scenario-based views of the future 
is often considered confusing by politicians and other decision-makers: they feel 
impelled to prefer a specific scenario, while the range of outcomes was meant, in 
the first place, to provide information on the magnitude of spatial changes and their 
interdependency with specific policies and interventions. Many decision-makers 
feel the need to have a 'business as usual' scenario, which can be considered as 
the most likely scenario. For that reason, more recent policy-oriented studies tend 
to put more emphasis on providing the most probable land-use patterns that reflect 
the extrapolation of current trends and policies. 

1.2.2 Extrapolation of Trends 

An example of extrapolation of trends is the Second Sustainability Outlook study for 
the Netherlands (MNP, 2007; PBL, 2010), which is described in detail in Chapter 4. 
This study analysed whether prevailing sustainability goals were being achieved 
and what policy objectives remained for the future. Social, economic and spatial 
trends that were current at that time were tracked to provide what is referred to 
as the Baseline Scenario. This version only takes into account policies that have 
been approved by the Dutch parliament or European Parliament. It does not include 
policy changes or the introduction of new policies, but assumes a continuation of 
prevailing policies. In this study, based on the Transatlantic Market scenario in 
the study Welfare, Prosperity and Quality of the Living Environment produced by 
the Dutch assessment agencies (CPB, MNP and RPB, 2006), average demographic 
and economic growth was assumed until 2040: more precisely a modest economic 
growth of 1 .7% per year and a population growth to just over 17 million by 2040. 
This Baseline Scenario, representing average spatial pressure, is in line with the 
OECD baseline scenario (Fig. 1.2). 

1.2 3 Optimisation of Land Use 

Land-use modelling can also be applied to optimise land use from an ecological, 
environmental or spatial planning perspective, as is described in Chapter 7 
and elsewhere in the literature (Loonen, Heuberger & Kuijpers-Linde, 2007). 
The Second Sustainability Outlook study also contains several examples of this 
approach: the study optimised land-use patterns according to six different policy 
themes or so-called viewpoints. For each of these themes, land use was simulated in 
such a way that specific, adverse developments were prevented. 

The optimisation of land use for each policy theme started with an inventory 
of the autonomous developments that hamper the realisation of the current policy 



1 Introducing Land Use Scanner 



7 




I I Recreation I I Greenhouses 
Commercial Infrastructure 

Nature I I Water 

I I Arable land 



Fig. 1.2 Land use in base year 2000 (left) and in 2040 (right) according to the Baseline Scenario 
(Source: MNP, 2007) 



objectives. This inventory is based on the trend-based simulation of land use, 
described in the preceding section. For the biodiversity theme, for example, the 
fragmentation of habitats through the construction of infrastructure (e.g. roads, 
railway lines) and the development of sites for housing and business parks are 
likely to lead to a lack of spatial coherence in the National Ecological Network 
and Natura2000 sites. What is more, the presence of agriculture and the lowering 
of the water-table cause pollution and drought, affecting the quality of nature 
areas. From the Robust Nature viewpoint, the projected spatial developments were, 
therefore, optimised according to specific planning objectives with the aim of 
displaying possible alternative land-use configurations that may result from policy 
interventions. The current Natura2000 sites formed the base for the optimisation. 
To avoid negative consequences on these Natura2000 sites and their species, buffer 
zones were designated to neutralise the environmental and hydrological impact of 
agricultural activities nearby. Further, areas with a high biodiversity were added to 
the Natura2000 sites on basis of either the occurrence of threatened and rare species 
or their adjacency to the Natura2000 sites. 

After the optimisation step, it appeared that the total surface area for nature areas 
is about the same as for the Baseline Scenario, but it is much more geographically 
concentrated. As a result, the spatial preconditions for protected animal species 
improved considerably. An additional ecological assessment found that about 25% 
of the species had a better chance of sustainable preservation. Figure 1.3 shows 
the outcomes of the simulation of land use according to the principles of Robust 
Nature. 



8 



E. Koomen et al. 




Fig. 1.3 Optimisation of land use in 2040 according the Baseline Scenario (left) and Robust Nature 
viewpoint (right) 



1 2 A Regional Applications 

From the above, it follows that each approach has its own merits and supports the 
policy-making process in spatial planning in a different way by providing different 
information. But apart from these approaches, another distinction can be made 
relating to the scale of the study. With the exception of the study on the possible 
relocation of Amsterdam Airport (Scholten et al., 1999; Van de Velde et al., 1997) 
and a study for the Province of South-Holland (Borsboom-van Beurden et al., 2007; 
Bouwman, Kuiper & Tijbosch, 2006), all applications of Land Use Scanner up to 
2007 have been performed at the national level. With the recent transfer of many 
responsibilities in spatial planning to the provinces, the need for information to 
support spatial planning at a regional level has increased. Besides, as is shown 
in Chapters 10 and 11, the current policy questions concerning spatial planning, 
environment and sustainability now require much more detailed information than 
was needed at the time Land Use Scanner was developed. This book contains a 
number of recent examples of the successful use of Land Use Scanner at a regional 
level (e.g. Chapters 7 and 8). For those who are interested in the technical aspects 
of the Land Use Scanner model, the general structure of the basic model and later 
versions is described in Section 1 .3 . 



1 Introducing Land Use Scanner 



9 



1.3 Model Structure 

Land Use Scanner is a GIS-based model that simulates future land use through 
the integration of sector-specific inputs from other, dedicated models. The model 
is based on a demand-supply interaction for land, with sectors competing within 
suitability and policy constraints. It uses a comparatively static approach that 
simulates a future state in a limited number of time steps. Recent applications of the 
model simulate land-use patterns in three subsequent time-steps, each comprising 
one or more decades (MNP, 2007), whereas initial applications used only one or 
two time steps. Unlike many other land-use models, the objective of the Land Use 
Scanner is not to forecast the amount of land-use change, but rather to integrate 
and allocate future demand for land provided by different, external sources, such as 
specialised sector-specific models or policy intentions. This is shown in Fig. 1.4, 
which presents the basic structure of the Land Use Scanner model. The main 
components of this structure are discussed in the following subsections. 



1.3.1 Regional Demand and Local Suitability 

The basic structure of the model consists of a specification of regional demand for 
land, a definition of local suitability, an allocation module and resulting depictions 
of future land use. The first two of these components are described below. The two 



Regional demand 



A1 



: 




Local suitability 


Current 
land use 


Physical 
suitability 

Maps J Jp 


Policy maps 


Distance 
relations 




Fig. 1.4 Basic layout of the Land Use Scanner model 



10 



E. Koomen et al. 



different allocation modules that are available in the model to simulate land-use 
patterns are introduced in the following subsections. 

Regional Demand 

External regional projections of the demand for land, which are usually referred 
to as land claims, are used as input for the model. These projections are specific 
for each land-use type and are derived from, for example, sector-specific models on 
housing or agriculture provided by specialised institutes or experts (when it comes 
to functions strongly dependant on policies, such as nature or water management). 
These projections of demand express for each land-use type the additional land 
demand. The total of the additional demand and the present area claimed by each 
land-use function is allocated to individual grid-cells based on the suitability of the 
cell for that particular land use. 

Local Suitability 

The definition of local suitability uses a large number of geo-datasets that refer to 
the following aspects: current land use, physical properties, operative policies and 
market forces . 

Current land use is the starting point in the simulation of future land use. Various 
geo-datasets are used to construct a map of current land use in the base year of 
the simulation. Current land use is an important ingredient in the specification 
of both total regional demand for land and local suitability. For example, new 
housing is often located near to existing housing areas. However, because Land Use 
Scanner also allocates existing land use, current land-use patterns are not necessarily 
preserved in simulations. Transition costs can play an important role here, too, by 
preserving existing land use when that use is economically sound. The advantage 
of this flexibility is that dynamics in current land use can also be simulated, such as 
the conversion of obsolete business parks to new housing areas or the demolition of 
housing in regions with a shrinking population. This flexibility needs to be balanced 
with the geographical inertia that characterises especially the capital-intensive types 
of land use (e.g. urban land, greenhouses) and calls for sound information on the 
aspects that influence transition probability such as demolition costs. To date, this 
remains a relatively under-explored research area. 

The biophysical properties of land (e.g. soil type and groundwater level) 
are especially important for the suitability specification of particular land-use 
types, such as in agriculture, where they directly influence possible yields, or 
for nature management, where they determine the possibilities of realising policy 
aims such as the creation of new wetlands. Biophysical properties are generally 
considered to be less important for urban functions, since the Netherlands has a 
long tradition of manipulating its natural conditions, in particular its hydrological 
conditions . 

Operative policies, on the other hand, help steer Dutch land-use developments 
in many ways, and they are important components in the definition of suitability. 



1 Introducing Land Use Scanner 



11 



The designated zones of the National Ecological Network, where nature will be 
developed, or the municipal zoning plans are examples of spatial policies that 
stimulate the allocation of certain types of land use by enhancing its suitability. 
Conversely, policies can also reduce land suitability, through the definition of 
restrictions as is exemplified by various zoning laws related to, for example, 
groundwater protection and the preservation of landscape values. 

The market forces that steer residential and commercial development, for 
instance, are generally expressed in distance relations to other, nearby land-use 
functions. Especially accessibility aspects such as proximity to railway stations, 
highway exits and airports are considered important factors that influence the 
location preferences of actors who are active in urban development. Other factors 
that reflect location preferences are, for example, the levels of service available from 
urban facilities or the attractiveness of the surrounding landscape. 

The selection of the appropriate factors for all land-use types and their relative 
weighting are crucial steps in the preparation of the allocation of land uses and these 
largely determine the simulation outcomes. The relative weighting of the factors 
that describe the biophysical conditions, market forces and operative policies are 
normally assigned in such a way that they reflect the content of the particular trend, 
scenario or optimisation that is implemented land-use application. 



1.3.2 Continuous Model 

The original version of the Land Use Scanner model had a 500 m resolution with 
heterogeneous cells, each describing the relative proportion of all current land-use 
types. In this form it is referred to as a continuous model, since it uses a continuous 
description of the amount of land that is covered by each type of use in a cell. 
In the past, this approach has also been described as probabilistic, to reflect that 
the outcomes essentially describe the probability that a certain land use will be 
allocated to a specific location. This is different from most land-use models, which 
only describe one, dominant type of land use per cell. 

The original, continuous model employs a logit-type approach, derived from 
discrete choice theory. Nobel prize winner McFadden made important contributions 
to this approach of modelling the choices made by actors between mutually 
exclusive alternatives (McFadden, 1978). In this theory, the probability that an 
individual selects a certain alternative is dependent on the utility of that specific 
alternative in relation to the total utility of all alternatives. This probability is, given 
its definition, expressed as a value between 0 and 1 , although it will never reach 
either of these extremes. When translated into land use, this approach explains the 
probability of a certain type of land use at a certain location, based on the utility of 
that location for that specific type of use, in relation to the total utility of all possible 
uses. 



12 



E. Koomen et al. 



The utility of a location can be interpreted as its suitability for a certain use. 
This suitability is a combination of positive and negative factors that approximate 
benefits and costs. The higher the utility or suitability for a land-use type, the higher 
the probability that the cell will be used for that type of use. Suitability is assessed by 
potential users and can also be interpreted as a bid price. After all, the user deriving 
the highest benefit from a location will offer the highest price. Furthermore, the 
model is constrained by two conditions, namely, the overall demand for each land- 
use function, and the amount of land that is available. By imposing these conditions, 
a doubly constrained logit model is established in which the expected amount of 
land in cell c that will be used for land-use type j is essentially described by the 
formula: 

M cj = aj * b c * e Sc J (1.1) 

in which: 

M c j is the amount of land in cell c expected to be used for land-use type j; 

aj is the demand balancing factor (condition 1) that ensures that the total 

amount of allocated land for land-use type j equals the sector-specific claim; 
b c is the supply balancing factor (condition 2) that ensures that the total amount 

of allocated land in cell c does not exceed the amount of land that is available 

for that particular cell; 
S C j is the suitability of cell c for land-use type j based on its physical properties, 

operative policies and neighbourhood relations. The importance of the 

suitability value can be set by adjusting a scaling parameter. 

The appropriate aj values that meet the demand of all land-use types, are found 
in an iterative process, as is also discussed by (Dekkers & Koomen, 2007). This 
iterative approach simulates, in fact, a bidding process between competing land 
users (or, more precisely, land-use classes). Each use will try to get its total demand 
satisfied, but may be outbid by another category that derives higher benefits from the 
land. Thus, it can be said that the model, in a simplified way, mimics the land market. 
Governmental spatial planning policies that restrict the free functioning of the Dutch 
land market can be included in this process when they are interpreted as being 
either taxes or subsidies that cause an increase or decrease of the local suitability 
values respectively. In fact, the simulation process sort of produces shadow prices 
of land in the cells. This is discussed in more detail in the literature (Koomen & 
Buurman, 2002). 

In reality, the process of allocating use is more complex than this basic 
description suggests. In brief, the most important extensions to the model are: 

• The location of a selected number of land-use types (e.g. infrastructure, water) 
is considered as static and cannot be changed during simulations. Anticipated 
developments in these land-use types (e.g. the construction of a new railway line) 
are supplied exogenously to the simulations; that is they are directly included as 
simulation results and are not the derived from the iterative simulation process; 



1 Introducing Land Use Scanner 



13 



• The land-use claims are specified per region and this regional division may differ 
per land-use type, thus creating a more complex set of demand constraints; 

• Minimum and maximum claims are introduced to make sure that the model is 
able to find a feasible solution. For land-use types with a minimum claim, it is 
possible to allocate more land. With a maximum claim it is possible to allocate 
less land. Maximum claims are essential if the total of all land-use claims exceeds 
the available amount of land; 

• To reflect the fact that urban functions will, in general, outbid other functions 
at locations that are equally well suited for either type of land use, a monetary 
scaling of the suitability maps has recently been introduced (Borsboom-van 
Beurden et al., 2005; Groen, Koomen, Ritsema & Piek, 2004). In this approach, 
the maximum suitability value per land-use type is related to a realistic land price, 
ranging from, for example, 2.5 euros per square metre for nature areas to 35 euros 
per square metre for residential areas. The merits of this approach are currently 
being studied by others (Dekkers, 2005 and Chapter 9 this volume). 

A more extensive mathematical description of the basic model and its extensions 
can be found in the literature (Hilferink & Rietveld, 1999). 

The continuous model directly translates the probability that a cell will be used 
for a certain type of land use into an amount of land. A probability of 0.4 will thus, 
in the case of a 500 m x 500 m grid, translates into 10 ha. This straightforward 
approach is easy to implement and interpret but has the disadvantage of potentially 
providing very small surface areas for many different land-use types in a cell. This 
will occur especially if the suitability maps have little geographical variation in 
their values, a problem that can be solved by making the suitability maps more 
distinctive and pronounced. Another possible solution for this issue is the inclusion 
of a threshold value in the translation of probabilities into surface areas. Allocation 
can then be limited to those types of land use that, for example, have a probability 
of 0.2 or higher. The inclusion of such a threshold value calls for an adjustment 
of the allocation algorithm, to make sure that all land-use claims are met. This 
is feasible, however, and has been applied in the Natuurplangenerator model that 
aims to find an optimal spatial allocation of different types of nature within an area 
(Van Eupen & Nieuwenhuizen, 2002), which is in many ways similar to Land Use 
Scanner. Experience with this threshold value shows that insignificant quantities of 
land use are indeed set to zero, but if the threshold value is increased the model 
will have difficulties finding an optimal solution. This is due to the possibility that 
all probabilities are below the threshold value. Application of a threshold value in 
land-use simulation with Land Use Scanner remains to be tested and is a topic for 
further research. 

For the visualisation of results, the simulation outcomes are normally aggregated 
and simplified in such a way that each cell portrays the single dominant category 
among a number of major categories. This simplification has, however, a substantial 
influence on the apparent results and may lead to a serious over-representation 
of some categories and an under-representation of others. To prevent the above 
mentioned issues, which are related to the translation and visualisation of the 



14 



E. Koomen et al. 



probability-related outcomes, an allocation algorithm was introduced that deals 
with homogenous cells - see the description of the discrete model in the following 
paragraphs . 

1.33 Discrete Model 

A revised version (4.7) of the Land Use Scanner model became available in 2005. 
This new version offered the possibility of using a grid of 100 m x 100 m, covering 
the terrestrial surface of the Netherlands in about 3.3 million cells. This resolution 
comes close to the size of actual building blocks and makes it possible to use 
homogenous cells that only describe the dominant land use. Furthermore, the revised 
version contained a new algorithm, which was developed to restrict calculation time. 
The algorithm finds the optimal allocation of land use for the given specified demand 
and suitability definition. 

This new approach is referred to as the discrete model as it uses a discrete 
description of land use per cell: each cell is assigned only one type of land use from 
the total range of possible land-use types. Nowadays, the Land Use Scanner model 
has a flexible structure that allows for the selection of five different resolutions, 
ranging from 100 to 10,000 m, as well as the choice of using the discrete or 
continuous model, thus providing a total of 10 basic model variations. 

The discrete allocation model allocates equal units of land (cells) to those types 
of land use that have the highest suitability, taking into account regional land- 
use demand. This discrete allocation problem is solved through a form linear 
programming, the solution of which is considered optimal when the sum of all 
suitability values corresponding to the allocated land use is maximal. 

The allocation is subject to the following constraints: 

• the amount of land allocated to a cell cannot be negative; 

• in total, only 1 ha can be allocated to a cell; 

• the total amount of land allocated to a specific land-use type in a region should 
be between the minimum and maximum claim for that region. 

Mathematically the allocation problem can be formulated as: 




(1.2) 



subject to: 



Xcj > 0 for each c and j; 



^2 X c j — 1 for each c; 



Lj r < ^2 Xcj < Hj r for each j and r for which claims are specified; 



c 



1 Introducing Land Use Scanner 



15 



in which: 

Xcj is the amount of land allocated to cell c to be used for land-use type j; 

S C j is the suitability of cell c for land-use type j; 

Lj r is the minimum claim for land-use type j in region r; and 

Hj r is the maximum claim for land-use type j in region r. 

The regions for which the claims are specified may partially overlap, but for 
each land-use type j, a grid cell c can only be related to one pair of minimum and 
maximum claims . Since all of these constraints relate X c] to one minimum claim, one 
maximum claim (which cannot be both binding) and one grid cell with a capacity 
of 1 ha, it follows that if all minimum and maximum claims are integers and that 
feasible solutions exist, the set of optimal solutions is not empty and lies between 
basic solutions in which each X c j is either 0 or 1 ha. 

The problem at hand is comparable to the well-known Hitchcock transportation 
problem that is common in transport-cost minimisation and, more specifically, 
the semi-assignment problem (Schrijver, 2003; Volgenant, 1996). The objective 
of the former problem is to find the optimal distribution in terms of minimised 
distribution costs of units of different homogenous goods from a set of origins 
to a set of destinations under constraints of a limited supply of goods, a fixed 
demand, and fixed transportation costs per unit for each origin - destination 
pair. The semi-assignment problem has the additional characteristic that all origin 
capacities are integers and that the demand for each destination is one unit of a 
specific homogenous good. Both are special cases of linear programming problems. 
The discrete allocation algorithm has two additional characteristics that are not 
incorporated in the mathematical formulation of the classical semi-assignment 
problem: (1) several (partially) overlapping regions are specified for the claims 
(although the regions of claims for the same land-use type may not overlap); and 
(2) it is possible to apply distinct minimum and maximum claims . 

This mathematical problem, with its very large number of variables, calls for 
a specific, efficient algorithm. To improve the efficiency, a scaling procedure is 
applied and, furthermore, a threshold value is used. Scaling means that growing 
samples of cells are used in an iterative optimisation process that has proven 
to be fast (Tokuyama & Nakano, 1995). An optimisation is performed for each 
sample. After each optimisation, the sample is enlarged and the shadow prices 
in the optimisation process are updated in such way that the (downscaled) 
regional constraints continue to be met. To limit the number of alternatives under 
consideration, a threshold value is used: only allocation choices that are potentially 
optimal are placed in the priority queues for each competing claim. An important 
advantage of the algorithm used is that it enables an exact solution to be found with 
a desktop PC (Pentium 1V-2.8 GHz, 1 GB internal memory) within several minutes, 
provided that feasible solutions exist and all suitability maps have been prepared in 
an initial run of the model. Running the model for the first time takes just over an 
hour as all base data layers have to be constructed. These data sets are then stored 
in the application files (in a temporary folder) to speed up further calculation. 



16 



E. Koomen et al. 



The constraints that are applied in the new discrete allocation model are equal 
to the demand and supply balancing factors applied in the original, continuous 
version of the model. In fact, all the extensions to the original continuous model 
related to the fixed location of certain land-use types, the use of regional claims, 
the incorporation of minimum/maximum claims, and the monetary scaling of 
the suitability maps also apply to the discrete model. Similar to the original 
model, the applied optimisation algorithm of the discrete model aims to find 
shadow prices for the regional demand constraints that increase or decrease the 
suitability values, such that the allocation based on the adjusted suitability values 
corresponds to the regional claims. The main difference of the discrete model 
from the continuous model is that each cell only has one land-use type allocated; 
meaning that for each land-use type the share of allocated land is zero or one. 
From a theoretical perspective the models are, however, equivalent if the scaling 
parameter that defines the importance of the suitability values becomes infinitely 
large. In that case the continuous model would also strictly follow the suitability 
definition in the allocation and would produce homogenous cells. This procedure is, 
however, theoretical and cannot be applied in the calculations due to computational 
limitations. A more extensive discussion of the two available algorithms and an 
assessment of their performance is described in a separate report on calibration 
(Loonen & Koomen, 2009). 

1.4 Land Use Scanner in Practice: Implementation Issues 

With the model structure clarified, it is time to discuss the practical aspects of 
land-use modelling. For the PBL it is not so much the outcomes of the land-use 
simulations themselves that are of interest, but much more the meaning of these 
outcomes for assessing the environmental, ecological and spatial impact of land use. 
In fact, Land Use Scanner is part of a larger model chain. Its input is derived from 
sector-specific models that provide the future demand for land, whereas its output 
is used in specialised ecological, environmental or hydrological models to assess 
specific impacts. Land Use Scanner thus bridges these different model components 
(see Fig. 1.5), meaning that pre- and post-processing of data play an important role 
in the entire model chain. This section discusses the five main activities that have to 
be carried out to implement a new application in Land Use Scanner. 

Construction of a Base Map of Land Use and Classification of Land Use 

Firstly, a base map has to be created which contains data for all distinguished 
land-use types. For the Netherlands , this map is usually based on the latest versions 
of the datasets Land Use Statistics from Statistics Netherlands and the National 
Land Use Database (CBS, 2002; Thunissen & De Wit, 2000). The classification 
of land-use types should be in line with the definitions used by the sector models 
that simulate the demand for land. Often this raises questions on the translation of 
activities and objects, such as employment, farms and houses, to land use: are small 



1 Introducing Land Use Scanner 



17 



National scenarios for economy, demography, energy, culture, technology, climate etc., using macro 
economic models, climate models etc. in cooperation with CPB (National Bureau for Economic 
Policy Analysis), KNMI (Royal Netherlands Meteorological Institute, SCP (The Netherlands Institute 
for Social Research). 

1111111111 

PBL (Netherlands Environmental Assessment Agency) 



National Hydrologi- 
cal modeling 
lnstrument(Ministry of 
Transport, Public 
Works and Water 
Management) 



Thematic spatial models: 

• PEARL (demography) 

• HOMERA (housing market) 

• BLM (demand office space) 

• WEBER (labour market) 

• 'Waterplanner' (hydrology) 

• 'Natuurplanner' (nature) 



Integrated models providing 
interaction between sectoral models 
(LUS, LOV, Tigris-XL) 



Transport Research 
Models (Ministry of 
Transport, Public 
Works and Water 
Management) 



Global Agriculture 
Market Model (LEI) 



Air quality, noise 
levels and other 
environmental 
impacts 




Fig. 1.5 The impact assessment modelling chain 



roads, ditches, playgrounds also to be included? How to translate square meters of 
floor space to business parks? For most simulations, between 20 and 35 different 
categories of land use are distinguished. 

Collecting the Demand for Land 

For all land-use types, data on their future demand for land have to be gathered. 
For some types of land use this can be done relatively easily, by deriving data 
from specialised models, such as those for the housing market and employment or 
from agro-economic models. Often, however, definitions, spatial units, explanatory 
theories and underlying assumptions do not match and temporary solutions have 
to be found (Dekkers & Koomen, 2006). This requires quite some pre-processing. 
Furthermore, it is more difficult to find reliable data on the demand for land for 
sectors that are not so market-driven, such as nature, outdoor leisure and water 
management. For these sectors, one often has to fall back on the expected effects 
on future land use of policies, such as the realisation of the National Ecological 
Network or groundwater protections coming from experts. 



18 



E. Koomen et al. 



Collection of Geo-Datasets Representing Specific Aspects of Suitability 

Once all data on the demand for land have been acquired, maps related to specific 
aspects of suitability of grid cells for a particular type of land use need to be 
collected. In the course of time, the model configuration has become more and more 
elaborate and refined as each new study with the model was used to build upon and 
improve its previous configuration. The most recent model configuration contains 
more than 400 geo-datasets, which are regularly updated. An important element is 
the planning of new residential, business and nature development locations, because 
the probability that a certain land-use type will be realised here is very high. 

Set-Up of Allocation Rules and Attachment of Relative Weights 

For all land-use types, allocation rules describing the relevant suitability aspects 
maps and their relative weights are assigned using a scripting language (Data 
Model Server). Weights are attached that correspond with the relative importance 
of a particular aspect for the overall suitability definition for a particular land-use 
type. The more detailed the classification, the more time-consuming this part of the 
modelling is. 

Conversion and Assessment of Model Outcomes 

After land use has been simulated according to these steps, the outcomes can 
be presented as land-use maps . The results can also be used for the calculation of 
land-use based indicators within the Land Use Scanner itself (Bubeck & Koomen, 
2008; Ritsema van Eck & Koomen, 2008). These indicators highlight specific 
aspects of the results, such as locations of changed land use, impacts on natural 
areas and valuable landscapes and various urbanisation processes. To facilitate the 
further exploration of these results and help with their interpretation tools such 
as the Map Comparison Kit (Visser & De Nijs, 2006) exist. In addition, more 
complex indicators related to environmental and spatial quality can be obtained 
by using the outcomes in specialised ecological, environmental or hydrological 
models. Often considerable post-processing is needed to link land-use simulation 
outcomes to these specialised models because, here too, definitions, spatial units, 
theories and assumptions are not harmonised. In 2009, the calculation of flooding 
risks was significantly improved by coupling the DamageScanner model (Van 
der Hoeven, Aerts, Van der Klis & Koomen, 2008) to Land Use Scanner. Even 
so, this end of the model chain needs considerable amelioration. An additional 
option for visualising and interpreting results is the construction of elaborate 
three-dimensional representations of changed land use, an approach that has been 
pursued in different research projects (Borsboom-van Beurden, Van Lammeren & 
Bouwman, 2006; Lloret, Omtzigt, Koomen & De Blois, 2008). 

From this brief overview of implementation issues, it follows that the quality of 
the outcomes is mainly determined by the quality of the entire model chain, in which 
the actual allocation of land use is only one of the many issues. When using Land 
Use Scanner, special attention needs to be paid to the coherence and consistency 
existing between sector specific models and the Land Use Scanner model. 



1 Introducing Land Use Scanner 



19 



References 

Borsboom-van Beurden, J. A. M., Bakema, A., & Tijbosch, H. (2007). A land-use modelling 
system for environmental impact assessment; Recent applications of the LUMOS toolbox. 
Chapter 16. In E. Koomen, J. Stillwell, A. Bakema, & H.J. Scholten (Eds.), Modelling land-use 
change; Progress and applications (pp. 281-296). Dordrecht: Springer. 

Borsboom-van Beurden, J. A. M., Boersma, W. T., Bouwman, A. A., Crommentuijn, L. E. M., 
Dekkers, J. E. C, & Koomen, E. (2005). Ruimtelijke Beelden; Visualisatie van een veranderd 
Nederland in 2030. RIVM report 550016003. Bilthoven: Milieu- en Natuurplanbureau. 

Borsboom-van Beurden, J. A. M., Van Lammeren, R. J. A., & Bouwman, A. A. (2006). Linking 
land use modelling and 3D visualisation: A mission impossible? In J. Van Leeuwen & 
H. Timmermans (Eds.), Innovations in design and decision support systems in architecture 
and urban planning (pp. 85-102). Dordrecht: Springer. 

Bouwman, A. A., Kuiper, R., & Tijbosch, H. (2006). Ruimtelijke beelden voor Zuid-Holland. 
Rapportnummer 500074002.2006. Bilthoven: Milieu- en Natuurplanbureau. 

Bubeck, R, & Koomen, E. (2008). The use of quantitative evaluation measures in land-use change 
projections; An inventory of indicators available in the land use scanner. Spinlab Research 
Memorandum SL-07. Amsterdam: Vrije Universiteit Amsterdam/SPINlab. 

CBS (2002). Productbeschrijving Bestand Bodemgebruik. Voorburg: Centraal Bureau voor de 
Statistiek. 

CPB, MNP and RPB (2006). Welvaart en Leefomgeving. Een scenariostudie voor Nederland 
in 2040. Den Haag: Centraal Planbureau, Milieu- en Natuurplanbureau en Ruimtelijk 
Planbureau. 

Dekkers, J. E. C. (2005). Grondprijzen, geschiktheidkaarten en parameterinstelling in 
de Ruimtescanner. Technisch achtergrondrapport bij Ruimtelijke Beelden. MNP report 
550016005. Bilthoven: Milieu- en Natuurplanbureau. 

Dekkers, J. E. C, & Koomen, E. (2006). De rol van sectorale inputmodellen in 
ruimtegebruiksimulatie ; Onderzoek naar de modellenketen voor de LUMOS toolbox. SPINIab 
research memorandum SL-05. Amsterdam: Vrije Universiteit Amsterdam. 

Dekkers, J. E. C, & Koomen, E. (2007). Land-use simulation for water management: Application 
of the land use scanner model in two large-scale scenario-studies. Chapter 20. In E. Koomen, 
J. Stillwell, A. Bakema, & H. J. Scholten (Eds.), Modelling land-use change; progress and 
applications (pp. 355-373). Dordrecht: Springer. 

Groen, J., Koomen, E., Ritsema, van E. J., & Piek, M. (2004). Scenario's in kaart; model- en 
ontwerpbenaderingen voor toekomstig ruimtegebruik . NAi Uitgevers/Ruimtelijk Planbureau, 
Rotterdam/Den Haag. 

Hartje, V, Ansmann, T., Blazejczak, J., Gomann, H., Gornig, M., Grossman, M., et al. (2008). 
Szenarioanalyse der Regionalisierung der Driving Forces und Pressures des globalen Wandels 
in einem mittleren Flusseinzugsgebiet. Chapter 2. In: Wirkungen des globalen Wandels auf den 
Wasserkreislauf im Elbegebiet - Risiken und Optionen. Schlussbericht zum BMBF-Vorhaben 
GLOWA-Elbe II. 

Hilferink, M., & Rietveld, P. (1999). Land use scanner: An integrated GIS based model for long 
term projections of land use in urban and rural areas. Journal of Geographic Systems, 1(2), 
155-177. 

Hoymann, J. (2010). Spatial allocation of future residential land use in the Elbe river basin. 

Environment and Planning B: Planning and Design, 37(5), 9 1 1-928. 
IPCC (2001). Climate change 2001 : Synthesis report. A contribution of Working Groups I, II and III 

to the Third Assessment report of the Intergovernmental Panel on climate change. Cambridge: 

Cambridge University Press. 
Koomen, E., & Buurman, J. J. G. (2002). Economic theory and land prices in land use modeling. 

In M. Ruiz, M. Gould, & J. Ramon (Eds.), 5th AGILE conference on geographic information 

science proceedings (pp. 265-270). Palma (files Balears), Spain: Universitat de les files 

Balears. 



20 



E. Koomen et al. 



Koomen, E., Kuhlman, T., Groen, J., & Bouwman, A. A. (2005). Simulating the future of 

agricultural land use in The Netherlands. Tijdschrift voor Economische en Sociale Geografie 

(Journal of Economic and Social Geography), 96(2), 218-224. 
Koomen, E., Loonen, W., & Hilferink, M. (2008). Climate-change adaptations in land-use 

planning; A scenario-based approach. In L. Bernard, A. Friis-Christensen, & H. Pundt 

(Eds .) , The European information society; Taking geoinformation science one step further 

(pp. 261-282). Berlin: Springer. 
Lloret, J. R., Omtzigt, N., Koomen, E., & De Blois, F. (2008). 3D visualisations in simulations of 

future land use: Exploring the possibilities of new, standard visualisation tools. International 

Journal of Digital Earth, 1(1), 148-154. 
Loonen, W., Heuberger, P., & Kuijpers-Linde, M. (2007). Spatial optimisation in land-use 

allocation problems. Chapter 9. In E. Koomen, J. Stillwell, A. Bakema, & H. J. Scholten (Eds.), 

Modelling land-use change; progress and applications (pp. 147-165). Dordrecht: Springer. 
Loonen, W., & Koomen, E. (2009). Calibration and validation of the land use scanner 

allocation algorithms. Bilthoven: PBL-report. Netherlands Environmental Assessment 

Agency. 

McFadden, D. (1978). Modelling the choice of residential location. In A. Karlqvist, L. Lundqvist, 

F. Snickars, & J. W. Weibull, (Eds.), Spatial interaction theory and planning models 

(pp. 75-96). Amsterdam: North Holland Publishers. 
MNP (2004). Kwaliteit en toekomst. Verkenning van de duurzaamheid . RIVM report. Bilthoven: 

Milieu- en Natuurplanbureau. 
MNP (2007). Nederland Later; Tweede Duurzaamheidsverkenning deel fysieke leefomgeving 

Nederland. MNP-publicatienr.500 12700 1/2007. Bilthoven: Milieu- en Natuurplanbureau. 
PBL (2010). The Netherlands in the Future. Second Sustainahility Outlook: The physical living 

environment in The Netherlands. Bilthoven: Netherlands Environmental Assessment Agency 

(PBL). 

Ritsema, van E. J., & Koomen, E. (2008). Characterising urban concentration and land-use 

diversity in simulations of future land use. Annals of Regional Science, 42(1), 123-140. 
Scholten, H. J., Van de Velde, R. J., Rietveld, P., & Hilferink, M. (1999). Spatial information 

infrastructure for scenario planning: The development of a land use planner for Holland. In 

J. Stillwell, S. Geertman, & S. Openshaw (Eds.), Geographical information and planning 

(pp. 1 12-134). Berlin/Heidelberg/New York: Springer. 
Schotten, C. G. J., Goetgeluk, R., Hilferink, M., Rietveld, P., & Scholten, H. J. (2001). Residential 

construction, land use and the environment. Simulations for The Netherlands using a GIS-based 

land use model. Environmental Modeling and Assessment, 6, 133-143. 
Schotten, C. G. J., & Heunks, C. (2001). A national planning application of Euroscanner in The 

Netherlands. Chapter 17. In J. C. H. Stillwell & H. J. Scholten (Eds.), Land use simulation for 

Europe (pp. 245-256). Amsterdam: Kluwer Academic Publishers. 
Schotten, C. G. J., Heunks, C, Wagtendonk, A. J., Buurman, J. J. G., de Zeeuw, C. J., Kramer, H., 

Boersma, W. T. (2001). Simulating Europe in the 21th century. NRSP-2 report 00-22. BCRS, 

Delft. 

Schreuder, A. (2005). Doemscenario's over natuurbehoud. NRC-Handelsblad January, 15, p. 2. 
Schrijver, A. (2003). Combinatorial optimization - polyhedra and efficiency. Berlin: Springer. 
Thunissen, H. A. M., & De Wit, A. J. W. (2000). The National land cover database of the 

Netherlands. Amsterdam: ISPRS XXXIII. 
Tokuyama,T., & Nakano, J. (1995). Efficient algorithms for the Hitchcock transportation problem. 

SIAM Journal on Computing, 24(3), 563-578. 
Van de Velde, R. J., Schotten, C. G. J., Van der Waals, J. F. M., Boersma, W. T, Ouwersloot, H., 

& Ransijn, M. (1997). Ruimteclaims en ruimtelijke ontwikkelingen in de zoekgebieden 

voor de toekomstige nationale luchtinfrastructuur (TNLI). Quickscan met de Ruimtescanner. 

RIVM-rapport 71 1901024. RIVM, Bilthoven. 
Van der Hoeven, E., Aerts, J., Van der Klis, H., & Koomen, E. (2008). An integrated discussion 

support system for new Dutch flood risk management strategies. Chapter 8. In S. Geertman & 



1 Introducing Land Use Scanner 



21 



J. C. H. Stillwell (Eds.), Planning support systems: Best practices and new methods (159-174). 
Berlin: Springer. 

Van Eupen, M., & Nieuwenhuizen, W. (2002). NatuurPlanGenerator: Gebnrikershandleiding 

(Versie 1.0). Werkdocument Natuurplanbureau. Wageningen: Alterra. 
Visser, H., & De Nijs, T. (2006). The map comparison kit. Environmental Modelling & Software, 

21 (3), 346-358. 

Volgenant, A. (1996). Linear and semi-assignment problems: A core oriented approach. Computers 

and Operations Research, 23(10), 917-932. 
Wagtendonk, A. J., Juliao, R. P., & Schotten, C. G. J. (2001). A regional planning application of 

Euroscanner in Portugal. Chapter 18. In J. C. H. Stillwell & H. J. Scholten (Eds.), Land use 

simulation for Europe (pp. 257-291). Amsterdam: Kluwer. 
Xiang, W. N., & Clarke, K. C. (2003). The use of scenarios in land-use planning. Environment and 

Planning B, 30,885-909. 



Chapter 2 

Lumos Models from an International 
Perspective 

Harry Timmermans, Michael Batty, Helen Couclelis, 
and Michael Wegener 



2.1 Introduction 

This chapter summarises the main findings of an international scientific audit of the 
Land Use Scanner and Environment Explorer models as described in the report of 
the audit committee (Timmermans, Batty, Couclelis & Wegener, 2007). Both models 
are part of the LUMOS toolbox maintained by the PBL Netherlands Environmental 
Assessment Agency. The audit was based on a selection of documents (reports, 
conference papers and journal articles) related to these models, and a site visit at 
PBL that took place from 8 to 9 January 2007, during which a series of presentations 
about and software demonstrations of the models were given. 

The aim of the audit committee was to evaluate the scientific quality of the 
processes and products of PBL relating to the application of land-use models, 
primarily at the national level. The two models are used to help provide integrated 
environmental policy assessments and exploratory studies for the Netherlands. 
Because PBL is increasingly involved in regional and local land-use assessments, 
the appropriateness of the models for use at lower spatial scales and finer resolutions 
was also considered. In addition, the audit committee was asked to advise on 
improving scientific quality in future model development and use, and on actions to 
be taken to ensure an internationally prominent role for PBL in the use of land-use 
modelling to support environmental policy-making. The auditors also addressed 
PBL's mission and the resources available to it for the fulfilment of that mission. 

More specifically, the audit committee posed the following questions: 

1. Does PBL use the most appropriate model type(s) and suite(s) to produce the 
information needed for assessments and exploratory studies? 

2. Are the resources available to PBL adequate and appropriate for the sort of use 
it makes of land-use models? 



H . Timmermans (El) 

Urban Planning Group/EIRASS, Eindhoven University of Technology, PO Box 513, 5600 MB 

Eindhoven, The Netherlands 

e-mail : h j .p .timmermans @ bwk .tue .nl 

E. Koomen, J. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 23 
Practice, The GeoJournal Library 101, DOI 10.1007/978-94-007-1822-7_2, 
© Springer Science+Business Media B.V. 201 1 



24 



H. Timmermans et al. 



3. Are the results of land-use model applications for environmental impact 
assessment adequately used in policy report making processes inside PBL, such 
as Outlooks and Balances, and in joint projects with other, relevant national and 
international agencies? 

4. Are the results of studies using land-use models reported in a way that 
communication with policy-makers is well targeted and optimised, while 
maintaining an acceptable level of scientific quality? 

5. Can the auditors advise PBL on its data, model and application methods for the 
coming years? 



2.2 Audit Committee's Findings and Recommendations 

2.2.1 Appropriateness of the Models 

Does PBL use the most appropriate model type(s) and related software for pre- and 
postprocessing to produce the information needed for assessments and exploratory 
studies? 

Both Land Use Scanner and Environment Explorer simulate the mechanisms 
that balance changes in the demand for and supply of various types of land use. 
Each model classifies these land uses in slightly different ways, but both are based 
on the same underlying data set. Land Use Scanner is based on principles taken 
from utility and discrete choice theory, whereas Environment Explorer is based on 
constrained cellular automata, reflecting simple decision rules associated with land 
development. 

Land Use Scanner, originally developed in 1997, has made a significant 
contribution to the state of the art of cell-based land-use modelling. The discrete 
choice (logit) model for land-use allocation on which Land Use Scanner is based 
allows users to test different assumptions about the degree of rationality of choice in 
behaviour of actors driving land -use changes, ranging from full rationality through 
bounded rationality to random choice. In addition, a multitude of extensions of 
the basic model, including models of strategic behaviour, are possible. In the 
first, continuous version of the model (1997), land uses could be represented 
simultaneously as a proportion of the total land per land unit. In the second, discrete 
version of the model (2005), only one land-use type is represented per land unit. 
Both versions of the model can process grid cells of 500 m x 500 m and 100 m 
x 100 m. Allocation of land uses to cells in the continuous model occurs using 
bi-proportional adjustment such that demand for land by each land-use type and 
supply of land in each cell are balanced (or remain above or below these constraints) . 
Allocation of land uses to cells in the discrete model occurs by optimisation 
(by finding an optimal solution of the Linear Semi Assignment Problem). The 
calibration of the continuous version of the model was originally done by iterative 
search for best-fit coefficients of the discrete choice model. Recent calibration 
methods for both versions use logistic regression. There was some concern within 



2 Lumos Models from an International Perspective 



25 



the auditing committee as to whether this calibration method is consistent because 
it does not take account of the versions' regional constraints. 

Environment Explorer, also originally developed in 1997 and based on earlier 
work by White and Engelen, is one of the more advanced cellular automata models 
of land-use change used in practice. Environment Explorer is a two-level model: 
at a macro level it allocates demand for land by sector to 40 economic regions 
(COROP regions); and at a micro level it allocates regional demand to 500 m x 
500 m grid cells in which only one land-use type is represented. The micro level of 
Environment Explorer is more comprehensive than that of Land Use Scanner as it is 
linked to a four-step transport model (trip generation, trip distribution, modal split 
and assignment) and an environmental impact model of traffic-generated noise and 
air pollution. It is not clear whether these models are state of the art, i.e. whether 
the transport model is able to model congestion or elastic demand. Land use is 
allocated to each grid cell via transition/decision rules based on intrinsic suitability, 
zoning, accessibility, neighbourhood effects and a random factor. Calibration of the 
micro-model is achieved by a pattern-matching algorithm that iteratively adjusts 
the many coefficients of the micro-model to maximise goodness of fit. The core 
of the model is at the cellular automata level, with the model being extended to 
embrace models at greater spatial scales within a more general process of integrated 
assessment. 

The fact that Land Use Scanner and Environment Explorer are being used by 
PBL in much the same way implies that in actual applications the differences 
between them have become smaller over the years. Both can now use the same 
data and practically the same classification of land uses. They both deal with 
land-use allocation, and in certain cases the main difference between them is just 
the allocation algorithm. Environment Explorer is a quasi-dynamic model in that it 
allocates land use and updates cell suitability in 1-year time steps, whereas Land Use 
Scanner is a comparative static model that allocates land use at one predetermined 
date at a time, or during a time period. Land Use Scanner has the potential advantage 
that the balancing factors could be interpreted, in terms of economic theory, as costs 
and benefits or even prices for land. In addition, the same kind of utility functions 
used in the sectoral models that generate the demand for land use could also be used 
in Land Use Scanner. Land Use Scanner is in this sense more open and extendable, 
which may offer some advantages. However, the actual implementation does not 
seem to acknowledge that utility models for different land uses and situations for 
choice do not, in general, yield comparably scaled utilities. Although Land Use 
Scanner does provide a unified theoretical basis for incorporating knowledge from 
experts or findings from other models, this assumes these can be suitably expressed 
in the same units. This potential problem does not occur in Environment Explorer 
because the algorithm performing the sequential assignment of land use remains 
consistent, if not a little arbitrary, in that no balancing of land uses is invoked. 

In Environment Explorer, a random component is incorporated to account for 
a diversity of land users or non-explained behaviour. For presentation purposes 
several model runs are aggregated before presentation. Land Use Scanner does not 
incorporate a random component. Measures of goodness of fit for both models are 



26 



H. Timmermans et al. 



high but this should not be interpreted as proof that the models are able to make 
reliable micro-level predictions. Rather these measures largely reflect the built-in 
inertia of the models , which maintain existing land uses, and as such they are heavily 
constrained to meet realistic land-use targets. This means that spatial patterns of 
land use changes will highly resemble those of past changes. When the goodness of 
fit of only the changes in land use is measured, the results are less impressive. In 
addition to the method of cell-by-cell comparison used, several other methods exist 
that measure different aspects of goodness of fit. 

Both models assume that the suitability of cells remains constant within a round 
of assignments. Theoretically, however, assigning a land use to a particular cell 
immediately changes the suitability of that cell. Although time steps differ for 
Land Use Scanner and Environment Explorer, this means that suitability values for 
land use types depending upon other types, are not fully representative. Whether 
this is a problem in practice depends on how much real time is represented by 
one iteration of the model. There are examples in the literature of more detailed 
allocation algorithms. 

Overall, the audit committee considers that Land Use Scanner and Environment 
Explorer suit their purpose and PBL's mission, i.e. assessing the likely impact 
of changing demand on aggregate land-use patterns in broad terms, and meeting 
the standards of academic quality and the state of the art in practice. The choice 
between the two models then becomes largely a matter of personal preference. One 
key question in this context is whether one wishes to use Environment Explorer 
for modelling regional and national processes or use more specialised and detailed 
exogenous models. In principle, either model could be used in that way, but Land 
Use Scanner has the potential advantage, at least in principle, of allowing the use 
of the same utility function that is also used in the model that provides the demand 
forecasts. It should be noted, however, that although this serial linking of models 
seems plausible, many experiences in other domains suggest that it may be more 
problematic than one would like to believe. This is especially true if feedback 
between models is important - the functions used in the different models may be 
based on inconsistent assumptions. It is therefore important to consider the effects 
of model chaining and the proliferation of errors in the kinds of model structures that 
are being considered. This problem becomes particularly acute if, for example, the 
models are nested in a series of higher-level models to obtain integrated assessments. 

The audit committee was told that, increasingly, PBL is being confronted 
with policy issues that concern regional and even local scales and that, 
moreover, questions on policy issues were accompanied by related questions about 
management. In general, this means that the models need to be sensitive to a 
wider spectrum of policies and that other variables may become important. The 
committee considers that the current versions of Land Use Scanner and Environment 
Explorer are less appropriate for addressing policy issues below a national scale. 
The reason for this is that the drivers of change in both models are land-use 
cells and not the actual decision-makers (households, firms, developers, etc.), 
who constitute the truly dynamic elements in the system. Many changes with 



2 Lumos Models from an International Perspective 



27 



considerable environmental impacts (e.g. emissions) may take place in a city 
without much change in land-use patterns. In fact, land use shows substantial 
inertia and is one of the slower processes in urban dynamics . The prediction and 
assessment of such smaller-scale impacts would require simulating location choices 
and the space-time behaviour of households, firms and developers, along with the 
resulting changes in the spatial distribution of people and firms and, eventually, 
land-use patterns. A more behavioural orientation of the models would also require 
the introduction of demographic and economic variables, such as socio-economic 
groups, land and housing prices and transport costs. Environment Explorer does 
have a link to a transport model, but to date this is only an aggregate model 
of transport demand. For environmental issues such as exposure to air pollution 
and total energy consumption, a high resolution (in time and space) activity-based 
model of demand for travel and goods transport seems necessary to generate more 
reliable outcomes at the local level, which some policies and European guidelines 
require. Linking together several different models to incorporate these additional 
mechanisms is likely to result in these missing components only being addressed 
in superficial terms. To truly embrace the perspective of socio-economic activity, 
substantial model redesign or the development of a new model is required. 

In principle, both models could be further elaborated along these lines. However, 
because Environment Explorer is already much closer to being an integrated model 
than Land Use Scanner, it seems most efficient to try and expand it further 
and examine whether it would be sufficiently sensitive to meet relevant policy 
requirements. This would involve adding a layer representing household and firm 
behaviour. However, in the long run it may be more effective to develop a completely 
different modelling approach that focuses on actors (people and firms) rather 
than on units of land use. Most integrated land-use transport models simulate the 
behaviour of households and firms, but fully operational models with the same 
spatial resolution as Land Use Scanner and Environment Explorer are still in their 
infancy. High-resolution agent-based models that include an activity-based model 
of transport demand, and which keep track of what activities are conducted where 
and for how long, constitute the frontier of academic work in this field and so far 
only partial results have been obtained. PBL has already experienced some of the 
considerable risks involved in following this approach. 

Recommendations 

In the short term, for the assessment of policies at a national scale, there is 
no major reason to change the current use of Land Use Scanner or Environment 
Explorer. The models may assume slightly different roles, with the former being 
used in combination with national demographic, housing and other models, and 
the latter being used as a desktop planning system. Some minor issues could be 
worked out in more detail - perhaps a task for Masters or PhD students. These 
issues involve making assumptions about socio-economic and political trends more 
explicit or making the models sensitive to a wider range of policies. Combining the 
two models does not deliver any advantages. The two models could, however, be 



28 



H. Timmermans et al. 



used in parallel, with similar data sets, to explore the same policy options: significant 
differences in their outputs should be carefully probed for unexpected explanations. 

In the medium term, PBL should consider developing a model that would be 
more appropriate for assessing policy impacts at regional and local scales. The audit 
committee recommends investigating the possibility of extending Environment 
Explorer's capabilities by adding a layer representing people and firms, introducing 
economics into the model and linking it to an activity-based model of transport 
demand. This would introduce the idea of agents into such a model, although it 
would be aggregated agents associated with population and employment at the cell 
level rather than individual agents . 

In the longer term, a fully-Hedged, integrated agent-based simulation framework 
may be desirable, especially if a more detailed approach is required to support 
regional and perhaps local policy-making. Such as model or framework will be 
fairly complex. Rather than starting from scratch, a definition study should explore 
which elements can be kept from the current models and how various existing 
micro-simulation models of demand for travel and goods transport, housing, firm 
demographics, use of green space, shopping behaviour, leisure activities, office 
development, agriculture, real estate, etc., may be incorporated. Many such models 
have recently been developed in the Netherlands and elsewhere and there is 
now a good opportunity to focus these efforts. In developing such a framework, 
constraints of data availability, calibration methods and computing time should be 
carefully considered. This may lead to a problem-specific, multi-level combination 
of aggregate and disaggregate model components, subject to the caveats regarding 
linked models (omissions, inconsistencies) already expressed above. 

At the other extreme of complexity, PBL should explore the potential utility of 
much simpler, conceptual models, often called sketch planning models. Developing 
such models may serve some purposes better than detailed models, especially 
where broad, uncertain trends are concerned. These models would constitute 
needed links between detailed quantitative models, on the one hand, and the broad 
qualitative scenarios being developed by PBL, on the other. Large ensembles of such 
(meta-)models could be systematically generated, leading to a more scientifically 
grounded development of qualitative scenarios . 

PBL should examine the ways in which different models can best serve the 
process of integrated assessment - both individually or in combination. This would 
involve a review of model chaining, of how models at different spatial scales 
might be interfaced, and how processes operating at different spatial scales can be 
successfully represented within integrated models. 

This is a process that PBL is already engaged in and it involves continued 
vigilance with respect to how these models are integrated with those of other 
activities of PBL, as well as the models of other relevant Dutch agencies and 
research groups. 



2 Lumos Models from an International Perspective 



29 



2.2.2 Resources 

Are the resources available to PBL adequate and appropriate for the sort of use it 
makes of land-use models? 

The audit committee believes that PBL basically has the right combination of 
people and skills to be able to liaise effectively with decision-makers. However, 
because different professional and public groups are interested in different aspects 
of modelling output, the sorts of decision-makers for whom PBL's models may be 
relevant should be further examined. Indeed, PBL's whole interface with clients 
should be examined, as mentioned below under Communication (Section 2.2.4). 

The current research group has the right scientific balance and seems motivated 
to incrementally improve their tools and the process of communication, although 
an expansion of PBL's work into activity-based models, as recommended in 
Section 2.2.1, would require a strengthening of the expertise in that field. Staff are 
up to date with respect to new international developments in their field and national 
and international cooperation is high on their agenda. Furthermore, the national and 
international reputation of PBL has substantially increased of late. 

PBL depends on others for developing LUMOS tools. There is a difference in 
this regard between the two models. Land Use Scanner has been developed in an 
open fashion, as part of a broader consortium following open source practices. It is 
therefore easier for PBL to influence its design. In contrast, Environment Explorer 
depends in part on proprietary software that is also being developed for other 
groups. While cooperation with external entities is both necessary and desirable, 
it is essential that PBL maintains the in-house expertise needed to conduct model 
analyses and make minor improvements to the models as needed. Eventually, a small 
number of PBL staff should have the expertise to develop new models jointly with 
consultants, as opposed to using models developed by others. 

The LUMOS initiative is to be applauded and could be extended beyond 
harmonising the input of models, especially if PBL decides to develop agent-based 
models. It may even be worthwhile to consider LUMOS as a vehicle for enabling 
groups outside PBL to display and compare their models and to use the LUMOS 
website (www.lumos.info) as a forum for increasing national and international 
cooperation. 

Because land-use modelling had recently regained momentum within the 
international research community, it is critically important that PBL staff remain 
at the cutting edge of the field. Joint projects and exchange programmes with 
universities (nationally and internationally), greater participation in international 
conferences and increased interaction with other agencies are examples of measures 
that should be considered for maintaining a high level of skills and expertise 
among staff. In addition, PBL should start positioning its modelling efforts in 
the perspective of the emergent information infrastructure, in particular the spatial 
data infrastructure and related networks of professional scientific activities that are 
interfacing with policy-makers and government. Such activities would also increase 
PBL's visibility as a highly desirable employer, thus attracting and retaining young 
highly-qualified scientists. 



30 



H. Timmermans et al. 



Recommendations 

PBL should continue to develop collaborations in model development and data 
collection with other agencies and research institutions, but at the same time it 
should aim at maintaining a self-contained in-house capacity to conduct expert 
model analyses and make minor improvements to models without having to rely 
on external cooperation. In the longer term, some PBL staff should have levels of 
expertise needed to develop new models jointly with consultants, as opposed to 
always using models developed by others. 

PBL should develop an infrastructure - at both national and international 
levels - for model development and dissemination built around the LUMOS 
initiative that interfaces with other types of models, research orientations, and data 
resources. 

PBL should actively develop a programme for knowledge management and 
intensify national and international collaborations in order to constantly update 
the skills and expertise of its staff. This would help make PBL less dependent 
on external consultants and would further enhance its national and international 
reputation. 

2.23 Use of Results 

Are the results of studies using land-use models adequately used in policy report 
making processes inside PBL and in joint projects with other, relevant national and 
international agencies? 

Land Use Scanner and Environment Explorer - and, indeed, PBL itself - owe 
their existence to the contributions they makes to policy-oriented decision-making. 
The appropriateness and scientific quality of the models results are therefore 
every bit as important as those of the models themselves. The committee was 
informed about several applied projects, in particular those focusing on scenario 
development. This is a very important activity as it combines quantitative with 
qualitative analysis, and links the two land-use models to issues in both policy 
formulation and to external developments beyond the purview of policy-makers. As 
staff have indicated, these projects have provided important learning experiences 
and there seem to be ample opportunities to further improve the process leading to 
such applications . 

The audit committee did not review all the applications that had been carried out 
by PBL using the two models or the ways in which their results were communicated 
to policy-makers . However the committee was concerned to find a lack of a clear 
distinction between the systematic development and examination of consistent 
policies and policy packages, on the one hand, and the generation and exploration 
of future scenarios, on the other. The former concern courses of action that 
policy-makers may adopt, whereas the latter outline potential developments in the 
environment and society at large that policy -makers have little or no influence, over: 
megatrends as individualisation and globalisation having an impact on the demand 
for land. The distinction between policies and scenarios is important because it 



2 Lumos Models from an International Perspective 



31 



is difficult, if not impossible, to distinguish the contribution to the outcomes of 
individual assumptions or policies from the results of scenario simulations: this is 
only possible if individual policies or combinations of policies (policy packages) are 
formulated and examined separately from the scenarios themselves. This distinction 
is essential to the process of decision and planning support and should provide the 
essential framework within which to set up appropriate applications. 

The distinction between policies and scenarios would also be helpful in further 
articulating the role of uncertainty in applied projects . Currently, uncertainty appears 
not to receive the systematic attention it deserves. Several aspects of uncertainty 
are relevant. First, there is the uncertainty of global and local events not under the 
control of decision-makers that are likely to have an influence on future land-use 
patterns (megatrends). Second, there is the uncertainty about the responses of the 
system of interest to policy instruments. Third, land-use decisions of particular 
actors are made under conditions of inherent uncertainty regarding the decisions 
of others, including spatial planners. Fourth, there is uncertainty about the value 
choices that future stakeholders will make. Finally, there is uncertainty about data 
(for example accuracy, representativeness) and model predictions . The generation 
of both future scenarios and policy packages should take these different kinds of 
uncertainty systematically into account. 

This uncertainty is also related to the calibration of the models, especially 
when used in scenario-based studies. Currently model calibration is the result of 
a combination of expert knowledge and statistical analysis. Statistical analysis 
is, of course, very important and successful reproduction of changes in land-use 
patterns provides some evidence of the validity of the models. However, it is 
equally important that any assumptions made are shared by experts outside PBL 
and have institutional (PBL) legitimacy. In addition, validation should also rely on 
meta-analysis of assumptions about individual behaviour and not only on process 
outcomes. 

Finally, it seems as if many of the cases in which models are currently 
used are focussed on impact assessment. This seems to be unnecessarily narrow 
in scope as the data, models and expertise could also be used to support the 
policy-making process, including generating plans, formulating instruments and 
developing standards and guidelines. Concepts such as robust adaptive planning are 
of potential value in this context since they would help connect models, policies and 
scenarios within a more systematic framework. 

Recommendations 

PBL should further expand the methodology for developing scenarios for 
policy-making by systematically differentiating between exogenous developments 
and individual policies and policy packages. In this context, exploration of the role 
of formal decision-making and planning support systems should continue. 

The issue of uncertainty deserves further attention in relation to model 
development, model application and scenario development and assessment. 
Qualitative, as well as quantitative, notions of uncertainty should be associated 
with the various model structures in terms of their inputs, outputs and mechanisms. 



32 



H. Timmermans et al. 



Uncertainties of different kinds also lie at the core of scenarios of the future and 
should, therefore, be systematically considered both within individual scenarios 
and in connection with ensembles of scenarios characterised by different degrees 
of plausibility. 

PBL should consider developing a wider range of qualitative and quantitative 
decision-support tools to be able to expand its role beyond impact assessment. 
In particular, the methodology of robust, adaptive planning has the potential 
to integrate models, policy options and scenarios of exogenous or autonomous 
developments within the same systematic framework. As part of such a framework, 
certain kinds of simpler conceptual models may be developed to help fill gaps 
between detailed quantitative models, on the one hand, and qualitative future 
scenarios, on the other (see recommendations in Section 2.2.1). 

22.4 Communication 

Are the results of studies using land-use models reported in a way that 
communication with policy-makers is well targeted and optimised, while 
maintaining an acceptable level of scientific quality? 

Insights gained from models can be communicated to policy-makers in a variety 
of ways. First, model software can be used in workshops and policy -development 
sessions to provide participants with a better understanding of the issues, underlying 
processes and likely effectiveness and impact of alternative policy instruments. 
Environment Explorer especially stands out in this regard, as it has a very user- 
friendly user interface that does not require any knowledge of scripting and allows 
for the exploration of policy strategies at every level. 

Second, outcomes of the models can be communicated to policy makers. 
Visualisation will clearly play an important role in this regard, although there is 
a risk that visualisation may become a goal in itself and distract attention away 
from key messages. Given the problem of stochastic variation at the micro level 
as discussed earlier, care should be taken to match any visualisation with the 
spatial resolution at which reliable results can be achieved. In general, this will 
preclude the presentation of results at the micro-level, except for demonstration 
purposes, that is, to focus attention on the essential information. While current 
efforts are very promising, the audit committee considers that there still is much 
work to be done on developing visualisation methods that are effective and suitable 
for policy applications. Both models clearly enable good visualisations to be 
developed, including animations, especially in the case of Environment Explorer. 
3-D visualisation shows promise in that it helps engage non-expert decision-makers 
and other stakeholders in scenario testing and development. 

In addition to communication with policy-makers and other stakeholders, 
communication with scientists remains of utmost importance. If the models used by 
PBL do not meet international standards, the credibility of their applications may 
become an issue. Regular publication of scientific articles in peer-reviewed journals 
is important in this respect. 



2 Lumos Models from an International Perspective 



33 



Recommendations 

Current visualisation tools should be further developed and expanded upon 
without losing sight of the ultimate goal: communicating essential information. 

PBL should explore different ways of using visualisation to communicate the 
results of scenario development and policy studies simulated by the models. 

Just as publishing scientific articles on modelling research is desirable, 
publishing in appropriate media about improvements to, and novel applications of, 
visualisation techniques in the context of policy exploration should be encouraged. 

2.25 Next Steps: The Coming Years 

Can the auditors advise PBL on its data, model and application methods for the 
coming years? 

The recommendations listed in the previous sections reflect differing degrees of 
scientific urgency and practicality. Given a thorough knowledge of its own resources 
and limitations, PBL is in a better position than the audit committee to determine 
the most appropriate timing and balance of the efforts that need to be made to meet 
the audit committee's recommendations. Considering that these were formulated to 
be as consistent as possible with PBL's current mode of operation and resources, 
the committee believes that all recommendations could be met within a 5-year time 
frame. Thus, we strongly recommend the development of a detailed 5-year plan to 
that effect. 

The committee considers that some minor issues related to Land Use Scanner 
and Environment Explorer can be dealt with without a need for much planning. 
Moreover, strategically, it is important to start identifying early on (a) the new kinds 
of policy questions that are being asked, (b) how the socio-economic dynamics can 
be captured in a new, additional model or model framework, (c) what additional data 
would be required for these, and (d) to what extent could the effort to develop a new 
model benefit from the (expanded) LUMOS consortium. In this, the role of PBL 
should also be identified and the bureau positioned with respect to other agencies, 
such as DVS (Road and Water Transport Service), because these will be dealing 
with similar issues in the near future. 

Recommendations 

The audit committee considers that PBL should develop a phased plan to 
implement these recommendations over the next 5 years, by, say, 2012. Our 
proposals are quite consistent with the working style of the current research group 
and PBL as a whole and have been made on the basis that these can be realised with 
relatively little change to the organisational structure of the group within PBL. 

Using the LUMOS consortium as a platform, PBL should reach out to agencies 
inside and outside the Netherlands to promote Land Use Scanner and Environment 
Explorer and educate others about their use. This could include engagement in some 
limited training of other professionals and some technology transfer of the models . 



34 



H. Timmermans et al. 



2.3 Conclusion 

This chapter reports the findings of an audit conducted to evaluate the scientific 
quality of the processes and products of PBL relating to the application of land-use 
models primarily at the national level. The audit committee thinks that the Land 
Use Scanner and Environment Explorer are among the best in the field, and are 
very useful for assessing policy impacts relating to the spatial distribution of land 
use at the national level. There are several technical elements that could be further 
developed but this will not dramatically change the adequacy of these models for 
their intended purpose. 

PBL is however also increasingly confronted with questions about the regional 
and even the local scale and questions about management. In general, this means 
that the models should be sensitive to a wider spectrum of policies, and that other 
variables may become important. To address such questions and truly embrace 
the socio-economic activity perspective that is more adequate for such questions, 
substantial model redesign or the development of a new (activity-based) model is 
required. 

In addition, the Agency may consider simpler models for policy development as 
opposed to policy assessment. Intensifying collaboration with other agencies in the 
Netherlands and elsewhere in terms of promoting the models and educating other 
agencies and groups on their use, building on the LUMOS consortium, should be 
the core of an improved communication strategy. 

Reference 

Timmermans, H., Batty, M., Couclelis, H., & Wegener, M. (2007). Scientific audit of national 
land use models; Report and recommendations of the audit committee. Bilthoven: Netherlands 
Environmental Assessment Agency (MNP). 



Chapter 3 

Core Principles and Concepts in Land-Use 
Modelling: A Literature Review 

Jonas van Schrojenstein Lantman, Peter H. Verburg, 
Arnold Bregt, and Stan Geertman 



3.1 Introduction 

Simulation models of land use predict or describe land-use change over space 
and time. Recent overviews of land-use simulation models show an overwhelming 
amount of different types of models and applications (Heistermann, Muller & 
Ronneberger, 2006; Koomen, Stillwell, Bakema & Scholten, 2007; Verburg, Schot, 
Dijst & Veldkamp, 2004). Obviously, such models are simplifications of reality, but 
increasing computing power over the years has made it possible to incorporate more 
and more complexity in such models. This increased complexity, however, tends to 
obscure the theoretical foundations of land-use simulation models. This theoretical 
foundation relates to the core principles that are used to explain land-use change and 
the concepts that are applied to translate these principles into a functioning model of 
land-use change. An in-depth review of land-use change concepts, their underlying 
principles, applicability and translation into actual models does not exist to our 
knowledge. In this chapter we aim, therefore, to analyse the application of various 
theoretical concepts of land -use change that are used in modelling. This analysis is a 
first step to better understand the conceptual background of land-use change and the 
application of these concepts in computer simulation models. Based on this review 
we present some observations on important research issues in land-use modelling 
and suggest possible ways for further model improvement. 

Computer simulation models of land use are characterised in terms of core 
principles and applied concepts in Section 3.2. After this discussion of the 
theoretical background of current land-use models we dwell on two additional issues 
that are important for improving simulation models of land-use change. Lessons 
from model validation efforts are described in Section 3.3, while the importance 
of scale issues in land-use change is dealt with in Section 3.4. The main conclusion 
and discussion can be found in Section 3.5, whereas some themes for future research 
are suggested in Section 3.6. The latter suggestions are specifically focused on the 
Dutch LUMOS models. 



J. van Schrojenstein Lantman (Kl) 

Nelen & Schuurmans, PO Box 1219, 3500 BE Utrecht, The Netherlands 
e-mail: jonas.vanschrojenstein@nelen-schuurmans.nl 

E. Koomen, J. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 35 
Practice, The GeoJournal Library 101, DOI 10.1007/978-94-007-1822-7_3, 
© Springer Science+Business Media B.V. 201 1 



36 



J. van Schrojenstein Lantman et al. 



3.2 Categorising Land-Use Simulation Models 

In the literature, various categorisations of computer simulation models of land-use 
change have been made. Baker (1989) categorised them according to scale: 
whole landscape models, distributional landscape models and spatial landscape 
models. Briassoulis (2000) categorised models according to the modelling tradition 
to which they belong: statistical/econometric, spatial interaction, optimisation, 
integrated and 'other' modelling approaches. Lambin, Rounsevell, and Geist 
(2000) distinguished between empirical-statistical, stochastic, optimisation, 
dynamic simulation, and integrated modelling approaches. Agarwal, Green, Grove, 
Evans, and Schweik (2001) analysed 19 different computer simulation models 
according to a three-dimensional framework: space, time, and human decision- 
making. Verburg et al. (2004) discussed computer simulation models according 
to six features: level of analysis, cross-scale dynamics, driving factors, spatial 
interaction and neighbourhood effects, temporal dynamics, and level of integration. 
Heistermann et al. (2006) classified 18 computer simulation models according to 
geographical, economic and the integration of both. Koomen and Stillwell (2007) 
did not categorise models at all, but rather discussed a number of characteristics 
of computer simulation models: static/dynamic, transformation/allocation, 
deterministic/probabilistic, sectoral/integral, zones/grid. 

Common factors in these categories of computer simulation models can be 
recognised (for example, distinction between dynamic and integrated models), 
but differences in approach make these studies difficult to compare. The general 
complexity of computer simulation models and the fact that the field of land-use 
modelling is interdisciplinary are the most probable reasons for the different 
categorisations made by these authors. Yet, despite the marked differences between 
computer simulation models of land-use change, they do have a common basis. 

It is common practise for modellers to describe the processes of land-use change 
according to a particular concept or mechanism that can be used to characterise 
land-use changes (e.g. Cellular Automata). In fact, a concept of land-use change 
is composed of a set of core principles according to which 'real world' processes 
of land-use change are assumed to operate (e.g. Cellular Automata is based on 
neighbourhood interaction). This concept of land-use change is then codified 
into algorithms, for which other, different algorithms can be used, which will in 
turn lead to different computer simulation models (e.g. Environment Explorer or 
GEOMOD2). Thus, a land-use change algorithm is no more than the translation 
of a concept of land-use change into calculation rules for a computer simulation 
model. 

Figure 3.1 shows the relationship between the process of land-use change and 
computer simulation models. The grey box encloses the steps that are the focus 
of this chapter. Although different divisions can be made, all computer simulation 
models have in common that they are based on four core principles. The main steps 
in Fig. 3.1 are discussed in the following subsections. 



3 Core Principles and Concepts in Land-Use Modelling: A Literature Review 



37 



Land-use 
change process 



Core principles 
of land-use 
change 



Land-use 
change concept 



Land-use 
change 
algorithm 



Computer 
simulation 
model 



Fig. 3.1 Relationship between the land-use change process and computer simulation models 



3.2.1 Land-Use Change Process 

'Land use change is the result of socio-economic and biophysical phenomena, 
dependent on spatial location, scale, and existing land use' (Briassoulis, 2000; 
Lambin et al., 2001; Lesschen, Verburg & Staal, 2005; Meyer & Turner, 1992; 
Turner et al., 1995). In our literature review we focused on concepts of land-use 
change and their origin, not on the process of land-use change itself. Nevertheless, 
the literature mentioned in this subsection provides a good starting point for reading 
on land-use change processes . 



3.2.2 Core Principles of Land-Use Change 

In-depth analysis of the literature on land-use modelling shows that all simulation 
models of land use are based on at least one of the following four core principles of 
land-use change: 

1 . Continuation of historical development; 

2. Suitability of land (in monetary or other units); 

3. Result of neighbourhood interaction; or 

4. Result of actor interaction. 



Continuation of Historical Development 

The premise behind historical land-use change is that future land use can be 
predicted by means of historical changes. This can be interpreted in several ways: 
in the past people liked to live near the water, so in the future they will like to do so 
too; or 15% of the forest became residential land so it is likely that given the same 
time horizon this will happen again. 

Extrapolating land-use change from past changes into future changes is an 
intuitive concept and is, therefore, widely used. A business-as-usual scenario is 
a practical application of this principle, as demonstrated by Kuijpers-Linde et al. 
(2007). 

Suitability of Land (in Monetary or Other Units) 

Suitability may cover a combination of factors, such as soil suitability, spatial 
location or terrain characteristics of a piece of land; distance to a market is an aspect 



38 



J. van Schrojenstein Lantman et al. 



of spatial location. The underlying premise is that people want to maximize profit, 
which can be expressed either in monetary units (quantitative) or non-measurable 
units (qualitative). 

Result of Neighbourhood Interaction 

The principle of neighbourhood interaction states that the possibility of transition 
from one use of land to another is dependent on the land use of its surrounding cells. 
This driver can be biophysical, e.g. a land cover affecting that of neighbouring cells, 
or a socio-economic one; the latter can be explained with, for example, the Core 
and Periphery model, which assumes that people want to maximise profit (Fujita, 
Krugman & Venables, 1999; Krugman, 1991; 1999). 

Result of Actor Interaction 

The main assumption in actor-focused decision-making is that land-use change is 
the result of interaction between actors . Actors can be represented as agents: a single 
entity or a group of actors, depending on the scale of modelling. The core principle 
of actor decision-making aims at explaining and understanding socio-economic 
drivers and policies for development. A complicating factor is that it is only recently 
that surveys have been held among actors in certain case studies in which their 
preferences were monitored. Therefore, the amount of data is insufficient and it 
is difficult to validate them. In practice this core principle of land-use change 
modelling is still in its infancy, but it seems to be a promising research tool 
(Matthews, Gilbert, Roach, Polhill & Gotts, 2007). 

3.23 Land-Use Change Concepts 

Different concepts of land-use change represent differing attempts to enable science 
to explain and thus translate reality on the ground into a model; each concept brings 
with it its own advantages and disadvantages. Any concept of land-use change is 
always based on one or more of the four core principles described in the previous 
subsection. 

Quantity of Subjects in the Literature on Land-Use Change Modelling 

The following well-known concepts of land-use change were selected from the 
literature: 

• logit functions; 

• markov chains; 

• cellular automata (CA); and 

• agent-based modelling. 

A search of the Scopus database was made for each concept. With access to over 
15,000 peer-reviewed journals (Scopus, 2008), Scopus is the largest abstract and 
citation database of research literature and quality web sources. In addition to basic 



3 Core Principles and Concepts in Land-Use Modelling: A Literature Review 



39 



information such as author(s), year of publication and name of journal, the citation 
count for each concept was compiled. As references in Scopus only go back as far 
as 1996, articles published before then were not taken into account. 
Subject experts selected the following search terms: 

• 'artificial neural networks' AND 'land use'; 

• 'cellular automata' AND 'land use'; 

• 'agent-based' AND 'land use'; 

• 'agent based' AND 'land use'; 

• 'Markov' AND 'land use'; and 

• 'logit' AND 'land use'. 

The search was performed in February 2008 and the papers found were 
classified according to the concept in use. Papers in which two different concepts 
are compared were counted as an occurrence of both concepts. A paper describing 
a model combining two (or more) concepts was counted as hybrid paper. Some 
papers described concepts only in theory, while other papers described concepts 
as applied in case studies. After the analysis, the database was checked for any 
key papers that were missing; these were added when necessary. Figure 3.2 shows 
the results of the literature search. It was remarkable that, with hybrid and CA 
papers combined, CA appeared in over 40% of the papers published on land-use 
modelling. Apart from the obvious fact that CA is a popular concept among 
scientists modelling land use, other explanations are possible. One of them is that 
when an article is written on the use of cellular automata, it receives this keyword 

Number of articles 

30 1 □ Hybrid 

^ Agent-based 

25 I — , | Cellular automata 




1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 



Fig. 3.2 Scientific articles in Scopus on various concepts of land-use change 



Scopus is the largest abstract and citation database of research literature and quality web sources. 
It is designed to find information scientists need. Quick, easy and comprehensive, Scopus provides 
superior support for literature searches. Surf to www.scopus.com. 



40 



J. van Schrojenstein Lantman et al. 



from the author. If, however, an article is written about a land-use model based on 
a logit concept, the keyword selected by the author may not be iogit'. Our Scopus 
literature search is, therefore, merely an indication of the number of articles written 
about certain concepts of land-use change. 

3.2.4 Concepts of Land-Use Change in Practice 

The following concepts of land-use change were identified in the literature: 

• cellular automata; 

• statistical analysis; 

• Markov chains; 

• artificial neural networks; 

• economics-based models; and 

• agent-based systems. 

In this subsection, each concept will be discussed in terms of the basic principle 
of the concept and its use in practice. 

Cellular Automata (CA) 

The most well-known land-use change concept is that of cellular automata 
(Langdon, 1998; White and Engelen, 1994). The basic principle of CA is that land- 
use change can be explained by the current state of a cell and changes in those of its 
neighbours. It is, thus, based on the core principles of continuation of historical 
development and result of neighbourhood interaction. In other words: if a road 
through a forest is paved, this enhanced the connectivity of the region, with the result 
that forest land use is replaced by residential use. CA comprises four elements: cell 
space; cell states; time steps and transition rules (White & Engelen, 2000). Decision 
rules can be expert-based or derived from statistical analysis. Two main types of 
CA can be distinguished: unconstrained and constrained. Unconstrained is the most 
'true' CA, as it only uses decision rules to calculate land-use change. In constrained 
CA, the amount of land-use change per land use class is limited; the limit of a certain 
land-use class is either expert-based or calculated from historical land use. 

Made famous by Gardner (1970), John Conway's game of life is the best known 
example of a cellular automaton to date. Tobler (1979) was the first to introduce the 
use of cellular automata in geography. In the decades that followed this was further 
developed by Couclelis (1985), Batty and Xie (1994), and White and Engelen 
(2000). More recently, Hagoort (2006) has put together a nice overview of the 
history of CA, and Pinto and Antunes (2007) provide an overview of the use of 
CA in urban studies. In the Netherlands, a constrained CA model (Engelen, White, 
Uljee & Drazan, 1995; Geertman, Hagoort & Ottens, 2007) called the Environment 
Explorer (part of the LUMOS toolbox), is commonly used in spatial planning and 
policy-making in (de Nijs, de Niet & Crommentuijn, 2004). 



3 Core Principles and Concepts in Land-Use Modelling: A Literature Review 



41 



Statistical Analysis 

All kinds of statistical information can be derived from land-use maps. Statistical 
information can be based on all four core principles, depending on the aim of 
the research. A widely used application is the computer program FRAGSTATS 
by McGarigal and Marks (1995), which can analyse ample amounts of landscape 
statistics. In the discipline of land-use modelling, linear regression, probit 
regression, binomial logit and multinomial logit models are used to analyse 
statistical relations in land use and predict future patterns . Logit analysis provides 
an interpretable linear model (deMaris, 1992). Logistic regression is used to analyse 
the probability of occurrence, dependent on different factors, of a certain land- 
use category, for instance biophysical characteristics, distance from location to the 
market (Verburg, Ritsema, van Eck, de Nijs & Schot, 2004). The coefficients for 
the conversions can be calculated from historical land use and extrapolated to the 
future. A logit model can be based on neighbourhood interaction, historical land- 
use change, soil suitability, spatial location or combinations of these. A binary 
logit model is a mathematical variant of an ordinary linear model that will give 
a prediction of land-use change relative to all other options. A multinomial logit 
(MNL) model is similar, with the exception that it describes the conversion of 
different land-use categories relative to a reference category (deMaris, 1992; Liao, 
1994; Wrigley, 1976). 

Among the first to apply logit models in relation to land-use change were 
Veldkamp and Fresco (1996), with the CLUE model, and Wear and Bolstad (1998), 
who linked a negative binomial regression model of building density with a logit 
model of land use. An example of MNL analysis is given in Walsh, Soranno, 
and Rutledge (2003), where spatial association of various categories of land use 
is analysed. Land Use Scanner (Hilferink & Rietveld, 1999), which has been used 
in producing sustainability outlooks for the Netherlands (Borsboom-van Beurden 
et al., 2005; Kuijpers-Linde et al., 2007), is an example of the use of logit modelling 
in the planning and policy environment. 

Markov Chains 

Burnham (1973) was one of the first to propose using Markov chain analyses 
for modelling land-use change. Such analyses are based on the core principle of a 
continuation of historical development. A Markovian analysis uses matrices (e.g. 
Table 3.1) to represent changes between land-use categories. 



Table 3.1 Example of a 
Markov matrix 



Current\future land use 


Farmland 


Peri- Urban 


Urban 


Farmland 


0.50 


0.40 


0.10 


Peri-urban 


0 


0.8 


0.2 


Urban 


0 


0.1 


0.9 



42 



J. van Schrojenstein Lantman et al. 



Under an assumption of stationarity (temporal rate of change and amount of 
change stay the same), the matrix can be used to calculate the probability of land- 
use change of one land-use category to another. A disadvantage of this type of 
analysis is that it is non-spatial, meaning that additional assumptions are required 
for allocation. Distinction is made between a first-order and second-order Markov 
matrix. The former uses a matrix with current land use and a change matrix based 
on expert knowledge, while the latter calculates changes from one land use to 
another by comparing two maps of land use over time, that is, this change matrix 
is constructed from historical land-use change. The temporal rate of change is 
assumed to be constant, so the change matrix can then be used in the same way 
as in first-order Markov analysis. 

Examples of Markov chains in land-use studies are given by Muller and 
Middleton (1994), Fearnside (1996) and Lopez, Bocco, Mendoza, and Duhau 
(2001). Because of its simplicity, Markovian analysis is a popular technique to 
combine with other concepts. The GEOMOD2 model (Pontius, Cornell & Hall, 
2001) is an example of combining Markov techniques with statistical analysis. 

Artificial Neural Networks 

The use of artificial neural networks (ANNs) in land-use modelling has increased 
substantially over the last few years because of advances in computing performance 
and the increased availability of powerful and flexible software (Skapura, 1996). 
ANNs are self-learning computer models and are used for pattern recognition in 
many disciplines (Pijanowski, Brown, Shellito & Manik, 2002). Recently, ANNs 
have found their way into the world of land-use modelling. The ANN algorithm 
assumes a relation between past and future land-use change and can be linked to 
suitability maps. ANNs can be based on all four core principles. The first to apply 
ANNs to a computer simulation model was Pijanowski et al. (2002). The model 
'trains' itself on a dataset and the corresponding land-use maps of different years 
enabling it to recognise and reproduce the pattern of land-use categories (Mas, Puig, 
Palacio & Sosa-Lopez, 2004; Pijanowski, Pithadia, Shellito & Alexandridis, 2005). 

Economics-Based Models 

Although not exactly a concept of land-use change, but rather more a land-use 
theory, economics-based models of land use cannot be left out of the list of concepts 
of land-use change. These models are based on the core principle of suitability 
of land (in monetary or other units), although the core principle of continuation 
of historical development can also be included. The first theory on land use was 
developed by Johan Heinrich von Thiinen in 1826. It was translated into English in 
1966 (von Thiinen, 1966). Von Thiinen converted the principle of Ricardo (1817) 
that profit will be used to reinvest in a land-use change concept. Von Thiinen stated 
that as long as the profit of a commodity (turnover minus production costs) is higher 
than the transportation costs of the commodity, the land will continue to be used to 
produce this commodity. Alonso (1964) expanded this theory by including land-use 
suitability and an individual bid-price curve for every household or firm. A little later 
Sinclair (1967) expanded von Thiinen's theory to explain urban sprawl. More recent 



3 Core Principles and Concepts in Land-Use Modelling: A Literature Review 



43 



applications of the theory have been reported by Chomitz and Gray (1996), Nelson 
and Hellerstein (1997), and Walker (2004) which all use von Thtinen's model as the 
base theory to explain tropical deforestation. 

Agent-Based Systems 

Be it agent-based modelling, individual-based modelling, micro-simulation or 
activity-based modelling, the common ground of all these concepts of land-use 
change is that they are based on the core principle of actor interaction. The 
differences between these approaches are to be found in the institutional dimension 
of scale, for example, ranging from individual modelling to group behaviour and the 
number of agents being modelled. One of the first actor-interaction based models 
was a competition model for individual trees (Bella, 1971), but it took a long time 
before this type of model was applied to land-use change. Balmann (1996) was 
one of the first to introduce a spatially dynamic model of land-use change for 
analysing structural change in agriculture in which individual farmer preferences 
were taken into account. Since 1996 the number of articles on agent-based land-use 
models has steadily grown over the years (Fig. 3.2). Within agent-based systems, 
we distinguish four sub-types: agent-based modelling; individual-based modelling 
and micro-simulation; activity-based modelling; and expert-based decision rules. 

An agent-based model of land-use change consists of two key components: a 
map of the study area, and a model with agents that represent human decision- 
making (Parker, Berger & Manson, 2001). The preferences of agents can be defined 
by expert judgement, by using questionnaires, or by using an artificial neural 
networks technique. A multi-agent system is a set of agents interacting in a common 
environment, which consists of everything in the model that is not an agent (Ferrand, 
1996; Le Page, Bousquet, Bakam, Bah & Baron, 2000). An agent is a representation 
of an actor in the process of land-use change and can be either an individual or a 
group. The following actor interactions can be distinguished (Ligtenberg, Bregt & 
Lammeren, 2001): 

• interactions between the spatial objects of the environment and actors; 

• interactions between spatial objects; 

• interactions between actors and spatial objects; and 

• interactions between actors and other actors . 

Originating in the discipline of ecology, individual-based modelling is a subset 
of agent-based modelling. It differs from the agent-based modelling usually done 
in that the scale of the concept is always at the individual level. Each individual is 
represented as an agent, each with their own preferences (Grimm et al., 2006). In 
social sciences, micro-simulation models aim at reproducing human behaviour at 
the individual level (Moeckel, Schurmann & Wegener, 2002). Micro-simulation is 
mostly used to model urban land use and transport that is under development; an 
example of this approach is the Ilumass project (Wagner & Wegener, 2007). 

Activity-based modelling is a special subset of the agent-interaction decision- 
making concept. A relatively new concept in research on land-use modelling, 



44 



J. van Schrojenstein Lantman et al. 



activity-based modelling is based on the work of Arentze and Timmermans (2000) 
and the UrbanSim model (Waddell, 2002). In the Netherlands, there is need for 
policy-makers to know the impact of intensification of cities as compared to that 
of urban sprawl. In activity-based modelling, the relation between mobility and 
infrastructure and their impact on land-use change is explicitly addressed which 
in turn can be used to recognise, for example, traffic bottlenecks. The concept 
of activity-based modelling is relatively new and opinions on what an activity- 
based model exactly is differ. It can range from modelling of population density 
in combination with CA, to a sort of individual-based model in which, for example, 
the result of movements per individual during the day is taken into account, which 
can help recognise possible traffic jams. 

In cases of uncertainty about parameters, expert-based rules, which encapsulate 
experts' knowledge of a specific area, are used to estimate them. Basically, all four 
core principles can be used for this. The IMAGE model (Alcamo, Kreileman, Krol 
& Zuidema, 1994a; 1994b) is an example of a land-use model built on expert-based 
decision rules. 

Analysis of concepts of land-use modelling in practice and the models involved 
allows modellers to pick the most appropriate concept of land-use change for the 
study area under investigation. The relation between land-use change concepts and 
underlying core principles is shown in Table 3.2. 



3.2.5 Selection of Operational Land-Use Simulation Models 

This subsection contains a selection of computer simulation models of land-use 
change. Various models have been developed that can estimate future land use. 



Table 3.2 Relation between concepts of land-use change and core principles 

Suitability 

Continuation of land Result of 

Concept of land-use of historical (in monetary neighbourhood Result of actor 

change development or other units) interaction interaction 



Economics-based models P 

Agent-based models P 

Cellular automata A 

Statistical analysis P 

Markov chains A 

Artificial neural networks P 

Individual based P 

modelling and 

micro-simulation 

Activity-based models P 

Expert-based decision P 

rules 



A 

P P A 

P A - 

P P P 

P P P 

P P A 

P P A 

P P P 



A = Always, P = Possible, - = Not possible or not practical 



3 Core Principles and Concepts in Land-Use Modelling: A Literature Review 



45 



For reviews of some of these models see Agarwal et al. (2001), Hunt, Kriger, and 
Miller (2004), and Verburg et al. (2004). All such models use different concepts 
and algorithms to decide where and when land-use change will take place. It is 
beyond the scope of this chapter to describe all the land-use models available: 
only models from the LUMOS toolbox and a small selection found in the literature 
are described. The models are described only briefly here as computer simulation 
models of land-use change are not the focus of this chapter. 

The LUMOS toolbox consists of Land Use Scanner and Environment Explorer 
(LUMOS, 2005). Land Use Scanner allocates land according to bid prices for 
various types of land use (Hilferink & Rietveld, 1999; Koomen, Loonen & 
Hilferink, 2008). The possibility of government intervention in determining land 
use is taken into account by, among other things, adding aggregate constraints. 
Environment Explorer is built on a local-level Cellular Automata model constrained 
by regional-level spatial interaction model (White & Engelen, 2000). 

Other well-known land-use models are: 

• CLUE (Conversion Land Use and its Effects) is a dynamic simulation model 
using empirically-derived relations between land-use change and driving forces 
from cross-sectional analysis at multiple scales (Veldkamp & Fresco, 1996; 
Verburg, de Koning, Kok, Veldkamp & Bouma, 1999; 2004). 

• Land Transformation Model (LTM), which combines GIS and ANNs to predict 
future land use. The ANNs are used to learn the patterns of development in a 
region, whereas GIS is used to develop the spatial-predictor drivers and perform 
spatial analysis on the result (Pijanowski et al., 2002). 

• GEOMOD2 can be used to model both forwards and backwards in time. It 
selects locations for a particular type of land use based on: nearest neighbourhood 
interaction; political sub-region; and the pattern of biophysical attributes (Pontius 
etal.,2001). 

• SLEUTH (Slope, Land use, Exclusion, Urban extent, Transportation and Hill 
shade), formerly called the Clarke Cellular Automaton Urban Growth Model, 
was developed for and tested on various cities in North America. It is based 
on cellular automata and is designed to be scalable and universally applicable 
(Silva& Clarke, 2002). 

• UrbanSim uses aggregate economic and spatial-interaction models, and pursues 
a disaggregated approach to predict changes over small time steps. It takes into 
account the demand for real estate at each location, and the actors and choice 
processes that influence patterns of urban development and real-estate prices 
(Waddell, 2002). 

• IMAGE is an ecological-environmental framework that simulates the 
environmental consequences of human activities worldwide. The objective is 
to explore the long-term dynamics of global change. It has a relatively coarse 
resolution (of 5 min x 5 min) and the allocation of new land use is based on 
demand, location preferences and autonomous system change (Alcamo et al., 
1994a; b). 



46 



J. van Schrojenstein Lantman et al. 



• CORMAS stands for Common-pool resources and multi-agent systems. It is a 
multi-agent framework in which interactions between groups of agents and a 
shared environment with natural resources can be simulated (Le Page et al., 
2000). 

• ILUMASS is a micro-simulation model of urban land use. Unfortunately, it 
never became fully operational, but it is a good experiment on micro-simulation 
of urban land use. It takes into account Land Use (Population, Accessibility, 
Firms), Transport (Travel demand, dynamic traffic assignment, demand for goods 
transport) and Environment (impacts, emissions) (Wagner & Wegener, 2007). 



3.2.6 Core Principles and Concepts in Practice 

Our literature research has enabled us to compile an overview of the relationships 
between core principles, land-use change concepts, land-use change algorithms and 
land-use change models in practice; see Fig. 3.3. The figure links the core principles 
with the modelling concepts that are normally associated with it and, by way of 
example, indicates examples of existing land-use models that rely on that particular 
combination of principles and concepts. Since statistical analysis, artificial neural 
networks and agent-based systems can all include one of the core principles, they 
are connected to all the core principles. The same applies to the computer simulation 



Neighbourhood 
interaction 



Historical 
development 



Suitability 
of land 



Actor 
interaction 



Core principals 



Concepts 



a: 



Cellular 
Automata 



Statistical 
analysis 



Constrained 
Unconstrained 



Markov 
chains 



Artificial Neural 
Networks 



Linear regression 
Probit regression 
model 
Binomial Logit 
Multinomial logit 



1A 



Economic 
models 



Expert-based 
decision rules 



Agent-Based 
Modelling 



Microsimulation 

I 



1 



Individual Based 
Modeling 



Multi -Agent 
Systems 



Activity-based 
modeling 



Environment 
Explorer 



Economic 
LUC 
models 



Land Trans- 
formation 
Model 



Deforestation 
models 



Existing computer simulation models of land-use change 

Fig. 3.3 Examples of relationships between core principles, concepts and land-use simulation 
models in practice 



3 Core Principles and Concepts in Land-Use Modelling: A Literature Review 



47 



models of land-use change, for example, Land Use Scanner, CLUE, economic land- 
use change models and GEOMOD2, all of which incorporate statistical analysis. 

This detailed overview has been compiled from the results of our literature 
research. Since models are mostly developed further after the initial paper is 
published, it is possible that some models now incorporate other or more core 
principles or land-use change concepts than is shown in Fig. 3.3; the same 
relationships between core principles and concepts of land-use change are also 
shown in Table 3.2. 



3.3 Lessons from Model Validation 

Validation is an essential step in assessing the performance of any land-use model 
(see, for example, Pontius et al., 2008). In this section we list some general lessons 
that can be learned from model validation and that are relevant for the further 
development of land-use simulation models. 

Validation of land-use models consists of two elements: validation of the 
spatial resemblance of the output maps with reference maps; and validation of the 
resemblance of the described land-use change process. If a model is able to predict 
spatial resemblance accurately but fails to predict the process, it is questionable 
whether policy -makers can learn much from the results. If, on the other hand, the 
model is more process oriented but fails to predict spatial resemblance, it is also of 
little use. 

In general it can be said that an inductive (data-driven) approach is strong 
in reproducing land-use patterns which is (spatial resemblance), but is weak in 
explaining correlations found. A deductive (theory-driven) approach, on the other 
hand, is strong in explaining how and why land use will change (which is process 
resemblance), but is weak in spatial allocation of land-use change (Overmars, de 
Groot&Huigen,2007). 

Overmars et al. (2007) have shown that prediction of spatial location with 
deductive models is promising when compared to inductive models. For land-use 
change developments that follow a single dominant process it is expected that a 
deductive model will score only marginally less on spatial resemblance than an 
inductive model (see Fig. 3.4). For land-use changes that result from multiple 
processes, the differences in spatial and process resemblance between inductive and 
deductive simulation models increases . 

One could argue that inductive simulation models, in which a lot of correlations 
are found, are more suitable for allocating land use at the correct spatial location. 
The obtained correlations, however, do not provide a causal relation and that 
explains why land-use change occurred at specific locations. At the other end of 
the spectrum are deductive models, which are aimed at modelling the process 
behind land-use change. Although these theory -driven models approximate the 
governing processes more closely than inductive models, they have more difficulty 
in identifying the exact spatial location of change. But since these kinds of models 



48 



J. van Schrojenstein Lantman et al. 



Multiple dominant land-use 
change processes 



Single dominant land-use 
change processes 



Spatial resemblance 
ik 

Induction 



Perfect 



Spatial resemblance 



Induction Perfect 



Deduction 



Random 



Deduction 



Random 



Process resemblance 



Process resemblance 



Fig. 3.4 Process resemblance versus spatial resemblance for various types of models in relation 
to the complexity of underlying land-use change processes 

provide more insight into causality, they enhance the learning process more than 
inductive models. 

In general, strong process resemblance is of more use to policy-makers than 
spatial resemblance because they can learn from the process and the interactions 
that take place. Ideally, however, a simulation model should be able to meet spatial 
and process resemblance criteria. To validate a model on its process accuracy is a 
difficult task since the process of land-use change is often very complex, so it is 
impossible to check whether the model accurately has reproduced the process. A 
possible technique for checking whether all key processes have been included in 
a model is to add other processes to the model. If the outcome, such as the land- 
use pattern, does not change significantly it indicates that all key processes were 
initially included in the model (or, at least, that the new process that was added to 
the model is not important in producing spatial patterns. Thus, the key processes 
can be identified by a combination of common sense and trial and error. To check 
the stability of the model, the same process can be modelled several times. If the 
outcomes are not comparable, it indicates that the model is not stable enough to use 
for estimating future land-use change. In that case, it might be necessary to split the 
model into sub-components to find the cause of the instability and thus validate its 
sub-components first. 



3.4 Scale Issues in Land-Use Modelling 



Scale and land-use modelling are interwoven phenomena. Choice of scale of 
modelling is of influence on the result as was described by Benenson (2007), who 
discussed the theoretical impact of scale on CA transition rules . Wu and Li (2006) 



3 Core Principles and Concepts in Land-Use Modelling: A Literature Review 



49 




Fig. 3.5 Hierarchy of scale (after Wu & Li, 2006) 

distinguish a conceptual hierarchy of scale (see Fig. 3.5) consisting of three levels: 
dimension, kind, and components of scale. Each of these levels is briefly described 
below. 

The first level is the dimension of scale, which can be temporal, spatial or 
institutional (Biirgi, Hersperger & Schneeberger, 2005; Dungan et al., 2002; Wu & 
Li, 2006). Temporal scale is measured in time steps (e.g. years), spatial scale 
is measured in resolution (e.g. kilometres) and institutional scale is measured in 
institutions (e.g. two countries). Institutional scale is a special kind of scale since 
two institutions are never of the same size. Furthermore the institutional scale is 
discrete, whereas time and space are continuous. 

On the second level of scale, Wu and Li (2006) distinguish different kinds of 
scale; see the left-hand side of Fig. 3.6. The intrinsic scale is the scale on which a 
process operates in reality. The process is observed by humans on an observation 
scale. When a certain process is modelled the modeller has to choose a modelling 
scale, which can be, but is not necessarily, the same as the observation scale. The 
modelling results are then presented to the policy-makers, who will look on a policy 
scale at the process. For explanatory purposes, the kinds of scale are related to land 
cover and land-use change in Fig. 3.6. This figure distinguishes between the land 
cover that can be observed (e.g. building) and the use to which the land is put (e.g. 
residential or commercial). Analysis of spatial developments normally starts with 
observation of land cover change processes using remote sensing (with a resolution 
that can range from 1 m up to 30 m). The analysis or modelling scale is normally an 
aggregate of the observation scale. In case of the Dutch CBS data used in the Land 
Use Scanner this is a resolution of 25 m x 25 m. The policy scale depends strongly 
on the focus of the policy theme and may involve a variety of scales. Our focus in 
this chapter is the process of land-use change, which is often more complex than the 
process of land-cover change because of the human factor. Different core principles 
(e.g. neighbourhood interaction) exist to describe the way land-use change processes 



50 



J. van Schrojenstein Lantman et al. 



Intrinsic scale 






Land-cover 
change process 






Land-use change 
process 
























r 








f 




r 




Observation 
scale 






Observation 
scale, e.g. 
remote sensing 






Core principle of 
land-use change 




J 


i. 












r 












Analysis/modeling 
scale 






Analysis/modeling 
scale, e.g. 
CBS data 






Land-use change 
concepts and 
algorithms 




J 


i. 






i 






r 






1 


r 




Policy scale 




Policy scale 






Computer 
simulation models 
of land-use 
change 





Fig. 3.6 Kinds of scale in modelling land-use change (after Wu & Li, 2006) 



are supposed to have taken place. These can be seen as the observation scale of 
land-use change. Land-use change concepts and the algorithms that express them 
can then be thought of as the modelling scale of the land-use change process. The 
policy scale can be found in the outcomes of the models. Since policies influence 
processes of land-use change, there are feedback links back to the intrinsic scale, 
i.e. the land-use change process. 

The third and last hierarchical level of scale relates to its components and 
is about quantifying and developing measurable scaling relations. These include 
cartographic scale, grain, extent, coverage and spacing (Wu & Li, 2006). 

Figure 3.6 also places the framework in Fig. 3.1 in perspective. The implications 
of scale issues are important in the case of land-use modelling (Parker et al., 2001). 
Scaling difference between the policy scale and the intrinsic scale should be as small 
as possible to avoid model-induced errors. If, for example, the modelling scale level 
chosen is too large, the actual process of land-use change may be overlooked. 

In recent research, the focus (as far as scale issues are concerned) has been 
on spatial correlation of land use and land-use change (Overmars, de Koning & 
Veldkamp, 2003; Verburg et al., 1999). Analysis at different temporal scales is more 
difficult due to incomplete or inconsistent time series of the required spatial data, 
(e.g. Li & Yeh, 2001). The institutional level of scale differs per country and case 
study. 



3 Core Principles and Concepts in Land-Use Modelling: A Literature Review 



51 



3.5 Conclusions and Discussion 

The literature analysis we present in this chapter is a first step towards acquiring 
a better understanding of the usability of land-use models. Knowing the core 
principles of a model can help a modeller use it in the right way and help the policy- 
maker to understand the basis of the model. It makes the 'black box' - which models 
sometimes are to policy-makers (and even modellers) - less black. To illustrate the 
degree of complexity of advanced models, it is estimated that someone who is new 
to such a model needs at least 2-3 months before he or she is fully capable of 
understanding and working with the model . 

Of the many models that exist, this report mentions only a few. Some models 
are clearly too complex, which makes them difficult to implement at other spatial 
locations, such as UrbanSim of which the implementation at the Environmental 
Assessment Agency illustrated the difficulty to use this model in the context of a 
different location and data availability. 

Care should be taken not to judge models by their spatial performance alone. 
Over- fitting of models is a well-known trap. There is a balance to be found in 
calibrating a model for a certain spatial location and the usability of this calibration 
for other spatial locations. Research aims might also influence this. Perhaps the aim 
is to build a model that can be used immediately at different spatial locations, or 
perhaps it is to build a model that has to be calibrated for each spatial location. The 
first option, a model that can be used at different spatial locations, seems the most 
preferred one . 

It is better to use simpler models because it is easy to explain what happens and 
the results are easier to explain. It seems that the complexity land-use modellers are 
trying to capture is not something that can be modelled. There are simply too many 
factors to be taken into account. Agent-based models focus strongly on process 
resemblance, artificial neural networks focus strongly on spatial resemblance. 
Changing the focus of the model will result in a loss of its strong points, so a 
trade-off has to be made in which certain models are used to optimal effect. 

More care should also be taken in listening to policy-makers. What do they 
actually need? This is currently the focus of the work of several PhD students in 
the Netherlands. 

It is our opinion that - at least for the Netherlands - using a model such as Land 
Use Scanner, which works with suitability, demand and supply, is a good approach, 
but not for creating beautiful maps . Rather, it should be used more with a technique 
such as variant-invariant region, as Brown, Page, Riolo, Zellner, and Rand (2005) 
show in their paper on path dependence in agent-based models of land use. This 
works as follows. Different scenario's can be analyzed in the computer simulation 
model. Maybe from the results it can be seen that certain developments will always 
take place (invariant regions) or that certain areas are high in demand and their future 
use depends on the chosen scenario (variant regions). Then bottlenecks for spatial 
planning are recognised and policy-makers can make their decision instead of the 
model making the decision for them. 



52 



J. van Schrojenstein Lantman et al. 



3.6 Themes for Future Research 

Inspired by the literature review, we have selected several themes that warrant future 
research: 

• thematic classification (number of land-use classes); 

• extent (size of the study area); 

• resolution (cell size); and 

• time step (discrete interval in time). 

The relation between these themes is shown in Fig. 3.7. Extent and cell size are 
assumed to be strongly correlated; for a study area with a specific extent, a certain 
resolution will be optimal. If a study area, for example, only measures 4 km x 
4 km, it does not make sense to use a cell size of 1 km x 1 km. Sizes of 100 m x 
100 m or 25 m x 25 m would then be more appropriate to discern spatial patterns 
within the area. The thematic classification of land use is often related to the extent 
and resolution of the case study. With increasing diversity (increasing number of 
classes), the minimal extent for a specific cell size increases. 

Time step is related to all three (extent of study area, cell size and thematic 
classification), but the chosen time step mostly depends on the choice of 
classification and the development or amount of land-use change in a study area. If 
there is little change over time, the ideal time step for modelling can be larger than 
when there is a lot of change. Intuitively, if there are small time steps there is room 
to simulate more diversity (more different classes) than if there are large time steps. 
Of course, data is scarce, so research is often limited to a time step dictated by data 
supply. 

For each of the four themes described above we list a few research options. These 
options focus on the Netherlands as they are meant to help the further development 
of the Dutch LUMOS models. 

Thematic Classification 

The definition of a thematic classification is a basic step in developing a land-use 
simulation model. Does a model simulate two (e.g. built or non-built) or more land 
uses? And what is the maximum number of land-use classes that can be simulated 
given a certain cell size and extent of a study area? In other words, what would be a 



Extent 



Thematic 
' classification 



Time step 



Cell size 



Fig. 3.7 Relation between cell size, extent of study area, thematic classification and time step 



3 Core Principles and Concepts in Land-Use Modelling: A Literature Review 



53 



sound choice of the number of classifications to be used in land-use modelling given 
a certain area and cell size? 

To start finding an answer the following option can be considered. First assume 
a certain extent of study area and cell size. Then the number of land-use classes 
is varied (from 2 up to 10). This can be investigated for different spatial locations. 
Although it may also be possible to approach this problem theoretically, i.e. with 
complicated mathematical equations, the ultimate goal is only to provide a rule of 
thumb with which modellers can determine an optimum number of classes. 

Extent of Study Area 

Often the choice of extent of a study area is arbitrary and data dependent. The 
bigger the study area (assuming cell size remains the same), the more land-use cells 
have to be allocated. Intuitively, there is a minimum extent that a study area should 
have before the use of a land-use model starts to make sense. A Markov model can 
be used to start off with a study area extent of four cells and only two classes (most 
extreme) and then, by making the extent bigger for each time step, an optimal extent 
can be found after which, statistically, the Markov model performs worse than not 
modelling (persistence). 

Resolution (Cell Size) 

For the Netherlands, further research could be done on the effect of using 
varying cell sizes (25/100/250/500/1,000 m) for different spatial locations in the 
country. The Land Use Scanner model offers the opportunity to simulate at all 
these resolutions, but the impact of these different choices has not yet been analysed 
systematically. 

Time Step 

To better understand ongoing spatial developments an analysis of historic 
developments is extremely useful. For the Netherlands several data sets are available 
for the years 1960-2000, with 10-year time steps. From these land-use maps, 
Markov matrices can be constructed. It can be investigated whether the stationarity 
principle holds or that the rate of change for certain classes changes over time. The 
matrix can be used to extrapolate land use into the future, validate the current model 
and obtain information on the impact of spatial planning regulations in the observed 
period. 

References 

Agarwal, C, Green, G. M., Grove, J. M., Evans, T. P., & Schweik., C. M. (2001). A review 
and assessment of land-use change models: Dynamics of space, time, and human choice. 
Bloomington, IN: South-Burlington, Center for the Study of Institutions Population, and 
Environmental Change, Indiana University. 

Alcamo, J., Kreileman, G. J. J., Krol, M. S., & Zuidema, G. (1994a). Modeling the global society- 
biosphere-climate system: Part 1: Model description and testing. Water, Air, & Soil Pollution, 
76(1-2), 1-35. 



54 



J. van Schrojenstein Lantman et al. 



Alcamo, J. G. J., van den Born, A. F., Bouwman, B. J., de Haan, K., Klein Goldewijk, O., Klepper, 

J., et al. (1994b). Modeling the global society-biosphere-climate system: Part 2: Computed 

scenarios. Water, Air, & Soil Pollution, 76(1-2), 37-78. 
Alonso, W. F. (1964). Location and land use. New Haven, CT: Harvard University Press. 
Arentze,T. A.,& Timmermans,H. J. P. (2000). Albatross, a learning based transportation oriented 

simulation system. Eindhoven: Eindhoven University. 
Baker, W. L. (1989). A review of models of landscape change . Landscape Ecology, 2(2), 1 1 1-133. 
Balmann, A. (1996). Farm-based modelling of regional structural change: A cellular automata 

approach. European Review of Agricultural Economics, 24, 85-108. 
Batty, M., & Xie, Y. (1994). From cells to cities. Environment & Planning B: Planning & Design, 

2i,S31. 

Bella, I. E. (1971). A new competition model for individual trees. Forest Science, 17(3), 
364-372. 

Benenson, 1. (2007). Warning! The scale of land-use CA is changing! Computers, Environment 

and Urban Systems, 31(2), 107-1 13. 
Borsboom-van Beurden, J. A. M., Boersma, W. T., Bouwman, A. A., Crommentuijn, L. E. M., 

Dekkers, J. E. C, & Koomen., E. (2005). Spatial impressions - Visualisation of future land 

use in the Netherlands. Bilthoven - The Netherlands, Netherlands Environmental Assesment 

Agency-5500 16003/2005 . 
Briassoulis, H. (2000). Analysis of Land use change: Theoretical and modeling approaches. 

Morgantown, WV: West Virginia University. 
Brown, D. G., Page, S., Riolo, R., Zellner, M., & Rand, W. (2005). Path dependence and the 

validation of agent-based spatial models of land use. International Journal of Geographical 

Information Science, 19(2), 153-174. 
Biirgi, M., Hersperger, A. M., & Schneeberger, N. (2005). Driving forces of landscape change - 

Current and new directions . Landscape Ecology, 19(H), 857-868. 
Burnham, B. O. (1973). Markov intertemporal land use simulation model. Southern Journal of 

Agricultural Economics, 5(1), 253-258. 
Chomitz, K. M., & Gray, D. A. (1996). Roads, land use, and deforestation: A spatial model applied 

to Belize. World Bank Economic Review, 10(3), 487-512. 
Couclelis, H. (1985). Cellular worlds: A framework for modeling micro - macro dynamics. 

Environment and Planning A, 17(5), 585-596. 
deMaris, A. (1992). Logit modeling: Practical applications, University of lowa-07-086. 
de Nijs, T. C. M., de Niet, R., & Crommentuijn, L. (2004). Constructing land-use maps of the 

Netherlands in 2030. Journal of Environmental Management, 72(1-2), 35^-2. 
Dungan, J. L., Perry, J. N., Dale, M. R. T, Legendre, P., Citron-Pousty, S., Fortin, M. J., et al. 

(2002). A balanced view of scale in spatial statistical analysis. Ecography, 25(5), 626-640. 
Engelen, G., White, R., Uljee, I., & Drazan, P. (1995). Using cellular automata for integrated 

modelling of socio-environmental systems. Environmental Monitoring and Assessment, 34(2), 

203-214. 

Fearnside,P. M. ( 1 996) . Amazonian deforestation and global warming: Carbon stocks in vegetation 

replacing Brazil's Amazon forest. Forest Ecology and Management , S0(l-3), 21-34. 
Ferrand, N. (1996). Modelling and supporting multi-actor planning using multi-agents systems. 

Santa Barbara, CA: Third NCGIA Conference on G1S and Environmental Modelling. 
Fujita, M., Kragman, P., & Venables, A. J. (1999). The spatial economy. London: The MIT press. 
Gardner, M. (1970). Mathematical Games: The fantastic combinations of John Conway's new 

solitaire game 'life'. Scientific American, 223, 120-123. 
Geertman, S., Hagooit, M., & Ottens, H. (2007). Spatial-temporal specific neighbourhood rules 

for cellular automata land-use modelling. International Journal of Geographical Information 

Science, 2 1(5), 547-568. 
Grimm, V., Berger, U., Bastiansen, F., Eliassen, S., Ginot, V., Giske, J., et al. (2006). A standard 

protocol for describing individual-based and agent-based models. Ecological Modelling, 

198(1-2), 115-126. 



3 Core Principles and Concepts in Land-Use Modelling: A Literature Review 



55 



Hagoort, M. (2006). The Neighbourhood Rules. Land-use interaction, urban dynamics and cellular 

automata modeling (300p). PhD, Faculty of geosciences, Utrecht University, Utrecht. 
Heistermann, M., Muller, C, & Ronneberger, K. (2006). Land in sight?: Achievements, deficits 

and potentials of continental to global scale land-use modeling. Agriculture, Ecosystems & 

Environment, 114(2-4), 141-158. 
Hilferink, M., & Rietveld, P. (1999). LAND USE SCANNER: An integrated GIS based model for 

long term projections of land use in urban and rural areas. Journal of Geographical Systems, 

1(2), 155-177. 

Hunt, J. D., Kriger, D. S., & Miller, E. J. (2004). Current operational urban land-use-transport 

modelling frameworks: A REVIEW. Transport Reviews, 25(3), 329-376. 
Koomen, E., Loonen, W., & Hilferink, M. (2008). Climate-change adaptations in land-use 

planning; A scenario-based approach. In L. Bernard, A. Friis-Christensen, & H. Pundt 

(Eds.), The European information society; Taking geoinformation science one step further 

(pp. 261-282). Berlin: Springer. 
Koomen, E., & Stillwell, J. (2007). Modelling land-use change; Theories and methods. Chapter 1 . 

In E. Koomen, J. Stillwell, A. Bakema, & H. J. Scholten (Eds.), Modelling land-use change; 

Progress and applications (pp. 1-21). Dordrecht: Springer. 
Koomen, E., J. Stillwell, A. Bakema, & H. J. Scholten, (Eds.). (2007) . Modelling land-use change. 

Progress and applications. Dordrecht, The Netherlands: Springer. 
Krugman, P. (1991). Geography and trade (Gaston Eyskens lecture series). Leuven: Leuven 

University Press. 

Krugman, P. (1999). The Role of Geography in Development. International Regional Science 
Review, 22(2), 142-161. 

Kuijpers-Linde, M., Geurs, K. T., Knoop, J. M., Kuiper, R., Lagas, P., Ligtvoet, W., et al. (2007). 
Nederland Later, Tweede Duurzaamheidsverkenning, deel fysieke leefomgeving Nederland. 
Bilthoven. 

Lambin, E. F., Rounsevell, M. D. A., & Geist, H. J. (2000). Are agricultural land-use models able 
to predict changes in land-use intensity? Agriculhire, Ecosystems & Environment, §2(1-3), 
321-331. 

Lambin, E. F., Turner, B. L., Geist, H. J., Agbola, S. B., Angelsen, A., Bruce, J. W., et al. 

(2001). The causes of land-use and land-cover change: Moving beyond the myths. Global 

Environmental Change, 11(4), 261-269. 
Langdon, W. B. (1998). Genetic programming and data structures (350p). MSc, University 

College , London . 

Le Page, C, Bousquet, F., Bakam, I., Bah, A., & Baron, C. (2000). CORMAS: A multiagent 

simulation toolkit to model natural and social dynamics at multiple scales. Wageningen: 

Workshop 'The ecology of scales'. 
Lesschen, J. P., Verburg, P. H., & Staal, S. J. (2005). Statistical methods for analysing the spatial 

dimension of changes in land use and farming systems, International Livestock Research 

Institute LUCC Focus 3 Office. 
Li, X., & Yeh, A. G.-O. (2001). Calibration of cellular automata by using neural networks for the 

simulation of complex urban systems. Environment and Planning A, 33(8), 1445-1462. 
Liao, T. F. (1994). Interpreting probability models. Logit, Probit, and Other Generalized Linear 

Models, University of Iowa-07-101 . 
Ligtenberg, A., Bregt, A. K., & Lammeren, Rv. (2001). Multi-actor-based land use modelling: 

Spatial planning using agents. Elsevier, 56, 21-33. 
Lopez, E., Bocco, G., Mendoza, M., & Duhau, E. (2001). Predicting land-cover and land-use 

change in the urban fringe: A case in Morelia city, Mexico. Landscape and Urban Planning, 

55,271-285. 

LUMOS (2005). Platform for land use modeling in the Netherlands. Lumospro. (2007). Project 

team website, www.lumospro.nl 
Mas, J. F., Puig, H., Palacio, J. L., & Sosa-Lopez, A. (2004). Modelling deforestation using GIS 

and artificial neural networks . Environmental Modelling & Software, 19(5), A6\-41\ . 



56 



J. van Schrojenstein Lantman et al. 



Matthews , R ., Gilbert , N . , Roach , A ., Polhill , G . , & Gotts , N . (2007) . Agent-based land-use models : 

A review of applications. Landscape Ecology, 22, 1447-1459. 
McGarigal, K., & Marks., B. J. (1995). FRAGSTATS: Spatial pattern analysis program for 

quantifying landscape structure. Portland, OR: US, Department of Agriculture , Forest Service, 

Pacific Northwest Research Station-PNW-GTR-351 . 
Meyer, W. B.,& Turner, B. L. (1992). Human population growth and global land-use/cover change. 

Annual Review of Ecology and Systematics, 25(1), 39-61 . 
Moeckel, R., Schurmann, C, & Wegener, M. (2002). Microsimulation of land use. 42nd 

European Congress of The Regional Science Association . Dortmund: Institut fur Raumplanung, 

University of Dortmund. 
Muller, M. R., & Middleton, J. (1994). A Markov model of land-use change dynamics in the 

Niagara Region, Ontario, Canada. Landscape Ecology, 9(2), 151-157. 
Nelson, G. C., & Hellerstein, D. (1997). Do roads cause deforestation? Using satellite images in 

econometric analysis of land use . American Journal of Agricultural Economics, 79(1), 80-88. 
Overmars, K. P., de Koning, G. H. J., & Veldkamp, A. (2003). Spatial autocorrelation in multi-scale 

land use models . Ecological Modelling, 164(2-3), 257-270. 
Overmars, K. P., de Groot, T., & Huigen, M. G. A. (2007). Comparing inductive and deductive 

modeling of land use decisions: Principles, a model and an illustration from the Philippines. 

Human Ecology, 35, 439^152. 
Parker, D. C, Berger,T., & Manson, S. M. (2001). Agent-based models of land-use and land-cover 

change. Adaptive agents, intelligence and emergent human organization: Capturing complexity 

through agent-based modelling. Irvine, CA: LUCC International Project Office. 
Pijanowski, B. C, Brown, D. G., Shellito, B. A., & Manik, G. A. (2002). Using neural networks 

and GIS to forecast land use changes: A land transformation model. Computers, Environment 

and Urban Systems, 26(6), 553-575. 
Pijanowski, B. C, Pithadia, S., Shellito, B. A., & Alexandridis , K. (2005). Calibrating a neural 

network-based urban change model for two metropolitan areas of the Upper Midwest of the 

United States . International Journal of Geographical Information Science, 19(2), 197-215. 
Pinto, N. N., & Antunes, A. P. (2007). Cellular automata and urban studies: A literature survey. 

Architecture, City and Environment, 4,M\-A%d. 
Pontius, R. G., Boersma, W., Castella, J.-C, Clarke, K., De Nijs, T., Dietzel, C, et al. (2008). 

Comparing the input, output, and validation maps for several models of land change . Annals of 

Regional Science, 42(1), 1 1-37. 
Pontius, R. G., Cornell, J. D., & Hall, C. A. S. (2001). Modeling the spatial pattern of land-use 

change with GEOMOD2: Application and validation for Costa Rica. Agriculture, Ecosystems & 

Environment, &5(l-3), 191-203. 
Ricardo, D. (1817). On the Principles of Political Economy and Taxation, Library of Economics 

and Liberty. 

Scopus. (2008). The largest abstract and citation database of research literature and quality web 
sources. 

Silva, E. A., & Clarke, K. C. (2002). Calibration of the SLEUTH urban growth model for Lisbon 
and Porto, Portugal. Computers, Environment and Urban Systems, 26(6), 525-552. 

Sinclair, R. (1967). Von Thunen and Urban Sprawl. Annals of the Association of American 
Geographers, 57(1), 72-87. 

Skapura, D. (1996). Building neural networks. New York: ACM Press. 

Tobler, W. (1979). Cellular geography. In S. Gale & G. Olsson (Eds.), Philosophy in geography 

(pp. 379-386). Dordrecht: Reidel. 
Turner, B. L., Skole, D., Sanderson, S., Fischer, G., Fresco, L., & Leemans., R. (1995). Land-Use 

and Land-Cover Change Science/Research Plan, International Human Dimensions Programme 

on Global Environmental Change-7. 
Veldkamp, A., & Fresco, L. O. (1996). CLUE: A conceptual model to study the conversion of land 

use and its effects. Ecological Modelling, 85(2-3), 253-270. 



3 Core Principles and Concepts in Land-Use Modelling: A Literature Review 



57 



Verburg, P. H., de Koning, G. H. J., Kok, K., Veldkamp, A., & Bouma, J. (1999). A spatial explicit 

allocation procedure for modelling the pattern of land use change based upon actual land use. 

Ecological Modelling, 116(1), 45-61 . 
Verburg, P. H., van Eck, J. R. R., de Nijs, T. C. M., Dijst, M. J., & Schot, P. (2004). Determinants 

of land-use change patterns in the Netherlands. Environment and Planning B: Planning and 

Design, 31(1), 125-150. 

Verburg, P. H., Schot, P. P., Dijst, M. J., & Veldkamp, A. (2004). Land use change modelling: 

Current practice and research priorities. GeoJournal, 61 , 309-324. 
von Thiinen, J. H. (1966). Isolated state: An English edition of Der isolierte Stoat. New York: 

Pergamom Press. 

Waddell, P. (2002). UrbanSim: Modeling urban development for land use, transportation and 
environmental planning. 

Wagner, P., & Wegener, M. (2007). Urban land use, transport and environment models. disP, 3, 
45-57. 

Walker, R. (2004). Theorizing land-cover and land-use change: The case of tropical deforestation. 

International Regional Science Review, 27(3), 247-270. 
Walsh, S. E., Soranno, P. A., & Rutledge, D. T. (2003). Lakes, Wetlands, and Streams as Predictors 

of land use/cover distribution. Environmental Management, 31(2), 198-214. 
Wear, D. N., & Bolstad, P. (1998). Land-use changes in southern appalachian landscapes: Spatial 

analysis and forecast evaluation. Ecosystems, 1(6), 575-594. 
White, R., & Engelen, G. (1994). Cellular dynamics and OIS: Modelling spatial complexity. 

Geographical Systems, 1, 237-253. 
White, R., & Engelen, G. (2000). High-resolution integrated modelling of the spatial dynamics of 

urban and regional systems. Computers, Environment and Urban Systems, 24(5), 383^-00. 
Wrigley, N. (1976). Introduction to the use oflogit models in geography. Norwich: University of 

East Anglia-10. 

Wu, J., & Li,H. (2006). Concepts of scale and scaling. In J. Wu,K. B.Jones, H. Li,& O. L. Loucks 
(Eds.), Scaling and uncertainty analysis in ecology: Methods and applications . New York: 
Springer, p. 6. 



Part II 
Practice 



Chapter 4 

A Sustainable Outlook on the Future 
of The Netherlands 

Rienk Kuiper, Marianne Kuijpers-Linde, and Arno Bouwman 



4.1 Introduction 

In 2005, the Dutch Upper House asked the Dutch Government to prepare a highly 
integrated, long-term investment strategy in spatial planning that should pay more 
attention to the effects of climate change and make provisions for the further 
development of the Randstad conurbation. In response to this request, in 2007 PBL 
carried out the Second Sustainability Outlook. This study comprised two parts, each 
looking at sustainability from a different perspective: 

1. The Netherlands in a Sustainable World; a spatial perspective analysing the 
relationship between the Netherlands and rest of the world (Hanemaaijer et al., 
2008); 

2. The Netherlands in the Future; a time perspective analysing the relationship 
between the Netherlands of today and the Netherlands in the future 
(Kuijpers-Linde et al., 2010). 

The study presented in this chapter deals with the second, temporal perspective 
and addresses the question of the sustainability of the physical living environment in 
the Netherlands. In the present system, political and administrative decisions on the 
various social issues are almost invariably taken from a sectoral, and thus partial, 
viewpoint. This leads to partial solutions and compartmentalisation. This study 
shows that the pursuit of a sustainable Netherlands requires a more far-reaching 
integration of current policies. The planning of housing and employment areas, 
nature, landscape, infrastructure and energy supply are clearly related. These 
often conflicting activities can be accommodated much more effectively - while 
delivering the maximum possible quality of life to future generations - if they are 
looked at together. To accommodate the current demand for land, while ensuring 



R. Kuiper (CE3) 

PBL Netherlands Environmental Assessment Agency, PO Box 30314, 2500 GH The Hague, 

The Netherlands 

e-mail: rienk.kuiper@pbl.nl 

E. Koomen, J. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 61 
Practice, The GeoJournal Library 101, DOI 10.1007/978-94-007-1822-7_4, 
© Springer Science+Business Media B.V. 201 1 



62 



R. Kuiper et al. 



that future generations inherit a high-quality living environment, a more coherent, 
long-term vision is needed. The Outlook study shows how optimising the spatial 
allocation of activities can maximise the sustainability of the Netherlands. After a 
short elaboration of the different stages in this study, the main outcomes and the role 
of tools such as Land Use Scanner are discussed. 

4.2 Researching Sustainability 
4.2.1 Defining Sustainability 

In the First Sustainability Outlook (MNP, 2004) the concept of sustainability was 
not easy to make concrete. Ultimately, sustainability was defined as the distribution 
of a certain quality of life and the possibilities for maintaining that distribution 
in the future. This quality of life is, then, determined by the availability of the 
resources needed to achieve the goals set. It was acknowledged, however, that 
the range of possible underlying objectives to choose from makes sustainability 
a heavy, value-laden concept. In the Second Sustainability Outlook, initially the 
endeavour was not only to examine the physical (i.e. planet) side of the problem, 
but, in collaboration with other policy assessments, also the long-term viability of 
the economic situation and the stability of social relations. Subsequently, however, 
it became clear that the concept of sustainability could only be transformed 
satisfactorily into workable principles for the planet side, in other words, the 
physical living environment. 

As already indicated in our introduction to this chapter, to accommodate 
the current demand for land, while ensuring that future generations inherit a 
high-quality living environment, a more coherent, integrated long-term vision is 
needed. For our study the sustainability of the physical environment was broken 
down into the following six main themes: 

1 . Climate change: flooding risks and water damage, water shortages and salt-water 
intrusion; 

2. Biodiversity (diversity of plant and animal life): connectivity and the quality 
of the National Ecological Network, compliance with the agreement of the 
European Commission to conserve internationally important habitats and species 
(Natura 2000 network); 

3. Accessibility: accessibility of cities, congestion on roads, unequal distribution of 
environmental impacts among different population groups; 

4. Quality of the living environment: shortage of housing in both quantitative and 
qualitative terms (particularly location: green space in urban areas, rural living) 
and affordability; 

5. International business establishment: availability of easily accessible business 
parks, presence of prime office sites, international hubs (particularly Amsterdam 
Airport Schiphol), and attractive residential areas; 



4 A Sustainable Outlook on the Future of The Netherlands 



63 



6. Landscape quality: landscape values that are characteristic for the 
cultural-historic identity of the Netherlands and that are important in relation to 
tourism and recreation. 



4.2.2 Drawing Up Two Reference Scenarios 

After formulating these themes, current social trends were tracked to see whether 
existing goals were being achieved and what policy objectives remained to be 
realised in the future. The resulting picture was called the Baseline scenario, 
taking into account only policies that had been adopted by the Dutch or European 
Parliaments: it is a policy-neutral scenario. From this neutral perspective, past trends 
and patterns of spatial development were translated into maps depicting future 
spatial structures. This analysis was carried out for the continental Netherlands, 
focusing on spatial pressure, probable spatial dynamics and the impact of both. From 
this policy-neutral perspective, the demand for land given moderate economic and 
demographic growth was allocated with the Land Use Scanner model according 
to past patterns of, and trends in, spatial development. Average demographic and 
economic growth was assumed until 2040: more precisely, an economic growth 
of 1.7% per year and a population growth from 16.6 million people in 2007 
to 17.1 million by 2040. This Baseline scenario, representing average spatial 
pressure, is in line with the OECD baseline scenario and was based on the 
Transatlantic Market (A2) scenario in the study 'Welfare, Prosperity and Quality 
of the Living Environment' produced by Dutch assessment agencies in 2006 
(CPB et al., 2006). Because it was felt that the impact of future economic and 
demographic growth might be underestimated, an additional High Development 
Pressure reference scenario was formulated. It assumed annual economic growth 
of 2.1% and population growth from 16.6 million people to more than 20 million by 
2040. 

4.23 Interpreting Spatial Focuses 

Future spatial developments were adjusted to obtain optimal land-use patterns for 
resolving a single persistent policy-related problem that is corresponding to one of 
the six main themes. The new picture of the desired use of space per sector that 
resulted was referred to as a 'focus'. A focus thus reflects a (partial) answer to 
one of the single problems. A focus is a spatial interpretation of a particular line 
of policy. Each focus not only analyses the spatial consequences of policy but also 
its effects on land-use patterns and land management. Graphic depictions of these 
spatial focuses are included in Chapters 1 and 5. 

For each of the six focuses, land use was simulated - provided by Land Use 
Scanner - in such a way that specific, adverse developments were prevented. For 
each focus not only the spatial consequences of particular policies were analysed, 
but also the effects of a specific land-use pattern and land management. For that 



64 



R. Kuiper et al. 



reason, the calculation of land-use dynamics was completed with estimates of 
the associated investment expenditures and maintenance and management costs. 
These steps were repeated for each focus. Finally, the outcomes of the land-use 
simulations for all focuses were compared with the reference scenarios . In this way 
the potential - spatial - conflicts and synergies could be highlighted. These insights 
formed the starting point for investigating alternative strategic actions and policy 
options . 



4.2.4 Comparing Focuses with the Reference Scenarios 

After the simulation of changes in land use and transport, the sustainability effects of 
all scenarios and focuses were assessed, both quantitatively and qualitatively. The 
two reference scenarios reflected business-as-usual conditions, while the focuses 
are optimised for one of the six sustainability themes. Nine indicators were 
designated that cover the six sustainability themes identified at the beginning of 
the study. Subsequently, additional effect-models, for example ecological models, 
combined with expert knowledge, were used to assess the impact on sustainability 
of all land-use simulations and corresponding transport intensities. Eventually 
these assessments were translated into three possible categories: an improvement 
of sustainability for that particular theme; a reduction of sustainability for that 
particular theme; and a stable situation for that theme, meaning no improvement or 
reduction. 



4.2.5 Optimising Land Use in a Combination Map 

Finally, the positive elements of the various focuses were combined as favourably as 
possible, so that ultimately a combined map (known, literally, as the Combination 
Map) was created in which the different focuses, and their land-use functions, were 
optimised for each of the three domains people, planet and profit. This map gives a 
picture of the Netherlands that is more sustainable and 'future-proof and provides 
starting points for drawing up specific strategic options. Other forms of integrated 
maps that provide an even better combination of objectives may be conceivable, 
but these have yet to be found. The Combination Map is not a 'blueprint', rather it 
displays the best conceivable options at the moment. 

In the end, the maps and indicators of the Combination Map were compared to 
both the Baseline scenario and the High Development Pressure scenario to identify 
what needs to be changed in current policies to achieve better performance on the 
six main themes of sustainability that had been identified at the beginning of the 
study. 

In the remainder of this chapter, the main outcomes of the Baseline and High 
Development Pressure scenarios and of the Combination Map are discussed in more 
detail . 



4 A Sustainable Outlook on the Future of The Netherlands 



65 



4.3 Land-Use Dynamics and Their Impact 

The maps of current and future land use according to the Baseline scenario for 
the year 2040 are included in Chapter 1. As already mentioned above, the maps 
of the Baseline and High Development Pressure scenarios reflect actual trends in 
land-use dynamics and are based on modest and high economic and demographic 
growth respectively. The associated demands for land were provided by specialised 
housing, employment and agro-economic models. As these two reference scenarios 
are based on the trend-based spatial policies their outcomes differ only in the amount 
of projected urbanisation. 

The main conclusions based on the outcomes of the simulations with the Land 
Use Scanner for the reference scenarios are: 

• Land use in the Netherlands will change radically in the coming decades. The 
built-up area will increase by 15-26% by 2040 (the range covers differences in 
population and economic growth). The current total size of the built-up area is 
therefore expected to grow by about 25%. 

• The majority of new housing will be sited in flood-prone areas; potential risk of 
flood-related damage will by 2040 be 2 to 3 times higher than today. In total the 
new built-up areas will account for about 25-30% of the economic value of all 
property in 2040. In the reference scenarios, the primary flood defences will have 
reached the required statutory standards by 2020. These improvements in safety 
levels will reduce the risk of flood-related damage in 2020 by a factor of 1 .7 and 
the casualty risk by a factor of 3, as compared with the year 2000. 

• Given the assumptions made about social trends and the effectiveness of 
existing policy, new housing development will take place near the major cities 
of the Randstad and in the province of North Brabant, the result of high 
demand for housing in those regions and an urban compaction policy in spatial 
planning. 

• In the decades ahead, demand for new commercial sites will be concentrated 
mainly on the Randstad. The northern wing of the Randstad, in particular, will 
become part of a global network of cities. Many new commercial sites will be 
developed in this part of the Netherlands because the growth of jobs will be 
highest there. 

• Because under the reference scenarios it is assumed that existing provincial 
spatial plans will be realised and that accessibility via the main road system 
is a significant business location incentive, new employment locations will be 
created and existing ones expanded, particularly in the centres of the large 
cities and their environs. Haarlemmermeer, the Zuidplaspolder, and the area to 
the south of Groningen are examples of locations that will be subject to these 
processes. 

• In the reference scenarios, total mobility will increase by approximately 15-30%, 
with car use rising by 35-45%, between 2002 and 2040. The ranges allow for 
differences in population and economic growth. Accessibility of employment 
locations by car will improve by about 10% during this period, owing to 



66 



R. Kuiper et al. 



T3 

s ^ 

TO -3 



1/1 1/1 

.a -e 



O M) g 

>>.S a 



o- 



i i 



s s 

.3 3 



a r u 



o o 



S TO 



T3 >, g.ss 
'3 s t 9 



•a S ^ iS 



H ri r-t 



3 « s h 
3 — u S 



TO 

aj TO TO 

TO Z 'o 

ri o 

S OS 



3 e >> 



o 
U 



c 
Z 



.5 "a 



O aj 
rr 1 rrl 



C > ^— i en 



c 5 
T3 03 



s -3 i H - 



S 2 
o 
-1 



3 a u ^ g 



P 5 



U TO « 



e ~ 
'3 u 

Sis § 
ft, t; 



.g 

3 



5- 

a, 



2 -2 



S 



bo 

3 















"5 
a 









4 A Sustainable Outlook on the Future of The Netherlands 



67 



investments in the road system and job creation. After 2020, however, the 
accessibility of employment sites - without additional investment and pricing 
policies - will decline, owing to increased congestion and a decrease in the 
number of jobs . 

• The proportion of homes experiencing noise levels of more than 55 dB will have 
increased from 43% today to 46% by 2040. 

• In the reference scenarios, the National Ecological Network will be realised by 
2018 (approximately 730,000 ha). The nature conservation areas in the network 
will remain fragmented, however. Environmental conditions will not improve 
sufficiently, either. Biodiversity goals will, therefore, not be achieved. 

• In the future, the open landscapes so characteristic of the Netherlands will largely 
disappear, because of ongoing urbanisation. 

As already mentioned in the description of the design of the reference scenarios 
(Section 4.2), Land Use Scanner was used to create explicit maps of future land use 
for the focuses, each of which represents one of the six sustainability themes for 
the physical living environment. For a more detailed description of these land-use 
maps, see Kuijpers-Linde et al. (2010). Here, it suffices to say that all land-use maps, 
trend-based scenarios, and focuses were assessed for their impact on sustainability. 
Table 4.1 shows the outcomes of these analyses; comparing the sustainability 
assessments of the six focuses with the Baseline scenario. 



4.4 Optimising for Sustainability 

This section highlights the integration of each of the six sustainability themes (or 
focuses) in the Combination Map. Figure 4.1 gives an overview of all relevant 
policies per sustainability theme. Their elaboration in the Combination Map and 
the assumptions made on the use of geographical space in the land-use simulations 
are discussed below in more detail for each sustainability theme. 

4.4.1 Climate Change 

The Climate change focus shows that the Netherlands can probably withstand 
climate change and rising sea levels for centuries to come and that structural 
measures, such as locating capital investments on the higher ground or substantially 
widening the coastal zone, are therefore not urgently required. The Combination 
Map, thus, assumes further investment in the low-lying areas of the Netherlands, 
particularly in the Randstad. The Map confines itself to a targeted differentiation of 
safety levels to reduce the damage and risk of flood-related casualties while creating 
a robust protection system on the Rhine-Meuse floodplain. Areas with the lowest 
safety standards will be kept free of new urban development as much as possible. 
The introduction of overflow dykes is expected to increase the predictability of any 
flooding and reduce the risk of casualties even further. 




Fig. 4.1 Preconditions for the combination map 



4 A Sustainable Outlook on the Future of The Netherlands 



69 



Although there are many uncertainties surrounding the rate and scale of climate 
change, and with it a rise in sea levels, the analysis presented in Chapter 2 
shows that the diminishing opportunities for unhindered discharge of water from 
its major rivers determine the long-term future of the Netherlands. In the event 
of a two-metre rise in the sea level, other structural solutions may have to be 
found to accommodate main and peak discharges from the Rhine. However, 
two or three centuries would be needed for a rise in the sea level to occur 
that approaches the upper end of the KNMI's estimates. The densely populated 
lower reaches of the Rhine and Meuse rivers, with cities such as Rotterdam 
and Dordrecht on their fringes, are particularly vulnerable (see Fig. 5.1). To 
keep open long-term options for adjusting river discharges and storing water, the 
Combination Map contains areas reserved for water retention in the southwestern 
delta area, areas flanking the courses of the major rivers, and the IJssel valley 
and IJsselmeer areas. These emergency water-retention areas make the Netherlands 
more resilient in the event of any unexpected acceleration in sea level rise during this 
century. 

Low-lying areas of the Netherlands must include areas for extra water storage. 
Some of the deepest polders (reclaimed land), or sections of these polders, as 
included in the Landscape quality focus, would be the most appropriate locations, 
because this would also serve to reduce the intrusion of salt water into the polder 
drainage system, and combat desiccation in surrounding nature conservation areas, 
as well as create additional opportunities for recreation and green residential 
areas. A separate saltwater drainage system would deliver additional options for 
expanding facilities for water-based recreation. Given the fact that options for 
adapting sewerage systems and reserving land for water retention at a later date 
are limited and would also entail higher costs, it has been assumed that new urban 
areas will have a robust design that includes provisions for additional water storage. 
This is also an important requirement for the restructuring of existing urban areas. 
Reservation of areas in the IJsselmeer area for longer term water-retention and the 
development in this area of internationally important nature conservation habitats 
would keep options open for substantial fluctuations in the water-table in that area 
and also reduce vulnerability to drought. 



4.4.2 Biodiversity 

The Biodiversity focus indicates that to comply with the Directive of the European 
Commission to conserve internationally important habitats and species (Natura 
2000 network), the Netherlands needs to strengthen several nature conservation 
areas by enlarging them and reducing the environmental pressure exerted by the 
surrounding areas. The Combination Map incorporates this expansion of the Natura 
2000 areas, as well as measures (including financial compensation) to be taken in 
the buffer zones around those areas. Greater emphasis is placed both on wetlands 
(peat marshes, regional river/stream systems, major water bodies and river areas) 
and the fringes of the Veluwe and the Utrechtse Heuvelrug. 



70 

4.43 Accessibility 



R. Kuiper et al. 



The Accessibility focus shows that with a high degree of clustering and 
intensification, accompanied by the introduction of road-use pricing (based on 
time, place and environmental criteria), the investments needed to achieve the 
accessibility goals that have been set are lower than if only the infrastructure 
is expanded. Improvement in public transport also delivers a significant gain in 
accessibility in addition to that resulting from the introduction of road-use pricing. 
Both instruments have therefore been incorporated in the Combination Map. 

Another advantage of clustering and intensification is that urban development 
takes up less space, which leaves the flexibility of the spatial structure of the 
Netherlands intact, makes it easer to reserve land for flood protection (ability 
to adapt to climate change) and limits impacts on the landscape. This makes 
the concentration of urban development a fundamental principle underlying 
the Combination Map. Clustering and intensification also have negative effects. The 
main ones are a (relative) decrease in the area of green space in and around the cities 
(recreational opportunities) and an increase in noise nuisance. As compensation, the 
Map takes into account the creation of additional areas of surface water and green 
space around the cities as part of the Landscape quality focus . 

4.4.4 Quality of the Living Environment 

The focus of Quality of the living environment performs less well than the Baseline 
scenario against many indicators, because of the greater claims on land and the 
development of urban land uses in attractive landscapes. These are, therefore, 
not included in the Combination Map. To improve the quality of the living 
environment, not only in new development areas, but also in existing housing areas, 
the Combination Map includes additional areas of water and green space around 
cities (see the focus of Landscape quality). 

There is more government intervention reflected in the Combination Map than 
under the Baseline scenario, and urban development is prohibited in attractive 
residential landscapes, such as the National Landscapes and urban buffer zones. 
Consequently, less land is available for this type of housing development. 

4.4.5 International Business Establishment 

The focus of International business establishment assumes that only the northern 
wing of the Randstad is equally attractive to foreign businesses as the top-ranking 
cites of Europe (Barcelona and Munich). A great deal of the urban area in the 
western part of the Netherlands is, therefore, concentrated around Amsterdam. 
However, this has an adverse effect on other parts of the city and on the natural 
habitats, landscapes and water bodies around Amsterdam. This concentration 
around Amsterdam has not been incorporated into the Combination Map because 



4 A Sustainable Outlook on the Future of The Netherlands 



71 



there is unlikely to be such a bias towards Amsterdam in government planning. 
The focus of International business establishment also shows that moving part of 
the capacity of Amsterdam Airport Schiphol to Lelystad Airport (a slight shift to the 
northeast) would improve the quality of the living environment around Amsterdam 
and, on balance, the Netherlands as a whole. This relocation of airport capacity is, 
therefore, incorporated in the Combination Map. 

4.4.6 Landscape Quality 

The focus of Landscape quality assumes a restrictive urban development policy for 
National Landscapes and urban buffer zones this is included in the Combination 
Map. Additional green space around the cities and additional water bodies are 
included in the Map to reduce the negative effects of clustering and intensification 
strategies applied in the accessibility focus. 

The extra emphasis on agricultural landscape management in the focus of 
Landscape quality is included in the Combination Map. This strategy is applied to 
National Landscapes, the peat meadows, a 5-kilometre zone around the larger cities 
and their urban buffer zones. The basic principle here is that the reform of the EU's 
Common Agricultural Policy will enable a substantial shift away from the present 
system of agricultural subsidies to a system of rewards for landscape stewardship 
by farmers in the public interest. This will make it possible to fund agricultural 
landscape management in those areas, as well as additional environmental measures 
in the buffer zones around Natura 2000 areas. The Combination Map also 
incorporates the greater degree of concentration of intensive forms of agriculture 
(greenhouse horticulture, intensive livestock farming) that are part of this focus. 

4.5 Comparing Combination Map with Reference Scenarios 

To identify where land-use developments should take place that differ from current 
trends, the land-use patterns and indicator scores of the Combination Map were 
compared to those of both the Baseline scenario and the High Development Pressure 
scenario. Figures 4.1 and 4.2 present the outcomes of this analysis with respect 
to urban growth, which is essential for many sustainability themes. Figure 4.2 
shows the optimal spatial configuration of new urban built-up areas, taking into 
account the constraints and requirements of all the themes (focuses) integrated in 
the Combination Map but based on the demand for land as assumed in the Baseline 
scenario 

Figure 4.3 shows the optimal spatial configuration of new urban built-up areas 
based on the demand for land under the High Development Pressure scenario that 
assumes higher economic growth rates and a population of almost 20 million people 
by 2040. When comparing the indicators of the Baseline scenario with those of the 
Combination Map, the Combination Map appears to perform much better than 
the Baseline scenario for most indicators. 



72 



R. Kuiper et al. 




Fig. 4.2 Increase in built-up area according to the combination map compared to the Baseline 
scenario 

The effects of the Combination Map score better on many indicators than the 
Baseline scenario does (see Table 4.2). However, not surprisingly, in the High 
Development Pressure scenario, the indicator scores would be significantly lower. 

4.6 Conclusion 

The land-use simulations and associated impact assessments performed for the 
Second Sustainability Outlook lead to the following main conclusions: 

• To provide the necessary new housing, employment, transport infrastructure and 
green space and at the same time maintain the quality of the living environment 
for existing and future generations, a more holistic approach to development and 



4 A Sustainable Outlook on the Future of The Netherlands 



73 




Fig. 4.3 Increase in built-up area according to the Combination Map compared to the High 
Development Pressure scenario 

the environment is needed. Only then will it be possible to meet all the policy 
objectives and create a spatial structure that can rightly be called sustainable. 

• The greatest improvements can be achieved by ensuring a much better fit between 
urban development and infrastructural measures and a better fit between flood 
protection, habitat creation/restoration and landscape development. Important 
gains can also be made through integrated approaches for agriculture and nature 
and landscape quality - and even integrating measures for flood protection 
with those for improving the attractiveness of the Netherlands to foreign 
businesses. 

• Existing policy objectives provide sufficient opportunities for taking concrete 
steps towards achieving more sustainable use of land in the Netherlands. 
However, realising this will depend on the vigorous implementation of these 
policies at the local level and on effective harmonisation with EU policy. 



74 



R. Kuiper et al. 



Table 4.2 Effects of the Combination Map (development pressure on land in line with the Baseline 
scenario) compared with the Baseline scenario. The symbol '+' denotes a better performance of 
the combination map, while '=' indicates similar scores for both 



Sustainability indicator Combination Effects 

map 



Protection against 


+ 


Differentiation in safety levels, use of overflow 


flooding 




dykes, concentration of urban development in 






low-lying areas of the Netherlands with the 






highest safety levels and limited urban 






expansion in areas flanking the major river 


Adaptation to climate 


+ 


Designation of specific areas for future flood 


change 




retention within flood-risk areas and more 






water bodies in and around the cities to 






provide more options for water management 


Biodiversity 


+ 


Expansion of Natura 2000 areas, additional green 






around cities , and landscape management of 






buffer zones National Landscapes 


Accessibility 




Improved accessibility resulting from urban 






compaction policy brings homes and 






employment closer to each other 


Quality physical living 


■ 


More green and water around cities and an 


environment 




improvement in the quality of agricultural 






landscapes; less noise nuisance from 






Amsterdam Airport. Urban intensification 






increases pressure on the quality of the living 






environment 


Conditions international 


+ 


More green and water around the cities, and 


business 




improvements in the quality of agricultural 


establishment 




landscapes; less noise nuisance from 






Amsterdam Airport 


Spacious and green 




In and around the Randstad, there is some scope 


living 




for developing new green living environments 


Quality of the landscape 


+ 


Areas with high landscape quality kept free of 






urban development and agricultural 






intensification; more water and green around 






the cities , more management of the 






agricultural landscape and reorganising of 






dispersed greenhouse horticulture and 






intensive livestock farming 


Spatial segregation 




Effect barely differs from structural trend 


Clear spatial planning 


policies and enforcement of adopted land-use plans are 



important requirements for creating a sustainable living environment. The new 
Spatial Planning Act allows the preparation of spatial visions that can prioritise 
the spatial opportunities and conflicts in the case of developments of national 
importance, accompanied by an enforceable implementation agenda. From a 
sustainability perspective, benefits that will accrue in the more distant future 
should weigh more heavily when political decisions are taken. 



4 A Sustainable Outlook on the Future of The Netherlands 



75 



• Clustering and intensification, as applied in the accessibility focus, will deliver 
considerable accessibility benefits. If the rise in mobility and congestion levels 
is modest, these gains will be greater than those that would be realised through 
the investments in the road network proposed by the Mobility Policy Document. 
Investment in infrastructure is more efficient if it is made according to a 
sequential approach: spatial policy (clustering and intensification) - road-use 
pricing - physical expansion of infrastructure. 

• The Randstad will increase in greatly in size, particularly under the High 
Development Pressure scenario. The addition of new urban areas will then form a 
wider ring than is currently (2010) the case: running from Amsterdam to Leiden, 
The Hague and Rotterdam, via the Brabant linear conurbation to Nijmegen, 
Arnhem and on through Ede, Amersfoort and Utrecht to Almere and Amsterdam. 

• In the Combination Map, some of the greenhouse horticulture complexes now 
situated in the western part of the Netherlands will move to the fringes of 
the Randstad (south of Rotterdam). To alleviate pressure on available space in 
the west, consideration could be given to designating areas elsewhere in the 
Netherlands for greenhouse horticulture. 

• There are many opportunities for combining creation of habitat with spatial 
policies for the river areas (land reserved for widening the river IJssel and for 
the Kampen and Dordrecht bypasses), the Dsselmeer area and the eastern half 
of the Green Heart. Possible measures in the latter region include: inundation of 
the lowest-lying polders (reclaimed from lakes); limited flushing with fresh water 
to remove intruded salt water; and peat marsh development. 

• Clear spatial planning policies that will bring the price of agricultural land under 
control, are a necessary precondition for ensuring the continuity and development 
of land-based agriculture that is needed to manage the cultural landscapes 
(National Landscapes) and for providing effective buffer zones around Natura 
2000 areas. Financial compensation is necessary for agricultural landscape and 
environmental management in all these areas . The necessary conditions can be 
met through a combination of EU agricultural subsidies and national policies for 
the National Landscapes and Natura 2000 areas. The reform of the EU's Common 
Agricultural Policy (CAP) in 2013 is an essential, although uncertain factor here. 
Co-funding will be needed and will have to come from national as well as EU 
sources. 

4.6.1 Policy Actions for the Short and Long Term 

Both short-term and long-term policy initiatives can be used to achieve greater 
integration of sectoral policies, which in turn will deliver sustainability gains. 
Implementing these initiatives will not require any fundamentally new policies. The 
National Spatial Strategy and many other policy documents already contain many 
policies that lean in this direction. The National Spatial Strategy Monitor does show, 
however, that we cannot always be sure that policy objectives can be achieved in 
practice. Indeed, the Second Sustainability Outlook also demonstrates that stronger 



76 



R. Kuiper et al. 



policies will be needed, particularly for flood safety in the longer term and to meet 
international nature conservation commitments. 



4.6.2 Merits of Land Use Scanner 

As has already been noted, cohesion at various levels and between various parts 
plays a central role in discussions about the way forward to achieve a more 
sustainable future for the Netherlands. The Land Use Scanner model made it 
possible to literally 'map' the future spatial dynamics of the Netherlands, thus 
enabling analysis of these dynamics and their interrelations at a national scale, as 
well as an assessment of their impact. A major advantage above exploration of the 
future based solely on expert judgment is the quantitative character of the Land Use 
Scanner model. Land Use Scanner can allocate the demand for land within regions, 
which makes regional differences in spatial pressure literally visible. The decision 
to compute not only the effects of a Baseline scenario, but also those of a High 
Development Pressure scenario was advantageous because it made clear that spatial 
patterns of land use, in particular that of urban growth, are very different from the 
Baseline scenario. 

Nevertheless, the quality of the outcomes clearly depends upon the quality of 
the input in terms of data on the demand for land, GIS data, scientifically validated 
rules for allocation, and expert judgment. For that reason, a continuous comparison 
of model outcomes with findings in the scientific literature and expert knowledge 
is imperative, with the result that the process of computation is quite labour 
intensive . 

Another point is the improvement in the understandability of grid cell maps. 
In communication with policy-makers and stakeholders, it appears that sketches 
conveying only the main outcomes work better than detailed maps of grid cells. 

The Netherlands is only a small country, but it still needs to find space for 
housing, employment and transport, while simultaneously maintaining the quality 
of the living environment and green spaces. The best way to use the available 
space as effectively as possible is to view these functions and features as a 
cohesive whole, including the additional water-management problems resulting 
from climate change. This study shows how optimising the spatial allocation of 
various land-use functions can maximise the sustainability of the Netherlands in the 
future . 

References 

CPB, MNP, & RPB (2006). Welvaart en Leefomgeving. Een scenariostudie voor Nederland in 

2040 (Prosperity, wellbeing and quality of the living environment. A scenario study for the 

Netherlands in 2040) CPB/MNP/RPB , Bilthoven/The Hague. 
Hanemaaijer, A., de Ridder, W., Aalbers, T., et al. (Eds.). (2008). The Netherlands in a sustainable 

world: Second sustainability outlook. Bilthoven: PBL Netherlands Environmental Assessment 

Agency. 



4 A Sustainable Outlook on the Future of The Netherlands 



77 



Kuijpers-Linde, M., Kuiper, R., Geurs, K., Knoop, J., Lagas, P., Ligtvoet, W., et al. (Eds.)- (2010). 
The Netherlands in the future, second sustainability outlook, the physical living environment in 
The Netherlands . Bilthoven: PBL Netherlands Environmental Assessment Agency. 

MNP (2004). Quality and the future. Sustainability outlook. Bilthoven: PBL Netherlands 
environmental assessment agency. 



Chapter 5 

Coupling a Detailed Land-Use Model 
and a Land-Use and Transport 
Interaction Model 

Barry Zondag and Karst Geurs 



5.1 Introduction 

As already described in the preceding chapter, the study 'The Netherlands in the 
Future; The Second Sustainability Outlook' (MNP, 2007) constructed alternative 
future spatial strategies for the Netherlands and evaluated them using a diverse set 
of sustainability indicators, such as flooding safety, biodiversity, accessibility and 
landscape protection. As in previous studies, Land Use Scanner was used here as an 
instrument to allocate demand for land of different land-use types within a region. At 
a regional level, this demand had been calculated earlier by sector-specific models 
for the Baseline scenarios. The analytical framework was further extended with the 
Tigris XL model, a Land Use and Transport Interaction model (Significance & 
Bureau Louter, 2007). The specific objectives for using the Tigris XL model in 
addition to Land Use Scanner were: 

• To simulate transport endogenously and analyse the impacts of joint alternative 
spatial and transport strategies on transport indicators (such as congestion) and 
accessibility indicators (in both geographical and monetary terms); 

• To simulate and analyse the mutual influence of land use on transport and 
transport on land use; 

• To calculate the impacts of alternative land-use and transport strategies on the 
housing and labour market at a regional level. 

Land Use Scanner and Tigris XL were applied for the first time together within 
one analytical framework in 'The Netherlands in the Future' project. The interaction 
between the models was set up within the project by following a pragmatic approach 
that used the strength of both models without making fundamental changes to either 
of them. Generally speaking Land Use Scanner provides more detail about the 



B. Zondag (El) 

PBL Netherlands Environmental Assessment Agency, PO Box 303 14, 2500 GH The Hague, 

The Netherlands 

e-mail: barry .zondag@pbl.nl 

E. Koomen, J. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 79 
Practice, The GeoJournal Library 101, DOI 10.1007/978-94-007-1822-7_5, 
© Springer Science+Business Media B.V. 201 1 



80 



B . Zondag and K. Geurs 



supply side of land by distinguishing more land-use types and contains detailed 
characteristics related to location, for example spatial planning policies. Apart from 
the incorporation of transport, Tigris XL provides more detail to inform its demand 
modelling, for example by simulating choice of location by different household 
types and economic sectors . 

However, to be able to successfully link both instruments within one analytical 
framework, it is important that the consistency between Land Use Scanner and 
Tigris XL increases in terms of classifications, spatial units, and time horizons used. 
To improve the consistency, input data for the two models was harmonised as far as 
possible. For some data categories, an adjustment of the classification is also needed. 
In addition to harmonizing the data, alternative ways of interaction between the two 
models were tested to see how the models might make best use each other's merits. 

Following this short introduction, a short description of the Tigris XL model and 
its characteristics is given in Section 5.2. For a description of Land Use Scanner 
reference is made to Chapter 1 of this book and previous papers (e.g. Hilferink 
and Rietveld, 1999). Section 5.3 describes how Land Use Scanner and Tigris XL 
have been linked. A description of an actual application of the combined framework 
follows in Section 5.4. In Section 5.5 conclusions are presented regarding the 
interaction between the two models, the contribution of the combined framework 
to policy-making and recommendations for further improvement of the framework. 

5.2 Short Description of Tigris XL 

In the period 2002-2005, a consortium of Significance (part of RAND Europe at the 
time) and Bureau Louter developed the Tigris XL model in a number of sequential 
projects for the Transport Research Centre in the Netherlands (part of the Ministry 
of Public Works, Transport and Water Management). These projects consisted of 
a combination of model development, model application and model testing. Tigris 
XL is a so-called Land Use and Transport Interaction (LUTI) model, which uses the 
National Transport Model (LMS) of the Netherlands as its transport module (RAND 
Europe & Bureau Louter, 2006a; Zondag, 2007). 
The key characteristics of the Tigris XL model are: 

• It is a dynamic land-use model simulating time steps of 1 year. This enables it to 
simulate path dependency and to analyse how the system evolves over time. The 
key reason for this approach is that, due to many different time lags in the system 
(economic cycles , planning cycles , construction time, etc .), a general equilibrium 
does not exist in land use. Within its incremental structure the model uses partial 
equilibrium conditions, for example to match supply and demand within a year 
on the housing market. 

• The module for choice of residential location has been statistically estimated, 
for six different household types, based on a large housing market survey (over 
100,000 respondents). The module for choice of location of firms (represented as 
jobs in the model) has been estimated at municipality level for seven economic 
sectors on time-series data for the period 1986-2000. This enables the model to: 



5 Coupling a Detailed Land-Use Model and a Land-Use and Transport. 



81 



o estimate the relationship between location of jobs and location of residents. 

Tigris XL does not use a pre-assumed hierarchical relationship but rather 

one that, based on estimation results, varies between economic sectors and 

household segments; 
o statistically estimate, based of revealed preference data, the influence of 

accessibility on the distribution of households and jobs. 

• The land-use and transport system is simulated in Tigris XL in a set of linked 
modules addressing specific aspects of the systems (e.g. demography). The 
overall architecture determines the sequences of the modules and data exchanges 
between the modules. This allows flexibility to change or re-estimate specific 
modules without the need to change the whole framework. 

• The land-use model operates in a way tailored to the National Transport model 
(LMS) of the Netherlands. The spatial detail of the land-use model is at the level 
of transport zones . 

• The model has a three-layer structure, namely land, objects (such as dwellings) 
and activities (people, jobs). 

• The land-market module in Tigris XL has different options to simulate the 
influence of government on spatial development. Depending on its settings it can 
simulate new urban development in three ways: 

o as a free-market development following the preferences of residents and firms. 

Location choices are only restricted by the lack of available land or possible 

spatial planning restrictions (such as nature areas); 
o as regulated development via designated locations and numbers of houses (in 

this case only the location choice of residents is simulated with the model); or 
o as an intermediate variant that takes development plans as a starting point but 

adjusts them within a certain range to actual market demand. 

Tigris XL consists of five modules, which together address demography and 
spatial markets. Figure 5.1 presents an overview of the model and the main 
relationships between the modules. The model distinguishes two spatial-scale 
levels: the municipality level (approximately 450 regions) and local transport 
zones of the National Model System (LMS sub-zones; 1,308 sub-zones cover the 
Netherlands). 

Core modules in the model are the housing market and labour market modules. 
These account for the effect of changes in transport on residential or firm location 
behaviour and in this way link changes in the transport system to changes in land 
use. A land and real-estate module simulates supply constraints arising from the 
amount of available land, land-use policies and construction. The module defines 
different levels of government influence on spatial development, ranging from 
completely regulated towards free market, and various feedback loops between 
demand and supply are also available. A demographic module is included to 
simulate demographic changes at the local level. At the national level, the model's 
output is consistent with existing demographic and socio-economic forecasts for 
population, labour force, income levels and employment. 



H2 



B . Zondag and K. Geurs 



Municipality 



Regional 
Labour 
market 



Labour 
market 



Regional 
workforce 



LMS subzone 



Firms, jobs 



I 



Real estate 
market 



Transport 
market 



Household/ 
persons 



I 



Housing 
market 



Demography 




Fig. 5.1 Functional design of the Tigris XL model 



The demographic module addresses processes such as births, deaths and ageing 
of the population, as well as changes in the composition of households. It deals 
with persons by category (gender, age) as well as households by category (size, 
income, etc.) - The demographic module operates at the local subzone level and also 
processes the spatial distribution of scenario-based international migration flows. 

The land-market and real-estate market module processes changes in land use 
and buildings, office space and houses, and addresses both densification of houses or 
floor space within existing urban areas and green-field developments. The changes 
within the existing urban area are exogenously specified and only administrated 
within this module. The simulation of the land-use changes in non-urban land 
can be modelled exogenously or endogenously depending on the setting for the 
level of market regulation. As said before, this can vary from a regulated land-use 
planning system, using exogenous input on the size and location of development 
sites, to a non-regulated market endogenously calculating the size and location of 
development sites . 

The aim of the housing market module is to simulate annual moving (if any) 
of households. The housing market module simulates two choices: the choice to 
move or stay, and the choice of residential location following a move. The choice 
of residential location has a nested logit structure and incorporates a regional and 
a local scale level: there is a choice of region and a choice of location within 
the region. Choices depend on household characteristics, local amenities, prices, 
accessibility and distance (travel time) between the new and old residence. The 
parameters of the move/stay and residential location choice function, for each 



5 Coupling a Detailed Land-Use Model and a Land-Use and Transport. 



83 



household type, have been estimated from a large four-annual housing market 
survey of more than 100,000 households in the Netherlands. 

The labour market module in Tigris XL models changes in number of jobs by 
seven economic sectors and changes in workforce at a regional and zone level. For 
each sector, the influence of accessibility on the spatial distribution of employment 
has been modelled in combination with a set of other explanatory variables. The 
parameters have been estimated on a historical data set (1986-2000) including 
employment figures by sector at a municipality level. The labour market module 
interacts with the demographic, land and real-estate, housing market and transport 
modules . 

The transport module calculates changes in transport demand and accessibility 
and is integrated with the National Transport Model (LMS). The LMS consists 
of a set of discrete-choice models for various choices in transport such as round 
trip frequency, mode of travel and destination, departure time and route. The LMS 
is based on micro-economic utility theory, enabling the derivation of utility-based 
accessibility measures . 

Since its finalisation, Tigris XL has been used to evaluate a variety of transport 
and spatial policies. Examples of the application of the model for evaluating policy 
issues include the evaluation of a new road or rail infrastructure, transport pricing 
policies, land-use plans and urbanisation strategies (RAND Europe & Bureau 
Louter, 2006b; Significance & Bureau Louter, 2007, 2009; Significance et al., 2007) 
The model has been applied for the evaluation of the following issues: 

• Effects of long-term socio-economic scenarios on transport and land use, 
including the translation of socio-economic trends and land-use plans into 
socio-economic input data for the transport model; 

• Land-use effects of transport policies, including road and public transport 
infrastructure and operations, as well as pricing policies; 

• Effects of alternative land-use policies on land use and transport: 

o for new residential or commercial development sites, the model can use 

different assumptions for the regulation of the land market; 
o urban densification strategies; 

• Effects on land-use and transport of integrated land-use and transport strategies. 



5.3 Linking Tigris XL and Land Use Scanner 

This section describes how the coupling between Tigris XL and Land Use Scanner 
has been set-up. It discusses which information is exchanged, the conversion and/or 
processing steps and what modifications have been made to the Tigris XL model. 
First a brief overview is presented of the main differences and overlaps between 
the two models . This overview was needed to design the proper level of interaction 
between the models . 



84 



B . Zondag and K. Geurs 



5.3.1 Overview of Differences and Overlaps 

When Tigris XL and Land Use Scanner are compared it can be concluded that: 

• Both models have different allocation mechanisms. In Land Use Scanner, 
regional claims for housing, labour, etc., are allocated to grid cells by 
a constrained logit model specifying the highest utility. The allocation is 
constrained by land-use claims provided at the regional level. The Tigris XL 
model allocates national projections to the regional and zone level by simulating 
the demand preferences of residents and firms. In both models, supply-side 
restrictions, depending on land characteristics and spatial planning, influence the 
location choices. This occurs in Tigris XL at the zone level and in Land Use 
Scanner at the grid cell level. 

• Land Use Scanner uses exogenous land claims at a regional level as input and 
land use by grid cell (100 m x 100 m) as its output. This means that the model 
simulates only changes in land use within regions and not processes leading 
to land-use changes in other regions (e.g overflow effects from large cities to 
adjacent regions). Tigris XL uses national demographic and economic scenario 
projections as input and simulates both the inter-regional and intra-regional 
changes in population, employment and land use. 

• The Tigris XL model simulates land-use changes resulting from urban 
development or government plans but does not simulate changes between 
different types of non-urban land-use functions. Land Use Scanner simulates 
changes in urban and several non-urban land-use functions, including agriculture 
and nature. The model also simulates non-urban transitions in land use. 

• A nation-wide classification system exists to classify residential areas in the 
Netherlands by type of neighbourhood, for example, countryside or urban centre. 
In Tigris XL, the classification of residential area types is a fixed exogenous input 
per zone. Land Use Scanner calculates endogenously changes in residential area 
types for the cells (e.g. from non-central into central urban area). This means 
that in future years Land Use Scanner will classify the grids (and indirectly the 
zones they are located in) differently from Tigris XL (which uses percentage of 
residential area types at zone level). 

The key characteristics of the two models are summarised in Table 5.1 . 

5.3.2 Set-Up of Interaction 

For the 'Second Sustainability Outlook on the future of the Netherlands' project 
(MNP, 2007) Tigris XL and Land Use Scanner were loosely coupled (see Fig. 5.2). 
In this set-up the models operate individually and interactions between the models 
take place by exchanging input and output through data files. By setting up these 
interactions, some modifications have been made to Tigris XL to take advantage 
of data available in Land Use Scanner and to improve the consistency in data 



5 Coupling a Detailed Land-Use Model and a Land-Use and Transport. . . 



85 



Table 5.1 Key characteristics of land use scanner and Tigris XL at various spatial and temporal 
resolutions 



Spatial and temporal resolution Land use scanner 



Tigris XL 



National level 



Regional level (e.g. 12 
provinces, 40 COROP 
regions or over 400 
municipalities) 

Zonal level (1 308 LMS 
sub-zones for the 
Netherlands) 



Local level (100m x 100 m 
grid cells) 



Transport network 



Temporal resolution 



Uses scenario input directly in 
definition of local suitability 
and indirectly as input for 
the sector models calculating 
the regional claims 

Exogenous input of land 
claims for employment, 
housing and other sectors 
can be provided at any 
regional level 

Not applicable 



Simulates land-use change for 

the cells using the regional 

claims and local 

(multi-criteria) suitability 

definition 
Static accessibility indicators 

are used to describe local 

suitability, no dynamic 

network is included 
2010,2020,2040 



Uses scenario input 



Provides an initial spatial 
distribution of residents and 
employment at level of 
municipalities 

Simulates spatial distribution 
of residents and 
employment, land-use 
change and changes in 
accessibility 

Not applicable 



Calculates transport indicators 
such as travel times, number 
of vehicles and congestion 

Yearly until 2040 



Output Land Use Scanner 

(housing locations, 
suitability maps by grid cell) 



Land Use Scanner 



Output TXL 

(land claims for housing, 
commercial sites by 
COROP region ) 



TIGRIS XL 



Land Use Scanner 



Transport and 
accessibility indicators 



Land-use indicators 



Fig. 5.2 Interaction between Land Use Scanner and Tigris XL (TXL) 



classes between the models. In addition, post-processing procedures have been 
developed to transform Tigris XL output such as residents, houses and employment 
by sector into regional land-use claims, which are used as input data for Land Use 
Scanner. 



86 



B . Zondag and K. Geurs 



The manner in which Land Use Scanner and Tigris XL are coupled depends 
on which type of housing market is assumed in the particular application. The 
housing market in the simulation can either be a regulated market, with fully planned 
development of all new housing construction, versus a more free market, consisting, 
for example, of a zoning policy and market-driven construction of houses. In the 
first case, the location or zone of green-field housing construction on agricultural 
land is determined exogenously in Tigris XL. Output of Land Use Scanner for 
2020 and 2040 related to new residential areas is used as input data (in terms of 
hectares and number of houses) for Tigris XL. Land Use Scanner output data is 
processed by aggregating the cell data to the zone level and by distributing the 
simulated developments into time steps of 1 year. In the second case, the demand 
preferences of households, as simulated in the module for choice of residential 
location in the Tigris XL model, influence the location of housing construction. 
The resulting population developments also influence the location of new economic 
activities, which is calculated in the labour market module of Tigris XL. Both the 
residential and employment developments are then converted into land claims at the 
regional level and used as input for Land Use Scanner. Both types of coupling are 
discussed in more detail in Section 5.3.4. 

5.3 J Use of Land-Use Data and Restriction Maps 

Several changes were made to the land-use classification and data of Tigris XL to 
improve consistency with Land Use Scanner and to take advantage of the detailed 
land-use information in the latter. The following changes were made: 

• The land-use category of housing in Tigris XL was split into three residential 
area types similar to those used in Land Use Scanner. At the housing stock level 
in Tigris XL, a classification of five residential area types is used, which can be 
simply aggregated into the three categories of Land Use Scanner. 

• The land-use type of agriculture in Tigris XL was split into agriculture and 
horticulture, based on data from Land Use Scanner. The two categories needed 
to be distinguished to reflect differences in the cost of acquiring land for urban 
development. 

• Maps restricting the options for development in Land Use Scanner, for 
example those related to external safety, noise pollution, nature protection and 
hydrological constraints, were used in Tigris XL to exclude locations as options 
for development. 

5.3.4 Interactions for Differing Housing Market Circumstances 

This subsection describes in more detail two alternatives ways in which Land Use 
Scanner and Tigris XL have been linked. The preferred way of coupling the two 
systems depends upon assumptions made about housing market conditions. If the 



5 Coupling a Detailed Land-Use Model and a Land-Use and Transport. 



87 



housing market is more regulated and thus supply-driven then the detailed supply 
information from Land Use Scanner plays a more dominant role, but if the housing 
market is more demand-driven then the demand modelling in Tigris XL plays a more 
dominant role. It is assumed that the location choices of firms will not be strongly 
regulated and in both of these Land Use Scanner-Tigris XL coupling options, the 
labour market module in Tigris XL has been used to calculate the spatial distribution 
of employment. 

In cases of regulated or planned housing supply, the supply information on new 
residential locations as simulated per grid cell in Land Use Scanner for 2020 and 
2040 are exported to fill the planning files of Tigris XL by year, zone and residential 
area type. The housing market module in Tigris XL uses houses as supply unit 
instead of hectares of residential land and the number of newly constructed houses 
by zone are calculated based upon the hectares of new residential construction by 
type (e.g. central urban area, green urban area, countryside etc.) and region-specific 
density figures by type (ABF, 2006). Because Land Use Scanner simulates changes 
in residential land use and not in houses, the model is better fitted to simulate urban 
expansion than urban restructuring. For that reason, in the Second Sustainability 
Outlook study, housing construction in currently built-up areas was subtracted from 
the total regional housing construction target that was underlying the regional land 
claims for housing. In Tigris XL the number of houses to be constructed in currently 
built-up areas is contained in a separate input file at the zone level. Then, this number 
of houses in currently built-up areas is made consistent with the total regional 
housing targets, minus the housing developments realised at the new residential 
locations. In the scenarios, the share of brown field and green field developments 
differs depending on the strategy chosen for urban development. 

Both the housing construction within and outside existing built-up areas is input 
for the housing market module in Tigris XL. The newly constructed houses are 
included as vacant houses by zone for the supply side of the module for choice 
of residential location. This module simulates the housing moves and location 
choices of the households. The module for choice of residential location and the 
demographic module together calculate for each year the changes in population (by 
age and gender) and households (by size and income class). 

The labour market module in Tigris XL calculates the spatial distribution of 
employment for seven economic sectors, in the first step at a municipality level 
and in a second step at the level of the zones. The results (statistical regressions) 
of this module show that overall the spatial distribution of people influences the 
location choices of firms, although the existence and strength of this relationship 
differs strongly by economic sector (RAND Europe & Bureau Louter, 2006a). 
The changes in employment by sector, calculated by Tigris XL, are used as input 
for a post-processing step to calculate the regional claims for commercial and 
industrial land. This post-processing module was developed by Bureau Louter and 
for a description reference is made to Significance and Bureau Louter (2007). 
This step uses, besides the calculated changes in employment, information on 
location preferences of the sectors (e.g. for commercial sites), assumptions on the 
changes in these preferences in the future, and developments in average land use per 



88 



B . Zondag and K. Geurs 



employee per sector. These factors together determine the future land-use claims for 
commercial and industrial sites, public facilities and airports and sea ports. These 
land-use claims are calculated for the years 2020 and 2040 at a regional level, which 
is needed as input for Land Use Scanner. 

In cases where housing supply is market (demand) driven, an alternative housing 
construction module is used in Tigris XL in which the housing construction rates 
at the level of zones are based on the location preferences of the household and the 
availability of land. To take better advantage of the detailed supply information in 
Land Use Scanner, the housing construction module in Tigris XL has been modified 
to be used in the integrated framework. In the new set-up, the main differences with 
the other approach are that demand surplus is now calculated specifically for each of 
the three residential area types and aggregated to the municipality level. Information 
on the highest suitability scores by residential area type in Land Use Scanner was 
used at the zone level to allocate housing construction within the municipality to the 
zones (see Fig. 5.3). 

Under both market conditions, the Tigris XL model is used to generate the 
land-use claims for employment. The labour market module in Tigris XL calculates 
the changes in employment and a post-processing procedure module is used to 
convert these changes in employment into land-use claims for commercial and 
industrial land . 

Land-use claims (in hectares) computed using output from Tigris XL at the 
COROP regional level were used as input for Land Use Scanner. Projected Tigris XL 



Suitability indicator 
for three neighbourhood 
types (Land 
Use Scanner) 



Housing construction module 
free market - LUSfTXL 



I 



Construction of new 
houses by zone 
depending on 
demand surplus by 
type and suitability 



Demand surplus 
three neighbourhood 
types /zone (TXL) 



Aggregated at 
municipality level 



Add new 
houses to zones 



1 



Reduce available land 



Land-use market type 



Fig. 5.3 Flow diagram of the adjusted housing construction module in Tigris XL (TXL) 



5 Coupling a Detailed Land-Use Model and a Land-Use and Transport. 



89 



demands for housing and employment by economic sector at LMS zone level were 
translated into land-use claims and subsequently used as input for the alternative 
land-use scenarios. 

5.4 A Policy Application of the Combined Framework 

The Land Use Scanner-Tigris XL framework was used in the Second Sustainability 
Outlook study to calculate the land-use and sustainability impacts of alternative 
scenarios and variants (MNP, 2007). The scenarios consisted of alternative future 
spatial and transport strategies for the Netherlands. One of the alternative land-use 
scenarios, the Uplands variant, is described here to illustrate the functioning of the 
combined framework. 

The Uplands scenario entails a radical break with past trends in spatial 
development in the Netherlands. Under the scenario, new housing and employment 
areas, in the period from 2010 to 2040, are relocated from the low-lying, highly 
urbanised western part of the Netherlands (the Randstad area) to more peripheral 
areas lying above sea level and outside flood-prone areas near rivers. The Uplands 
scenario was developed to represent an extreme climate adaptation scenario. With 
current knowledge on the effects of climate change, the Netherlands is expected 
to be 'climate-proof and protected from rising sea levels for some centuries to 
come. Structural spatial measures, such as a shift in investment to the upland areas 
of the Netherlands, are thus not urgently required. Yet, these were included in 
this scenario to show the potential implications of such drastic measures that are 
regularly suggested in adaptation studies . 

To simulate the Uplands scenario, the housing construction module (see Fig. 5.3) 
used input data from both Tigris XL and Land Use Scanner. This scenario results 
in a markedly different land-use development compared to the Baseline scenario, 
which follows as much as possible current plans. Table 5.2 shows the differences 
in population development between the two scenarios; the country is divided into 
three parts - the urban core (Randstad), a surrounding intermediate area and the 
periphery in the far North, East and South of the country. The table shows that the 
strategy underlying the Uplands scenario results mainly in a shift in population from 
the urban core towards the higher, intermediate area. The periphery is also mainly 
located above sea level but this area does not attract many additional residents due 



Table 5.2 Differences in population development according to the Uplands variant (in 2040) in 
different parts of The Netherlands 







Difference Uplands 




Part of the 


Population baseline 


variant minus baseline 


Percentage population 


Netherlands 


scenario 


scenario 


change in 2040 


Randstad 


7,826,331 


-872,141 


-11% 


Intermediate zone 


5,005,661 


775,355 


15% 


Periphery 


4,306,562 


97,091 


2% 



90 



B . Zondag and K. Geurs 




to its relative poor accessibility to employment and services as compared to the 
Baseline scenario. 

Figure 5.4 shows the population developments for the Uplands variant in the 
period 2010-2040 by region. The map shows clearly that the population in the area 
below sea level declines as a result of the assumed construction stop in combination 
with a falling average household size. Population growth takes place mainly in the 
regions above sea level and those bordering the urban core. Especially the province 
of Brabant, in the south of the country, can expect to experience high population 
growth. Areas in the middle of the Netherlands, where the river Rhine crosses 
the country from Germany to the North Sea, can expect lower population growth 
because the underlying strategy limits building options in this area due to increased 
risk of flooding. Although population development in the periphery is somewhat 
higher in the Uplands scenario than the Baseline scenario (see Table 5.2), there are 
regions in the periphery that nevertheless can expect a decline in population in the 
period 2010-2040. 

Unlike residential developments, labour market developments are not restricted 
in the Uplands scenario. This assumption reflects Dutch planning practice, in which 
the construction of new houses is historically strongly regulated, while the location 
of employment is far less regulated due to oversupply of commercial real estate 
and competition between different regions. The large population shift from the 
Randstad to the intermediate zone is also reflected in the spatial growth pattern 
for employment, as presented in Fig. 5.5. The labour market module simulates 



5 Coupling a Detailed Land-Use Model and a Land-Use and Transport. 



91 




seven economic sectors and employment in most of these sectors, as consumer 
services or government, is influenced by population developments. Other sectors, 
such as commercial services, that respond to changes in employment are indirectly 
influenced by the population through changes in the employment resulting from the 
population following economic sectors . 

Figure 5.6 presents the high-resolution land-use projections for the Uplands 
Scenario from Land Use Scanner, taking the population and employment 
developments at the COROP level from Tigris XL. 

The integration of Land Use Scanner and Tigris XL makes it possible to calculate 
transport impacts and to include the mutual influences of changes in land use on 
transport, and vice versa. Different accessibility indicators were calculated referring 
to different definitions of accessibility. The congestion indicators show that the shift 
in population growth from the urban core to the intermediate area results in overall 
lower losses in traffic congestion time in 2040. Of course this pattern is regionally 
diverse, with a decline in congestion in the urban core and a strong increase in the 
intermediate zone. Applying a geographical accessibility indicator, representing the 
number of jobs that can be reached by car (during off-peak and peak hours) and 
public transport, shows a decline in the number of jobs that can be reached for the 
Uplands scenario as compared to the Baseline scenario. 

As mentioned above, the Tigris XL model incorporates the LMS transport model 
of the Netherlands as its transport module. The LMS is a discrete choice type of 
transport model, which makes it possible to calculate the monetary accessibility 



92 



B . Zondag and K. Geurs 




impacts following the logsum method (see Geurs, Zondag, De Jong & De Bok, 
2010). The logsum method calculates, at a disaggregate level of person types, the 
differences in utility between a policy variant and a reference scenario. The cost 
coefficients in the utility functions can be used to convert the changes in utility 
into monetary terms. The advantage of the logsum method is that it can calculate 
the accessibility benefits of both transport and spatial strategies, while traditional 
measures only address the benefit of transport policies. The Uplands scenario does 
not assume any changes in the transport system and all accessibility benefits result 
from changes in the spatial planning strategy. Compared to the Baseline scenario, 
the Uplands variant would bring with it a high loss in accessibility totalling -521 
million euros a year in 2020 (expressed in 2005 euros) and -1,182 million euros 
in 2040. 



5 Coupling a Detailed Land-Use Model and a Land-Use and Transport. 



93 



5.5 Conclusions and Directions for Further Research 

This chapter describes the combined use of the Land Use Scanner and Tigris XL 
models within the same framework for the Second Sustainability Outlook study. 
The interaction between the models was set up pragmatically without making 
fundamental changes to either of the models. The combined modelling framework 
was used to analyse the impacts of alternative integrated spatial and transport 
strategies. Combining the modelling framework has a number of advantages: 

• Detailed land-use data and plans from Land Use Scanner improve the simulation 
results of the residential location choices within Tigris XL and, thus, indirectly 
improve the simulation of the spatial distribution of employment (as these sectors 
influence each other). 

• Conversely, Tigris XL simulation results for employment at a regional level, and 
associated changes in land use, can be used as land-use claims for Land Use 
Scanner in a feedback interaction. The Tigris XL model ensures consistency 
between population and labour developments. 

• The combined Tigris XL and Land Use Scanner framework facilitates the 
analysis of the impacts of integrated land-use and transport policies. The results 
include impacts of changes in the transport system on the spatial distribution of 
residents/firms and land use and, conversely, the influence of changes in land use 
(and spatial distribution of residents and firms) on the transport system. Within 
Tigris XL, these mutual impacts can be analysed at the level of transport zones 
(1308 zones for the Netherlands); within the combined framework these effects 
are exchanged between Tigris XL and Land Use Scanner at the regional level. 
Land Use Scanner further allocates these changes to the grid cell level. 

The inclusion of Tigris XL in the combined framework, and in so doing also the 
national transport model, enables the calculation of additional indicators, such as: 

• Transport indicators for the alternative spatial and transport strategies, for 
example mode split and congestion effects. 

• Accessibility impacts, expressed in geographical accessibility indicators as well 
as in the monetary benefits of changes in accessibility. The option to calculate 
the monetary benefits of changes in accessibility, due to transport policies as 
well as land-use policies, played an especially important role in the evaluation of 
alternative transport and spatial strategies. The results of the calculation showed 
that spatial strategies had large monetary accessibility benefits. This is, for 
example, illustrated by the much higher accessibility benefits of an urban density 
strategy, which was also considered in the sustainability outlook, in comparison 
with a large road-infrastructure investment programme (see Geurs et al., 2010). 

The combined application of Land Use Scanner and Tigris XL was considered 
as promising and the framework has made a substantial contribution to the ex-ante 
evaluation of various policies, as discussed above. However, the interaction between 



94 



B . Zondag and K. Geurs 



the instruments was established only in a practical manner and a more fundamental 
integration of the instruments, or their respective knowledge, is still needed to 
further reduce overlaps and inconsistencies between the two and to fill in missing 
aspects. 

A general observation is that simulations of the use of land (such as agriculture 
or residential), objects (such as houses or offices) and actors (such as people or 
firms) are closely related. Therefore an important direction for further research 
is to develop a more integrated and consistent approach towards modelling these 
three layers in time and space. It is clear that there are still many inconsistencies 
and omissions in the current combined framework when spatial changes in the 
number and distribution of residents, jobs, houses and in land use are calculated: 
inconsistencies and omissions between the three layers, between different spatial 
scale levels, and in time. To address these matters, we recommend several 
fundamental improvements of the framework: 

• Explicit modelling of objects and actors at a lower geographical level in 
integration with land-use changes. Currently regional land-use claims derived 
from regional changes in population, housing or employment are allocated to the 
grid cell level. This often requires a post-processing step to transfer land use at the 
grid cell level into, for example, the number of people or firms. Information on 
the number of people or firms at the grid cell level is needed to be able to calculate 
indicators such as flood damage or environmental pollution. Explicit modelling of 
actors and objects within the region also enables improvements in the modelling 
of regional markets (e.g. housing and labour markets) and the behaviour of actors 
in these markets; 

• Use of one or more consistent theories throughout the analytical framework. 
The use of a sequential set of models as described in this chapter, results 
in applying different theoretical principles at different spatial levels of scale. 
For example, different approaches are used for the simulation of choices of 
residential location within a region and for choices between regions. Although the 
underlying processes can be different at a regional and local level, for instance, 
more job-related migration at a regional level versus more housing-condition 
related moves at a local level, the use of a consistent approach (e.g. utility 
theory) would help to interpret modelling results in a consistent way in the policy 
evaluation. 

• Improve the inclusion of supply information, such as land-use restrictions and 
options, in the modelling of regional developments. Ideally, this would mean 
that existing boundaries are less strictly enforced to allow conditions within 
a region to have an influence on interregional developments through feedback 
mechanisms (e.g. additional land-use restrictions in the Amsterdam region 
impact on the number of people in the nearby municipality of Almere). A first 
step has been made in this direction but additional steps are needed to improve 
the inclusion of land in the specification of the sector models. 

• Harmonise the temporal resolution of the model components . TIGRIS XL is an 
incremental model that uses time steps of 1 year, while Land Use Scanner is 



5 Coupling a Detailed Land-Use Model and a Land-Use and Transport. 



95 



an equilibrium model that is normally applied for time horizons of 10, 20 or 
30 years. Preferably, the interactions or data exchanges would take place annually 
to allow for path dependency and analysis of the developments over time. 

References 

ABF (2006). Scenario's en varianten Nederland Later, technical note prepared for Netherlands 

Environmental Assessment Agency, Bilthoven. 
Geurs, K., Zondag, B., De Jong, G. C, & De Bok, M. A. (2010). Accessibility appraisal of 

land-use/transport policy strategies: More than just adding up travel time savings. Transport 

Research Pari D: Transport and Environment, 15(7), 382-393. 
Hilferink, M., & Rietveld, P. (1999). Land Use Scanner: An integrated model for long term 

projections of land use in urban and rural areas. Journal of Geographical Information Systems, 

1(2), 155-177. 

MNP (2007). Nederland later. Tweede Duurzaamheidsverkenning. Bilthoven: Netherlands 

Environmental Assessment Agency. 
RAND Europe and Bureau Louter. (2006a) . Systeem documentatie TIGRIS XL vl.0, prepared for 

the Transport Research Centre. The Netherlands: Leiden. 
RAND Europe and Bureau Louter. (2006b). TIGRIS XL toepassen inN18 studie. The Netherlands: 

Leiden. 

Significance and Bureau Louter. (2007). Toepassen van TIGRIS XL binnen de studie 'Nederland 

later' . Report prepared for Netherlands Environmental Assessment Agency, Bilthoven. 
Significance and Bureau Louter. (2009). TXL analyse KBA RAAM. Report prepared for 

Netherlands Environmental Assessment Agency, The Hague, The Netherlands. 
Significance, Bureau Louter and Stratelligence. (2007). Toepassen TIGRIS XL in de case 

studie indirecte effecten. Report prepared for Transport Research Centre, The Hague, The 

Netherlands . 

Zondag, B. (2007). Joint modelling of land-use, transport and economy, T2007/4, TRAIL Thesis 
Series, The Netherlands. 



Chapter 6 

Biomass on Peat Soils? 



Feasibility of Bioenergy Production Under a Climate 
Change Scenario 

Tom Kuhlman, Rene Verburg, Janneke van Dijk, and Nga Phan-Drost 



6.1 Introduction 

Energy security has become a priority as the world's population increases and its 
standard of living improves, thus increasing energy consumption. As the demand for 
energy increases, there is growing concern about the possible exhaustion of finite 
supplies of fossil fuels in the not-too-distant future. In addition to the problem of 
availability, combustion of fossil fuels also has negative environmental effects: air 
pollution (e.g. particulates, nitrogen oxides, carbon monoxide and sulphur dioxide) 
produced through the combustion of fossil fuels, threatens human health as well as 
plant and animal life. Furthermore, the combustion of fossil fuels releases carbon 
dioxide and other greenhouse gases into the atmosphere, thus contributing to an 
increase in global temperature. These considerations lead to a search for alternative, 
renewable sources of energy, one of which is bioenergy. 

Bioenergy, defined as energy from biomass, has the advantage that the carbon 
dioxide released into the atmosphere by its combustion is compensated for through 
the carbon dioxide removed from the atmosphere by the vegetation from which it 
was ultimately derived. Moreover, any country can produce bioenergy, so energy 
security is enhanced. There are two basic forms of bioenergy, that released by the 
simple combustion of biomass (e.g. firewood or animal dung), either directly for 
heating or indirectly for generating electricity; and that formed by the chemical 
transformation of biomass into other fuels (ethanol, diesel oil or biogas). One 
important form of the latter is the manufacture of liquid fuels from food crops such 
as vegetable oils (biodiesel), sugar and maize (ethanol). This form of bioenergy 
has become controversial because it competes with the use of crops for food, thus 
contributing to high food prices; it is also controversial because its environmental 
impact is considerable. Hence, there is doubt as to whether the use of these 
biofuels really mitigates the emission of greenhouse gases. Moreover, these biofuels 



T. Kuhlman (El) 

Agricultural Economics Research Institute (LEI), PO Box 29703, 7502 LS The Hague, 

The Netherlands 

e-mail: tom.kuhlman@wur.nl 

E. Koomen, J. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 97 
Practice, The Geo Journal Library 101, DOI 10.1007/978-94-007-1822-7_6, 
© Springer Science+Business Media B.V. 2011 



98 



T. Kuhlman et al. 



are costly, which means that market-distorting policies are needed to promote 
their use. 

There is, therefore, a need to find more efficient ways of producing bioenergy. 
Some hope has been placed on so-called second-generation biofuels. These 
encompass a wider spectrum of materials, notably crop waste (such as straw) and 
plants that can be easily grown but are not suitable for food (e.g. many grasses). 
The most promising types of second-generation biofuels use woody and fibrous 
biomass (lignocellulose) rather than sugars and oils. The technology for breaking 
down lignocellulose into the sugars from which ethanol can be produced exists, but 
is not yet economically attractive. There are hopes that further technical progress 
will lead to efficient large-scale production of such fuels in the near future. This 
would mean establishing plantations of the appropriate crops, thus leading to the 
question of where these could be planted. 

To explore that question for the Netherlands we have opted for two types of 
wetland crops: reed and willow. In a densely populated country like the Netherlands, 
the feasibility of growing a low-value crop on a large scale is not immediately 
apparent. However, in the future, conditions are likely to be different from those 
of today. For one thing, climate change and sea level rise may result in some current 
uses of agricultural land becoming uneconomical. For another, multifunctional land 
use may provide a way out: biofuel need not be the sole product of the land. In 
the case of the Netherlands, using wetlands to grow these crops should provide 
additional benefits. Firstly, the land can be also used as a flooding buffer - the need 
for which may increase as a consequence of sea level rise and increased volumes 
of river flow. Secondly, since reed and willow are indigenous to low-lying areas in 
the Netherlands, their cultivation could enhance Nature values - albeit with some 
limitations . Their cultivation would help create naturally aesthetic wildlife habitats . 
The harvesting of the biomass would create a disturbance, of course, but this could 
be mitigated by appropriate timing of the harvest. Thirdly, both types of vegetation 
can have beneficial effects for land and water resources. Reed is an effective water- 
filtering agent and willow has the capacity to remove certain metals from the 
soil. Indeed, willow plantations have proven to be effective in soil remediation 
programmes and are being used commercially for this purpose (Bertholdsson, 2001; 
Greger & Landberg, 1999). Willow has been used experimentally in natural filtration 
beds in a manner similar to reed beds for the filtration and removal of nutrients from 
of waste water (biofiltration) and in buffer zones to protect sensitive areas from 
nutrient leaching (Larsson et al., 2003). All these benefits can be added to those of 
biomass production itself. 

Will multifunctional use be enough to make this type of land use feasible and, 
if so, where are suitable areas for growing this biomass to be found? That question 
is explored in this chapter, which looks specifically at the economic feasibility of 
these biofuel crops under present conditions, how these conditions might change, 
and what kind of change would be necessary to make growing these biofuel crops 
feasible. 



6 Biomass on Peat Soils? 



99 



6.2 Method 

To identify areas where suitable conditions for industrial-scale biomass production 
might exist, use was made of the spatial model Land Use Scanner, a GIS-based 
model for generating spatial forecasts of land use. The model allocates land to 
various classes of use according to those types of land use that generate the largest 
benefits for society. It is, therefore, based on economic theory: Land Use Scanner 
simulates competition between land-use classes in which the benefits are expressed 
as bid prices for land (Hilferink & Rietveld, 1999). 

Land Use Scanner is appropriate for our study because it considers not only the 
suitability of land for biomass production, but also the claims of all other types of 
land use: residential, industrial, agricultural, nature, and water management. Claims 
on land and suitability for each land-use class are dependent on the particular 
scenario under consideration. The suitability of each location (the model uses a 
large number of grid cells) is influenced by current land use, physical properties 
(soil type, groundwater), operative policies (zoning, water management, protection 
of nature areas), proximity to other land uses and accessibility (Koomen, Loonen & 
Hilferink, 2008). These characteristics are included in the database accompanying 
the model. However, the values attached to each of these aspects are governed by 
the particular scenario chosen for a particular model run. 

For our study, we designed two scenarios and compared their results: one 
scenario is an extrapolation of recent trends - called the reference scenario; the 
other is a constructed scenario within which crucial factors affecting the feasibility 
of biofuel production are varied so as to see under which conditions this particular 
type of production might become feasible - called the bio energy feasibility scenario. 
Both scenarios incorporate socio-economic factors and the impact of climate change 
as driving forces of land-use change: they both use the socio-economic scenarios 
described in a study published by three Dutch planning agencies in 2006 (CPB, 
MNP & PvPB, 2006), and the climate scenarios published in 2006 by the Royal 
Netherlands Meteorological Institute (Van den Hurk et al., 2006). The particular 
socio-economic scenario used in both our scenarios is the 'Al' version. This 
scenario is also referred to as 'Global Economy' and assumes market liberalisation 
and high rates of economic and demographic growth. The biomass feasibility 
scenario assumes strong climate change and introduces a high energy price as an 
additional driver. The reference scenario is presented for the sake of comparison. 

In our study, the Land Use Scanner model used 100 m x 100 m grid cells, 
offering a high resolution of land use. All land-use classes were simulated in Land 
Use Scanner's new, discrete allocation algorithm, which allocates only one type of 
land use to each cell; for more details, see Chapter 1. Since land-use claims are 
one of the basic scenario-dependent inputs of the model, claims for industrial-scale 
biomass production need to be established. This claim-set was evaluated by cost- 
benefit analyses of biomass production from reed and willow under a variety of 
conditions . 



100 



T. Kuhlman et al. 



To use Land Use Scanner for this study we introduced a new land-use class for 
biomass production. This is a multifunctional class, reflecting the additional benefits 
that this type of land use could generate: apart from biomass production, it includes 
water storage with some water filtration services and nature enhancement added. 
Since land is allocated to a particular land-use class based on the net benefit to be 
obtained from that land use, we first examined what these benefits were. This entails 
a cost-benefit analysis, which is dealt with in the next section. 

6.3 Cost-Benefit Analysis of Biomass Production 

Cost-benefit analysis (CBA) is normally used to assess whether a project is 
worthwhile, in comparison to alternative uses of the investment financing required. 
In our case, the question is whether a particular type of land use delivers greater 
benefits than the alternatives. Costs and benefits can be seen in social terms, i.e. the 
costs and benefits of land use to society. Private CBA, on the other hand, addresses 
the question of whether it will be worthwhile for the individual land user to allocate 
his or her land to the use in question. 

Our focus here is on social CBA, because we need to know under what conditions 
it will be advantageous to allocate land to biomass production. We compared the 
gross value of agricultural production of two biofuel crops with that of likely 
alternative agricultural uses, while also assessing the value of additional benefits 
to be obtained from using the land for reed and/or willow production. Considering 
private CBA would require that we look at net returns to the farmer, i.e. just 
incorporating production costs; we only look at processing costs after the produce 
leaves the farm. 

In temperate climates, reed {Phragmites australis) and willow (Salix spp.) grow 
naturally in swampy areas. On well-drained soils, providing the groundwater is 
maintained at present levels, reed can produce 35 tonnes of dry biomass per hectare 
while willow can produce 15 t/ha. In reed, 45% of this biomass is cellulose, whereas 
in willow the proportion is 37.5%. As each tonne of cellulose can generate 340 L 
of ethanol, 1 ha of reed can therefore produce 5,355 L of ethanol, while 1 ha of 
willow generates 1912.5 L. These are equivalent to 4.23 and 1 .51 tonnes of ethanol, 
respectively, at a specific gravity of ethanol of 0.789. From that we can estimate 
the energy generated per hectare of biomass, since one tonne of ethanol yields 26.7 
gigajoules (GJ). Thus, 1 ha of reed generates 112.9 GJ and that of willow 40.3 GJ. 
Much higher energy values can, however, be obtained from direct combustion of 
dry matter, for heating or for the generation of electricity: 620.0 GJ/ha of reed and 
274.89 GJ/ha of willow. Reed and willow have near maximal production values in 
conditions that are too wet for grass or field crops . The yield of reed in waterlogged 
soils is only 1 .8% lower than that in well-drained soils (Hellings & Gallagher, 1992). 

The next question is: what return can a farmer expect? This requires us to take 
a look at energy prices and their possible evolution in coming decades. World 
population growth together with economic growth will lead to a large increase in 
the demand for energy up to 2050. Energy supply will, however, have difficulty in 



6 Biomass on Peat Soils? 

Price of crude oil (€/bl) 

200 -. 
160 - 
120 - 




2000 2010 2020 2030 2040 2050 
Fig. 6.1 Projected prices (€/barrel) of crude oil up to 2050 



101 



History 

— Median 

| Upper and lower limit 



meeting this increased demand, as oil and gas reserves are expected to decline and 
remaining reserves will be more expensive to exploit. Although there are alternative 
sources of energy, these tend to be expensive. All this points to a future in which 
rising energy prices will make bioenergy economically more attractive. Projections 
of future oil prices were made with the PROMETHEUS model to calculate the 
prices of crude oil per barrel up to 2050 (Fig. 6.1). We have used the upper limit 
for the bioenergy feasibility scenario (i.e. a price of €160/barrel) and the median 
projection for the reference scenario (€80/barrel) . 

The cost price of harvesting, transportation and processing of biomass into 
ethanol in 2005 was US$ 60/t, which is equivalent to €50 at the average exchange 
rate in 2004/2005. Allowing for technological advances in processing, we assume 
this cost to drop by 1%/year, which implies a cost price of €31 .5/t in 2050. Direct 
combustion requires less cost than fermentation: US$ 50, or €41.7, per tonne in 
2005. By 2050 this could be as low as €26/t. This enables us to compute the net 
private benefit of biomass production. Figures 6.2 and 6.3 show the results for 
conversion of biomass into ethanol: Fig. 6.2 for the reference scenario, under which 
groundwater levels are maintained as at present and oil prices follow the median 
prediction; and Fig. 6.3 for the bioenergy feasibility scenario, with higher levels of 
groundwater and oil prices at the upper limit. 

These calculations show that ethanol production based on reed can provide a 
return to farmers if oil prices rise to levels such as those of 2008, before the current 
financial crisis caused a steep decline. It could become profitable in 20 years or 
so, even with lower oil prices, provided biomass processing technologies progress 
at a reasonable pace. Ethanol based on willow would be marginally feasible only 
under the most favourable conditions. Biomass production for direct combustion 
process offers more profit to farmers. The net benefits of both reed and willow are 
all positive, even for modest oil prices (Figs. 6.4 and 6.5). 



102 T.Kuhlmanetal. 
€/ha 




-800 J 

I 1 1 ' 1 ' 1 1 1 1 1 

2000 2010 2020 2030 2040 2050 

Fig. 6.2 Net benefit of biomass conversion to ethanol for the reference scenario 



€/ha 




2000 2010 2020 2030 2040 2050 

Fig. 6.3 Net benefit of biomass conversion to ethanol for the bioenergy feasibility scenario 



These figures are gross potential returns to farmers. How competitive can reed 
and willow be, given the present uses of farmland? An economic study of willow 
cultivation in Northern Ireland shows that, with gross margins excluding subsidies 
for suckler cows and for lowland sheep, willow coppicing can compete with 
grassland use, depending on circumstances on each individual farm (Rosenqvist & 
Dawson, 2005). In the Netherlands, potato is the most important field crop and grass 
for cattle the most important alternative land use for larger areas (i.e. excluding 
horticulture and intensive livestock-keeping). Based on average yield data for the 
period 2005-2007, we calculated the value added per hectare per year that these 
two uses of land bring the farmer. Potatoes can be grown only once every 4 years; 
less profitable crops have to be grown in the other 3 years in order to keep the 
soil healthy, which has been taken into consideration in the calculation of the 



6 Biomass on Peat Soils? 103 



€/ha 




2000 2010 2020 2030 2040 



Fig. 6.4 Net benefit of biomass combustion for the reference scenario (median oil price) 



€/ha 




2000 2010 2020 2030 2040 20i 

Year 



Fig. 65 Net benefit of biomass combustion for the bioenergy feasibility scenario (high oil price) 

average value added per year. One hectare of arable land used in this way yields 
roughly €9,000 for potatoes, €2,400 for sugar beet and €800 for cereals (LEI, 
2008). If it is assumed that cereals are grown for two of the 3 years, the annual 
return for potato is €3,200/ha. Grassland used for dairying generates approximately 
€4,800/ha (Table 6.1). 

These value-added figures must be compared with those for bioenergy production 
(shown in Table 6.2). We see that bioenergy can compete only if processed through 
direct combustion, and then only at fairly high oil prices. This is, however, only for 
the current situation. As Figs. 6.2, 6.3, 6.4, and 6.5 show, value added for bioenergy 
may rise in decades to come. In our simulation, energy yields of reed and willow 
were kept constant; although in reality they may rise, the same is likely to happen 
for other agricultural yields, so the competitive positions of different types of land 



104 



T. Kuhlman et al. 



Table 6.1 Average value 
added for the products 
associated with the two major 
farming systems in the 
Netherlands (2005-2007) 



Table 6.2 Potential value 
added of two different types 
of bioenergy production in 
2005-2007 at current 
groundwater levels and two 
different oil prices 





Area 


Value added 


Product 


(x 1,000 ha) (€/ha) 


Milk (per ha grass and fodder 1,209 


4 809 


maize) 






Winter wheat 


209 


1,025 


Summer wheat 


120 


785 


Summer barley 


45 


690 


Sugar beet 


85 


2,440 


Potatoes (consumption) 


69 


8,959 


Potatoes (starch) 


49 


2,494 


Grass seed 


25 


1,314 


Source: LEI (2008) 




Oil price = USD 


Oil price = USD 


Bioenergy type 


58/barrel a 


85/barrel a 


Conversion to ethanol 






Reed 


€-431/ha 


€199 /ha 


Willow 


€-278 /ha 


€/-50 ha 


Direct combustion 






Reed 


€391 /ha 


€1249 /ha 


Willow 


€194 /ha 


€575 /ha 



a These are the median and upper limit projections, respectively, 
for oil prices generated by the PROMETHEUS model for 2007. 
Actual oil prices in that year varied from $48 to 91/barrel (world 
average spot-market prices) 



use will not be affected. Only for the cost of processing biomass have we assumed 
a gradual reduction. Increasing liberalisation is, however, likely to have a negative 
effect on prices of milk and potatoes. Furthermore, if climate change leads to rising 
groundwater levels (as we have simulated in one of our scenarios), the yield of 
grassland and of potatoes will be reduced, whereas reed and willow will be little 
affected. These changes may make biomass production attractive in some areas, as 
is explored in Section 6.5. 

However, bioenergy would be only one source of economic return from the land 
under the proposed form of land use. Other returns/benefits include: 

• Surface water storage: reed and willow grow well on low-lying land and do better 
than other crops under waterlogged conditions. This makes them suitable crops 
for areas reserved for water retention, such as areas that may be temporarily 
flooded to counter excessive water levels in rivers or other water bodies, or to 
store fresh water to cope with drought and to combat salinity. 

• Natural water filtration: reed acts as a natural filter for purifying water. Moreover, 
waste water or sludge could stimulate wood biomass production substantially and 
the nutrients it contains could replace conventional fertilisers to a large extent. 
Research by Larsson et al. in 2003 has concluded that growth levels of willow 



6 Biomass on Peat Soils? 



105 



after a first 3-year rotation were higher under irrigation with waste water than for 
non-irrigated cultures or those irrigated with drinking water. 

• Nature values: since reed and willow are indigenous plants in wetter parts of the 
Netherlands, their cultivation can enhance natural attributes and provide a habitat 
for wildlife, thus stimulating the value of these areas for recreational activities 
and local tourism. 

To some extent, these benefits compete with one another and it will not be 
possible to optimise all of them simultaneously. Using waste water to fertilise a 
biomass crop will subtract from its nature value. Similarly, harvesting the biomass 
will disturb wildlife, although appropriate timing could minimise some of its 
negative effects. In this study, we have refrained from quantifying the filtration and 
nature benefits of this form of land use, confining ourselves to just mentioning them 
as additional potential benefits . 

Concerning water storage, it is the policy of the Dutch government to cope with 
climate change by providing space for excess water, rather than, as in the past, 
just raising the dikes and pumping the water out (Tielrooij, 2000). The current 
slogan is 'retain, store, discharge' in that order. In the strategy of 'Room for the 
River' (VROM, 2006), forty locations in the Netherlands have been identified as 
sites for retention buffers . The existing dike around the polder is to be breached to 
allow the river to flow onto the low-lying land at high water. Although this does 
not preclude agricultural use of such land, no high- value uses such as building or 
cultivation of perennial crops will be permitted - thus reducing the value of the 
land. If it is farmland (which will usually be the case), the farmer will have to be 
compensated for this loss of value. The economic benefit of water storage has been 
estimated to be €395/ha, or 4 eurocents/m 2 /year, as this is what the Dutch Ministry 
of Agriculture will pay as compensation to farmers whose land is used for water 
retention (Verburg & Jongeneel, 2008). 

6.4 The Purpose of the Scenario Approach 

To assess the potential for bioenergy production in the Netherlands, two scenarios 
have been drawn up that attach differing importance to several crucial conditions for 
biomass production. One scenario - the reference scenario - assumes a continuation 
of currently prevailing conditions; whereas the other - the bioenergy feasibility 
scenario - assumes conditions that make biomass production feasible. We, thus, 
set out to examine to what extent the feasibility of this type of land use depends on 
these favourable conditions. The differences between the two scenarios relate to: 

• demand for land from outside the agricultural sector (in turn, determined by 
population dynamics and economic growth); 

• energy prices prevailing on the world market; 

• the policy environment in relation to European agriculture; 

• the extent and speed of climate change; 

• government policies for adapting to climate change. 



106 



T. Kuhlman et al. 



6.4.1 Reference Scenario 

In this scenario, the assumptions concerning economic and demographic growth are 
in line with the Global Economy scenario in the Dutch study 'Welfare, prosperity 
and quality of living environment' (CPB et al., 2006). This scenario is also 
applied in preceding climate change adaptation studies in the Netherlands (see, 
for example, Koomen & Van der Hoeven, 2008). The main scenario assumptions 
and their application in the Land Use Scanner model have been documented 
extensively elsewhere (Koomen et al., 2008; Riedijk, Van Wilgenburg, Koomen & 
Borsboom-van Beurden, 2007). In short, the following assumptions apply: 

• The Dutch population will grow to around 20 million by the year 2050. Average 
household size will decrease further and economic growth will be substantial. 
This will raise demand for accommodation for living, working and recreation. 
As a consequence, the area of land available for agricultural will decline rapidly. 
These assumptions follow 

• The median projection for oil price will apply, leading to a price of €81/barrel 
(in 2004euros), as compared to €160 projected for the upper limit, by the year 
2050; 

• Protection and subsidies for European agricultural products will be withdrawn 
and European farms will have to operate in a liberalised market environment. 
Producer prices will be lower and a large part of the food consumed by Europeans 
will be imported from countries that can produce it more cheaply. 

• Climate change is assumed to be minor in our reference scenario, following the 
moderate (G) scenario, as described by the Royal Netherlands Meteorological 
Institute (Van den Hurk et al., 2006). Table 6.3 provides more details of the 
climate change scenarios. 



Table 6.3 Key parameters of 
two Dutch climate-change 
scenarios for 2050 relative to 
1990 values, often referred to 
as G (moderate) and W+ 
(warm scenario with changed 
circulation patterns) 



Climate parameter 



W+ 



Winter 

Mean temperature +0.9°C +2.3°C 

Annual coldest day +1.0°C +2.9°C 

Mean precipitation +3.6% +14.2% 

Summer 

Mean temperature +0.9°C +2.8°C 

Yearly warmest day +1.0°C +3.8°C 

Mean precipitation +2.8% -19.0% 

Potential evaporation +3.4% +15.2% 

Annual values 

Annual maximum daily mean 0% +4% 
wind-speed 

Absolute sea level rise (cm) 1 5-25 20-35 



Source: Van den Hurk et al. (2006) 



6 Biomass on Peat Soils? 



107 



• Water policies will be aimed at controlling flood threats by conventional means: 
strengthening and enlarging dikes; sand suppletion against coastal erosion; 
limited additional space for rivers; 

• As a consequence of these policies, in combination with modest changes in 
climate and hence water levels, groundwater levels are maintained at the same 
levels as today. 

6.4.2 Bioenergy Feasibility Scenario 

The bioenergy feasibility scenario has been designed in such a way that its principal 
driving forces point towards conditions that are favourable for biomass production 
in combination with water storage. In other words, this scenario examines whether 
bioenergy production could be feasible under favourable conditions and, if so, the 
likely location of that production. In general, the driving forces are the same as 
in the reference scenario. However, the following special conditions apply that are 
favourable to biomass production in combination with water storage: 

• Energy prices will follow the upper limits of the projection shown in Fig .6.1. This 
is consistent with high economic growth (globally, as well as in the Netherlands). 

• Climate change is assumed to follow the relatively warm scenario with changed 
circulation patterns (W+), see Table 6.3. The climate change scenario selected 
is characterised by a temperature rise of 2°C between 1990 and 2050, plus 
concomitant change in atmospheric circulation patterns. Winters will be less cold 
but wetter because of an increase of westerly winds; summers will be drier and 
warmer because of prevailing easterly winds (Van den Hurk et al., 2006). 

• Protection against flooding will be provided partly by 'returning the land to the 
water', for example, through the designation of water retention areas. Higher 
water levels in both the sea and rivers (particularly in winter) are also expected 
in this scenario. Combined with the new water-retention policy, this will lead to 
higher groundwater levels and occasional inundation of low-lying areas - at least 
in those areas that can be flooded without extensive economic damage. 

6.5 Implementation and Results 

Our next step was to assess in a spatial dimension the potential for biomass 
production in relation to other agricultural crops. This was done by applying Land 
Use Scanner (described in Section 6.2), using the relative benefits of the different 
land-use classes (described in Section 6.3), to the scenarios drafted Section 6.4. 
Modelling with the Land Use Scanner involves the construction of suitability 
maps and the specification of a (regional) demand for land, which are reported in 
this section together with the results of the simulation. Here we limit ourselves 
to specific implementation issues related to biomass production, as the basic 
implementation of the Global Economy scenario is described extensively elsewhere 
(Riedijk et al.,2007). 



108 



T. Kuhlman et al. 



6.5.1 Suitability Maps 

The physical or spatial characteristics of locations, in our case 100 m x 100 m grid 
cells, are described in the suitability maps with scenario-dependent values relating 
to these characteristics. The local suitability for the bioenergy land-use types is 
determined by the potential values of biomass production and water retention. 

The potential value of biomass production is based on the cost-benefit analysis 
introduced in Section 6.3. The yield values presented there, however, only apply for 
optimum conditions. Therefore, a yield-loss fraction is specified for each location, 
determined by soil type and hydrological conditions. These yield-loss fractions 
have values between 0 and 1, with zero yield-loss representing optimum growth 
conditions. These calculations were also made for the principal competing land uses, 
potatoes and grassland. 

To describe local soil types, a simplified soil map of the Netherlands was 
constructed based on the soil map 1:250,000 (WUR-Alterra, 2006), distinguishing 
between clay, sand, loam and peat (organic) soils. Groundwater data were 
derived from the groundwater map, which has a resolution of 100 m x 100 m 
(WUR-Alterra, 2006). 

Martin and Stephens (2006) report in their study that willow grows well in coarse 
to fine loamy soil textures of alluvial or fluvial deposits. In consequence, clay soil 
and sand soil have a higher yield-loss fraction than other soil types. We used the 
following yield-loss fractions for willow: loam = 0.0; peat = 0.1; partially organic 
soils (with a lower percentage of organic material, and the rest made up mostly of 
sand) = 0.2; sand = 0.4; clay = 0.6. 

Reed grows in marshes and swamps, in firm mineral clays, where groundwater 
levels fluctuate from 15 cm below soil surface to 15 cm above - as is the case 
along streams, lakes, ponds, ditches, and wet wastelands; reed tolerates moderate 
salinity (Duke, 1983). Most soil types in the Netherlands are suitable for growing 
reed; only sandy soils have a somewhat lower potential, and thus a higher yield-loss. 
Reed yield-loss fractions are: clay = 0.0; loam = 0.2; peat = 0.2; partially organic 
soil = 0.2; sand = 0.3. 

These soil-based yield-loss fractions are, of course, independent of the scenario. 
As was discussed previously, this is not the case for groundwater levels. Thus, 
separate suitability maps per crop are needed for each scenario. The bioenergy 
feasibility scenario does, however, not only have higher groundwater levels than 
the reference scenario, but also higher energy prices . The net effect is much higher 
returns than under the reference scenario. 

Before the suitability maps could be set up, there was one more step to be taken. 
Since Land Use Scanner simulates a land market in which land-use classes act as 
rival bidders for land, the suitability has to be expressed as a bid price reflecting 
the value of that particular location for that particular land use. Therefore we had to 
convert yield figures into bid prices. This was done by calculating the present value 
of the land for the particular crop: revenue per year was added up for a number of 
years and discounted for each year, reflecting the fact that a profit in the future is less 
valuable than the same profit today. The common discount rate for public investment 



6 Biomass on Peat Soils? 



109 




Fig. 6.6 Suitability maps for reed in 2050 based on the median oil price projection and 
normal groundwater levels (left); and on the upper limit (high) projection of oil prices and high 
groundwater levels (right) 

projects in the Netherlands is 3.5%.' Over a period of 20 years, this results in a land 
value of about 15 times the annual yield. This is equivalent to a bid price in 2050 of 
€3.8-€8.5/m 2 for the cultivation of reed, depending on the oil price (assuming that 
the reed will be used for combustion rather than for distilling ethanol). For willow, 
the bid prices are, of course, much lower: from €1.7 to €3.9/m 2 . The resulting 
suitability maps for reed are shown in Fig . 6 .6 . 

The areas suitable for reed are primarily low-lying peat soils and partially organic 
soils. Being low -lying, they are also suitable for water storage. Wet conditions mean 
yield loss, but for reed this loss is less than for field crops or grass, which means that 
more areas will be suitable for reed under such conditions, as the Fig. 6.6 shows. For 
willow, there is hardly any loss at all, as was discussed earlier. Figure 6.7 shows the 
suitability maps for willow. The pattern is very similar to that of reed; the differences 
are due to a lower bid price for willow and to slightly different soil requirements . 

These bid prices can be compared to those of potato (€4.7, calculated in the 
same way as above) and dairy production (€7.0). Potatoes are grown on relatively 



The appropriate discount rate is controversial, as is the period over which discounting should be 
done. However, for comparing the competitive attractiveness of different agricultural uses it makes 
no difference what method is used, as long as it is the same method for all. What is important is that 
a bid price is reached that realistically reflects the competitive strength of the agricultural sector as 
compared to other land uses. This strength is low: almost all other land uses offer higher returns 
per hectare than agriculture, which means that agriculture effectively functions as the balancing 
entry in the land use accounts. The only land uses offering lower returns are for forest and Nature, 
but these tend to be protected by law. The values arrived at here for agricultural bid prices remain 
well below those used for competing land uses in Land Use Scanner (e.g. €20/m 2 for recreation) . 



110 



T. Kuhlman et al. 




Fig. 6.7 Suitability maps for willow in 2050 based on the median oil price projection and 
normal groundwater levels (left); and on the upper limit (high) projection of oil prices and high 
groundwater levels (right) 



dry land, but dairy production has a suitability pattern fairly similar to that of 
reed and willow, as Fig. 6.8 shows. The difference is that, since the value added 
per unit area of dairy production under the reference scenario is higher than for 
the bioenergy feasibility scenario, the potential area for dairying is much larger. 




Fig. 6.8 Suitability maps for dairy production in 2050 based on the median oil price projection 
and normal groundwater levels (left); and on the upper limit (high) projection of oil prices and high 
groundwater levels (right) 



6 Biomass on Peat Soils? 



Ill 




Fig. 6.9 Additional benefit (increased bid price) for biomass suitability reflecting the local 
potential for water storage (2050) 

With high groundwater levels and without market protection, however, the dairy 
yield is reduced, as the right-hand map demonstrates. Yield reduction due to wetter 
conditions is much less for reed and willow and, moreover, these crops do better 
because of higher energy prices . 

A final element in the suitability map definition for the bioenergy land-use types 
is the expected additional benefit of water storage (see Section 6.3). This aspect is 
based on elevation - naturally, low-lying areas are needed; and hydrological status 
-areas adjacent to rivers, canals, lakes and already identified retention areas are 
categorised in different degrees of suitability. This means that the local suitability 
value (bid price) can be augmented with this additional benefit. Using the value 
of €395/ha/year presented in Section 6.3 and the bid price calculation method 
described above, we arrived at an increase in bid price of €0.58/m 2 . The spatial 
pattern of this additional benefit, which is the same for both scenarios, is shown in 
Fig. 6.9. 

6.5.2 Land Claims 

Apart from mapping the suitability of each particular cell for each type of land use 
(using a maximum bid price as a standard from which each cell can deviate), the 
Land Use Scanner model also requires as input a total claim for each land use, that 



112 



T. Kuhlman et al. 



is, the total amount of land required for that land-use class. This, too, is scenario 
dependent. 

The model allows minimum and maximum claims, i.e. the land minimally 
required and the maximum that can be allocated. For biomass production, we set 
the minimum at 0, meaning that the model is not forced to allocate any land to it. 
At the maximum, the demand for energy would be high enough to cover the entire 
country with reed and willow. This is unlikely, of course. For example, people attach 
a high value to the historical agricultural landscapes existing in the Netherlands, 
some of which are unique. Covering these with reed and willow would make the 
landscape monotonous and obliterate a thousand-year history. Therefore, we set the 
maximum extent of biomass production at 15% of the present agricultural area. 
Since the agricultural area is bound to shrink in the period up to 2050, it will be 
more than that proportion in the target year. A better-founded demand for bioenergy 
in the Netherlands, taking into account international market conditions, could be 
obtained from specified agro-economic models such as LEITAP (Banse, Van Meijl, 
Tabeau & Woltjer, 2008). This is foreseen for future studies. 

For water storage, Kok (2004) calculated a need of 93,678 ha of water retention 
and storage by 2050 . However, part of this consists of existing primary watercourses, 
which leaves 72,226 ha needed for additional water storage. 

6.53 Results 

The results for both scenarios are shown in Fig. 6.10. The model allocates very 
little land (3,350 ha, to be exact) to biomass crops under the reference scenario. In 
other words, under current hydrological conditions and with oil prices at a level of 
about €80/barrel (or about $100-1 10 at the exchange rate prevailing in the first half 
of 2009), biomass cultures on wetlands are unlikely to be an attractive option for 
farmers except on a few locations. However, climate change and measures to adapt 
to it could alter hydrological conditions on low-lying farmland (especially peat soils 
now used for grassland) so as to make these areas less suitable for pasture and hay 
and more suitable for reed and willow. If this is combined with higher oil prices 
(at a level similar to or higher than those prevailing in the first half of 2008), reed 
would become commercially interesting in large parts of the Netherlands. Land Use 
Scanner allocates to it 339,000 ha, which is almost 15% of all agricultural land in 
use today. 

Willow gives lower yields than reed, so from the point of view of energy 
production the scenario would lead to the cultivation of reed only. However, mixing 
reed with willow stands would make for a less monotonous landscape, albeit with 
slightly lower biomass production. 

The areas allocated to biomass production in the map on the right-hand side have 
a variety of soils - clay, sand and peat; but the larger contiguous areas are peat soils. 
These are also the areas for which the long-term sustainability of dairy production is 
questionable, since optimal growth of grass requires lowering the groundwater to a 
level at which oxidation and subsidence of these vulnerable soils becomes inevitable 



6 Biomass on Peat Soils? 



113 




Fig. 6.10 Land use for reed with water storage in 2050 under two scenarios: for the reference 
scenario (left); and for the upper limit (high) projection of oil prices and high groundwater levels 
(right) 

(Van den Akker, 2005). Crops such as reed and willow that can be grown under wet 
conditions may well be the salvation of these soils. 



6.6 Conclusion 

Reed (Phragmites australis) and willow (Salix spp.) are potential feedstocks for 
bioenergy. Their biomass can be processed either into cellulose-based ethanol or 
burned directly to generate electricity. Since the first process itself requires much 
energy, the second appears to be the most efficient manner for harnessing this energy 
source - at least with current technologies. 

Both plants are indigenous to the Netherlands, and do well under wet conditions. 
This makes them particularly suitable for the peaty soils in low-lying areas. These 
areas are presently used mostly for pasture (dairy cows and sheep), and we 
have, therefore, examined the economic viability of reed and willow in relation 
to grassland pastures for dairy production. This viability is also given a spatial 
dimension: we investigated where growing bioenergy crops might be advantageous. 

These bioenergy crops offer other potential benefits as well. The land used for 
these crops can also be used for temporary water storage, either seasonally or as a 
reserve in case of calamity. The need for such areas is likely to increase as climate 
change progresses - albeit depending on whether a policy of adaptation to increased 
water levels is opted for or one that raises dikes and increases the volume of water 
to be pumped out. Furthermore, this type of land use may also improve Nature 



114 



T. Kuhlman et al. 



values, although this will depend on how the land is managed. Finally, reed also has 
a propensity for purifying water. A possible disadvantage of this scenario could be 
the creation of a monotonous landscape, although this could be partially alleviated 
by mixing reed and willow production. 

Quantifying only the benefits of energy production and water storage, under 
currently prevailing conditions (2009) these wetland bioenergy crops offer 
significantly lower benefits than for grassland agriculture. Conditions change, 
however, if we factor in, on the one hand, the impact of climate change (higher 
groundwater levels as a consequence of sea level rise) and, on the other hand, higher 
energy prices (a consequence of increasing scarcity of fossil fuels). Extrapolating 
these potential trends until 2050, the viability of grassland use declines and the 
attractiveness of wetland bioenergy crops increases substantially. At energy prices 
not much higher than those prevailing in early 2008, assuming modest progress 
in bioenergy technology, and following a climate scenario where not only higher 
temperatures prevail but where air circulation patterns have also been affected, reed 
will outcompete grassland in large areas of the Netherlands. The largest contiguous 
areas are presently peat land or are covered by partially organic soils . 

With our study, we intended merely to explore a method for investigating these 
questions. The method has some limitations that could be addressed by further 
research. For instance, our method of assessing viability is based on value added per 
unit area. This tells us much about the social benefit of a certain land use (whether 
it would be a worthwhile way of using the land), but less about its commercial 
feasibility (whether it will pay for the land owner to indeed use the land in that 
way). Gross value added can be defined as the total value of the produce of the land 
minus the consumable materials that have been used in the production. To arrive 
at the net profit per unit area, we need to know the total production cost. In all 
likelihood, production costs for reed are likely to be low compared to milk (there 
will be mainly a cost for establishing the reed), but this still needs to be investigated. 
Our calculations of value added are net of transport and processing, since these are 
costs beyond the farm gate. 

Another aspect that has not yet been included in our study is economies of 
scale. Transportation costs of reed are high, which means that, in order to operate 
efficiently, a fairly large production area is needed and the power station being 
supplied needs to be close by. Smaller patches of land shown on our maps will 
not, in effect, be feasible production areas. 

For the near future, wetland bioenergy crops such as reed and willow are 
likely to be no more than niche products. In the longer term, however, depending 
on climate change, energy prices, developments in the dairy market, water 
management policies in the Netherlands and progress in bioenergy technologies, 
the transformation of familiar meadow landscapes into wetlands producing reed and 
willow for electricity generation is far from unlikely. 

Acknowledgement We are grateful to the Dutch National Research Programme on Climate 
Change and Spatial Planning for financing part of the research described here. Another part 
was funded under the Knowledge Base programme of the Ministry of Agriculture, Nature and 



6 Biomass on Peat Soils? 



115 



Food Quality. We are , furthermore , grateful to the Netherlands Environmental Assessment Agency 
(MNP) and the Royal Netherlands Meteorological Institute for providing the modelling framework 
of Land Use Scanner and parts of the scenario definitions. 



References 

Banse, M., Van Meijl, H., Tabeau, A., & Woltjer, G. (2008). Will EU biofuel policies affect global 

agricultural markets? European Review of Agricultural Economics, 35, 1 17-141 . 
Bertholdsson, N.-O. (2001). Phytoremediation of heavy metals with Salix. In Swedish, English 

summary. Journal of the Swedish Seed Association, 111(2), 84-90. 
CPB, MNP, & RPB (2006). Welvaart en Leefomgeving. Een scenariostudie voor Nederland in 

2040 (Prosperity and Living Environment). Centraal Planbureau, Milieu- en Natuurplanbureau 

en Ruimtelijk Planbureau, The Hague. 
Duke, J. (1983). Handbook of energy crops: Phragmites australis, Ecology. Published online: 

http://www.hort.purdue.edu/newcrop/duke_energy/Phragmites_australis.html. Retrieved June 

2009. 

Greger, M., & Landberg, T. (1999). Use of willow in phytoextraction. International Journal of 

Phytoremediation, 1, 115-123. 
Hellings, S. E., & Gallagher, J. L. (1992). The effects of salinity and flooding on phragmites 

australis. Journal of Applied Ecology, 29,41-49. 
Hilferink, M., & Rietveld, P. (1999). Land use scanner: An integrated GIS based model for long 

term projections of land use in urban and rural areas. Journal of Geographical Systems, 1(2), 

155-177. 

Kok, T. (2004). Waterherging en natuur. Quick scan naar de comhinatie waterberging en natuur 

(Water storage and nature) . Ede: Expertisecentrum LNV. 
Koomen, E., Loonen, W., & Hilferink, M. (2008). Climate-change adaptations in land-use 

planning; a scenario-based approach. In L. Bernard, A. Friis-Christensen, & H. Pundt 

(Eds.), The European information society; Taking geoinformation science one step further 

(pp. 261-282). Berlin: Springer. 
Koomen, E., & Van der Hoeven, N. (2008). The Netherlands climate proof; What will the country 

look like in 2040? Geolnformatics , 11(5), 26-27. 
Larsson, S., Cuingnet, C, Clause, P., Jacobsson, I., Aronsson, P., Perttu, K., et al. (2003). 

Short-rotation Willow biomass plantations irrigated and fertilised with wastewaters . Results 

from a 4-year multidisciplinary field project in Sweden, France, Northern Ireland and 

Greece. Danish Environmental Protection Agency, Sustainable Urban Renewal and Wastewater 

Treatment Project, Report No. 37, Copenhagen. 
LEI (2008). Land- en tuinbouwcijfers 2008 (Agriculture and horticulture in figures), LEI-report 

no. 2008-048, LEI, The Hague. 
Martin, P. J., & Stephens, W. (2006) . Willow growth in response to nutrients and moisture on a clay 

landfill cap soil. I. Growth and biomass production; II: Water use. Bioresource Technology, 97, 

437^158. 

Riedijk, A., Van Wilgenburg, R., Koomen, E., & Borsboom-van Beurden, J. (2007). Integrated 

scenarios of socio-economic and climate change; a framework for the 'Climate changes Spatial 

Planning' program, Spinlab Research Memorandum SL-06, Vrije Universiteit Amsterdam. 
Rosenqvist, H., & Dawson, M. (2005). Economics of willow growing in Northern Ireland. Biomass 

and Bioenergy, 28(2005), 7-14. 
Tielrooij, F. (Ed.). (2000). Waterbeleid voor de 21e eeuw (Water policy for the 21st century), 

Commissie Waterbeheer 21e Eeuw, The Hague. 
Van den Akker, J. (2005). Maaivelddaling en verdwijnende veengronden (Soil subsidence and 

disappearing peat soils). In W. A. Rienks & A. L. Gerritsen (Eds.), Veenweide 25x belicht. 

Een bloemlezing van het onderzoek van Wageningen (pp. 11-13). Wageningen: Wageningen 

University and Research Centre. 



116 



T. Kuhlman et al. 



Van den Hurk, B., Klein Tank, A., Lenderink, G., van Oldenborgh, G. J., Katsman, C, Van den 
Brink, H„ et al. (2006). KNMI Climate Change Scenarios 2006 for the Netherlands. KNM1 
Scientific Report WR 2006-01. De Bilt: KNMI. 

Verburg, R., & Jongeneel, R. (August 2008) . Exploring multifunctional land uses as an adaptation 
strategy to climate change in the Netherlands: An economic assessment of costs and benefits of 
ecosystem services. Paper presented to the meeting of the European Association of Agricultural 
Economists, Ghent. 

VROM (2006). PKB Ruimte voorde Rivier (Room for the river, brochure). The Hague: Netherlands 

Ministry of Housing, Spatial Planning and the Environment (VROM). 
WUR-Alterra. (2006). Dataset Grondsoortenkaart van Nederland 2006 (Soil map of the 

Netherlands), Wageningen. http://www.bodemdata.nl/. Retrieved 12 August 2009. 



Chapter 7 

Simulation of Future Land Use for Developing 
a Regional Spatial Strategy 

The Case of the Province of Overijssel 



Arjen Koekoek, Eric Koomen, Willem Loonen, and Egbert Dijk 



7.1 Introduction 

Many geo-information tools, such as visioning, storytelling, forecasting, analysis, 
sketching, and evaluation, appear to be rarely used for spatial planning (Vonk, 
Geertman & Schot, 2005). Progress in the application of such tools beyond basic 
activities, such as researching spatial queries and generating thematic maps, to 
help solve key spatial planning problems remains limited (Stillwell, Geertman & 
Openshaw, 1999; Vonk et al., 2005). Land Use Scanner's model has the ability 
to assist in many of these planning-specific tasks. Especially in scenario-based 
national forecasts of on the future the model, it has proven to be an adequate 
tool for informing policy-makers on potential future developments (Borsboom-van 
Beurden et al., 2005) and has provided ex-ante evaluations of policy alternatives 
(MNP, 2001; Stillwell et al., 1999; Van der Hoeven, Aerts, Van der Klis & Koomen, 
2009). More recently, the model has also been used to optimise projected spatial 
developments according to specific policy objectives (Borsboom-van Beurden, 
Bakema & Tijbosch, 2007; MNP, 2007). The resulting maps have the potential 
to inform policy-makers about alternative solutions for current spatial problems 
(Koomen, 2008). This chapter demonstrates the capacity of Land Use Scanner 
to generate optimised spatial developments in actual regional planning contexts. 
Contributing to the policy formulation process by simulating land use on a regional 
scale in close cooperation with a regional authority is an interesting step forward 
in the application of land-use models. Thus, in this case Land Use Scanner has 
traversed the full spectrum, from academic research to actual planning practice. 
Applying Land Use Scanner on a regional scale has been a tempting idea since 



A revised version of this chapter is published as: Koomen, E., Koekoek, A. and Dijk, E. (2011) 
Simulating land-use change in a regional planning context. Applied Spatial Analysis and Planning 
doi: 10.1007/sl2061-010-9053-5. This article is published with open access at Springerlink.com. 

A. Koekoek (is) 

Geodan, President Kennedylaan 1 , 1079 MB Amsterdam, The Netherlands 
e-mail: arjen.koekoek@geodan.nl 

E. Koomen, J. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 117 
Practice, The GeoJournal Library 101, DOI 10.1007/978-94-007-1822-7J7, 
© Springer Science+Business Media B.V. 201 1 



118 



A. Koekoek et al. 



the modelling resolution was increased from 500 to 100 m. Before the currently 
discussed Overijssel study was done, such applications were, however, limited to a 
single example (Bouwman, Kuiper & Tijbosch, 2006). 

What spatial transformations can be expected in the future and what role can 
policy-making play in directing developments? The introduction of a new national 
spatial planning law (nWRO) in 2008 created opportunities for the Province to play 
a more pro-active role in the policy-making arena. In order to visualise spatially 
accurate versions of possible future scenarios it is paramount to have access to data 
on the desired spatial level. Scenario simulations rely heavily on the data being 
input and expert judgement. As many future developments entail a high degree of 
uncertainty, especially in the longer term, one should always interpret simulations 
as possible future configurations of land use. 

7.1.1 A New Regional Spatial Strategy for the Province 
of Overijssel 

In 2008, the strategic plan for the physical environment of the Province of Overijssel 
that was current at that time needed readjustment, which led to a need for a 
new strategic plan. Another factor that increased this need was the emergence of 
new themes on the policy agenda, such as climate change and new insights into 
demographic trends. The Province of Overijssel's planning authorities - hereafter 
referred to as the Province - used its new strategic plan to introduce a new planning 
creed, 'Quality-based decision-making' . The introduction of the new national spatial 
planning act (nWRO) provided a logical moment for redefining the Province's 
aims and embedding these in spatial plans. For this reason, the Provincial Council 
decided to develop a new, integral strategy for the physical environment. This 
plan, called the Regional Spatial Strategy (Omgevingsvisie Overijssel), replaced five 
plans then current and will legally function as the spatial structural vision, regional 
water management plan, environmental policy plan, provincial mobility plan and 
soil-management plan. The Regional Spatial Strategy goes beyond just formulating 
policy by presenting and framing policy choices and their implementation in an 
integral way for the entire spatial spectrum. The Strategy's sustainability and spatial 
quality goals function as overarching perspectives that connect the spatial themes it 
contains . 

The Province asked Geodan Next at the start of the policy formulation process 
to explore new spatial policy alternatives. In this process, important questions were 
posed: to what degree are current spatial developments 'climate proof, and what are 
the likely effects of possible spatial policies? Land Use Scanner played an important 
role in the process of answering questions like these by modelling policy-specific, 
future land-use configurations at different stages in the development process. From 
the start of the project, the specific assumptions associated with this type of research 
were made known to the Province. As many future developments contain a high 
degree of uncertainty, especially on the longer term, one should always interpret 
simulations as possible future land-use configurations. The spatial images generated 



7 Simulation of Future Land Use for Developing a Regional Spatial Strategy 



119 



can not be used as blueprints of the future: rather, they are to be used to stimulate 
thinking about cause-effect relationships in policy discussions . 

7.1.2 Chapter Contents 

This chapter presents two related studies that Geodan Next performed for and 
in cooperation with the Province. Section 7.2 discusses an exploratory study 
that provided potential future land-use configurations according to specific policy 
objectives. The results of this study were used in the policy formulation process 
for the Regional Spatial Strategy. Section 7.3 describes the Environmental Impact 
Assessment that Geodan Next subsequently performed for the Regional Spatial 
Strategy. Finally, Section 7 .4 summarises the role of Land Use Scanner in the two 
studies and finishes with some comments on its general applicability. 

7.2 Spatial Exploration: Overijssel in 2040 
7.2.1 Creating a Regional Application 

The starting point for the regional spatial exploration was the national Second 
Sustainability Outlook, study, carried out by the Dutch Environmental Assessment 
Agency (MNP, 2007; see Chapter 4 by Rienk Kuiper, Marianne Kuijpers-Linde, 
and Arno Bouwman, this volume). In this study, two trend-based scenarios from 
the national the Welfare, Prosperity and Quality of the living environment study 
of the Dutch assessment agencies (CPB, MNP & RPB, 2006) were applied to 
accommodate for uncertainties in demographic and economic growth: one reflecting 
moderate spatial pressure, comparable to the Transatlantic Market-scenario; and 
the other reflecting high spatial pressure, similar to the Global Economy scenario. 
The related scenario assumptions were only used to assess the potential diversity 
in demand for residential and commercial land. The demand for the other types 
of land and the spatial relations embedded in the suitability maps are the same 
for both scenarios, reflecting their trend-based or business-as-usual character. An 
important assumption in both scenarios in relation to the Province of Overijssel is 
that the influx of people from the Randstad is limited compared to directly adjacent 
provinces such as Gelderland, Noord-Brabant and Flevoland. 

To apply the national Land Use Scanner application related to the Second 
Sustainability Outlook to the specific context of Overijssel, several adjustments 
were necessary. In close cooperation with various representatives of the Province, 
local suitability variables were adjusted to match current regional spatial policies 
more closely. The most important adjustments were: 

• The importance of national concentration areas (bunde lings gebiederi) for 
residential and commercial development was decreased, as this national policy 
is not considered to be of great importance by the Province. 



120 



A. Koekoek et al. 



• Areas close to the larger cities were considered likely locations for residential 
and commercial development. This applied to a selection of eight relatively 
large towns in Overijssel. This measure was implemented to limit the growth of 
smaller towns that was observed in initial national simulations; such growth was 
considered not to be in line with current regional policies that aim to concentrate 
urbanisation. 

• Commercial areas from the IBIS database were considered attractive for new 
commercial development. 

• New developments in flood-prone areas and water-retention areas were 
discouraged through restrictive policies. 

• A spatial redefinition of the ecological main structure (the national plan for 
a spatially coherent set of nature areas) to replace an older version. New 
commercial and residential functions in the ecological structure and national 
landscapes were considered unlikely and received a higher negative suitability 
factor. 

The above adjustments were incorporated in the suitability maps that describe the 
preferred locations for the various types of land use. Restrictions lead to lower or 
even negative land-use suitability values, whereas stimulating policies are reflected 
as higher suitability values. Application of the adjusted suitability values resulted in 
adjusted spatial patterns for Overijssel in the year 2040. 

The Province was particularly interested in the spatial impact of the trend-based 
scenario simulations on certain sensitive areas. By using an interactive set-up in 
a workshop, the Province was able to provide feedback on preliminary results 
and thus fine-tune the resulting land-use patterns. The impact of urbanisation 
was analysed in relation to three different policy themes. The rationale behind 
these impact assessments is the following: from a nature perspective, building in 
the ecological main structure is discouraged; for landscape conservation, from a 
landscape preservation perspective large-scale urban development in the national 
landscapes is discouraged; and from a flood-risk perspective, building in flood-prone 
areas is unwanted. The impacts were assessed by comparing pixel-by-pixel the 
urban (residential, commercial and greenhouses) locations shown in the maps 
of current and simulated land use. The new urban locations were then overlaid 
on maps representing the specific policy themes. This procedure showed that 
new urban development in nature areas and national landscapes is limited in the 
two trend-based scenarios, although such development is more abundant within 
flood-prone areas, especially near the large cities of Zwolle and Deventer. Especially 
in the scenario reflecting high spatial pressure, the amount of urban development in 
the flood-prone areas is substantial. 

7.2.2 Visualising Theme-Specific Spatial Policy Alternatives 

What spatial developments can be expected if one thematic ambition is taken as 
the leading objective in spatial planning? By giving one policy objective overriding 
importance in land-use simulations, the spatial impact of successful implementation 



7 Simulation of Future Land Use for Developing a Regional Spatial Strategy 



121 



National Land 
use simulation 
(MNP, 2007) 



Input adjustments 
to adapt to the 
regional context 



Output Regional 
trend T (Map 1 ) 



Specific thematic 
input adjustments 
for policy 
alternative A1 

Specific thematic 
input for policy 
alternative A2 
(Map 2) 



Specific input 
adjustments for 
policy alternative A3 



Output: results 

of policy 
alternative A1 



Output: results 
of policy alternative 
A2 (Map 3) 



Differences in 
> built-up area between + 
output T and A1 



Differences in 
built-up area between ^_ 
' output T and A2 *~ 

(Map 4) 



Output: results 

of policy 
alternative A3 



Differences in 
> built-up area between « 
output T and A3 



Fig. 7.1 Flowchart describing the creation of regionalised policy-specific land-use simulations. 
Maps 1^1 are included in Fig. 7.2 



of that specific policy can be shown. The Province was interested in the potential 
spatial developments related to three different spatial policy ambitions. These 
policy alternatives were formulated as coherent sets of spatial policies related to 
the themes of: (1) water management; (2) safety and health; and (3) compact 
urbanisation. The high spatial pressure scenario was used for these simulations, 
because it maps the impact of the policy measures more clearly. As such, it prepares 
policy -makers for a worst-case scenario with respect to the spatial developments that 
had to be accommodated in the policy alternatives . The regional model application 
described in the preceding section was used as a starting point in this optimisation 
effort. For the simulations of the three thematic policies, additional, policy -related 
spatial datasets were added in Land Use Scanner to stimulate or restrict certain 
developments. By comparing the simulation results of the trend-based scenario 
(T) and the theme-specific policy alternatives (A123), spatial differences could be 
visualised, as described by Fig. 7.1 . 

The first thematic policy alternative focused on the optimisation of spatial 
developments from a water management perspective. Several policies are included 
in this alternative. To prevent drought, groundwater protection areas are kept 
free from urban development. To limit the potential impact of flooding, urban 
development is not allowed within flood-prone areas, near major waterways or 
in areas with a high risk of inundation, such as winter riverbeds. Compared with 
the trend-based scenario, this implies a shift in development from the lower-lying 
cities of Zwolle and Kampen towards cities and villages situated on slightly higher 
ground, such as Steenwijk and Hardenberg. 

A second thematic policy alternative puts health and safety policies first in new 
spatial developments. Residential development is therefore restricted in areas with 
high levels of noise pollution, high levels of personal risk (e.g. near dangerous 
plants, transportation routes for dangerous goods, gas lines or high- voltage power 
lines). Figure 7.2 shows the development of this specific policy alternative and 



122 



A. Koekoek et al. 



1 Overijssel 2040 Trendscenario 2 Thematic input scenario A2 




Fig. 72 The four main steps (clockwise from top left) for analysing the spatial impact of the 
health and safety policy alternative 



compares its distribution of urban areas with that of the trend-based scenario. 
Crucial in this alternative are the areas with safety and health restrictions that are 
safeguarded from new developments by adding negative suitability factors. The 
resulting maps show developments near major roads coming to a halt, in turn leading 
to increased urban development near some smaller cities. Above all, however, 
the analysis shows that enough alternative urbanisation locations are available to 
implement this extensive set of environmental policies. 

A third thematic policy alternative focused on compact urbanisation strategies. 
One such strategy could be to spread residential development among more and 
smaller cities. Another could be to centralise residential developments in the largest 
cities. 

In short, this study revealed that the projected urbanisation could result in local 
conflicts regarding landscape quality, personal safety and water management. The 
Province used the spatial exploration to frame the possible effects of new building 
developments on the ecological main structure, the preservation of the national 
landscapes and on flooding risks. The study provided with insights that the Province 
used in the process to formulate policies for its Regional Spatial Strategy. 



7 Simulation of Future Land Use for Developing a Regional Spatial Strategy 



123 



7.3 Strategic Environmental Assessment 

The Province chose to develop a Regional Spatial Strategy rather than produce 
separate plans for spatial development, water management, mobility, soil and 
environment. Because this strategy sets conditions for other plans and activities 
related to, for example infrastructure, rural developments and Natura 2000 areas, 
the law requires a Strategic Environmental Assessment (SEA) to be made. 

7.3.1 Sustainability Indicators 

The Province decided to expand the scope of this SEA from a strictly environmental 
impact assessment (EIA), to that of a more encompassing sustainability impact 
assessment. Sustainable development is perceived here as a balanced development 
of human, natural and economic capital. These three aspects are often referred 
to as the 'People', 'Planet' and 'Profit' dimensions of sustainability (Elkington, 
1994; Hermans & Dagevos, 2006). The policy impacts for these sustainability 
dimensions were assessed in the SEA for the proposed Regional Spatial Strategy 
in comparison with the current policy. We discuss the applied indicators in more 
detail in Section 7.3.3. 

At the start of the assessment process, the EIA commission advised to: develop 
a list of indicators with which to assess the sustainability impacts; perform 
analyses to reveal current policy shortcomings and dilemmas; elaborate on the 
role of the Province; and specify impacts for Natura 2000 areas. To do so, 
a list of indicators - qualitative and quantitative - for the People, Planet and 
Profit aspects of sustainability was formulated; whenever possible, the use of 
quantitative, reproducible methods was preferred above qualitative methods. Apart 
from assessing sustainability impacts, the SEA also entailed providing input for the 
policy formulation process. This encompassed, amongst other things, an analysis 
of the current policy shortcomings. During the assessment process, workshops were 
organised to present preliminary results and to introduce possible policy alternatives. 

The SEA study was conducted simultaneously with the formulation of the 
Regional Spatial Strategy. This had the advantage that preliminary assessment 
results in the SEA could be used by the Province as input during the process of 
formulating policy for the Regional Spatial Strategy. This iterative working process 
came at the price of requiring regular updates of the impacts on sustainability. 

7.3.2 Developing the Current and New Policy Alternatives 

Land Use Scanner was used in the SEA process to simulate future land 
use according to the conditions under current policy and those under policies 
outlined in the new integrated regional plan (i.e. the Regional Spatial Strategy). 
These simulations were the main input for the assessment of spatially-explicit 
sustainability impacts. The assessment of non-spatial impacts is not discussed in 
this chapter. 



124 



A. Koekoek et al. 



Two land-use simulations were carried out: one based on the continuation of 
current policy (Current Policy alternative), and one based on the introduction 
of the Regional Spatial Strategy (New Policy alternative). The regionalised 
national simulations we describe in Section 7.2.1 were the starting point for these 
simulations. To simplify the assessment process, only one scenario was applied: the 
moderate spatial pressure scenario was selected as the demographic and economic 
developments underlying this scenario are considered to be more probable than 
those associated with the high spatial pressure scenario. A slightly revised version 
of the trend-based scenario of the regional exploration study was used as the 
Current Policy alternative. This revision mainly concerned a small decrease in 
the demand for residential and commercial land in line with the most provincial 
policies (Koomen et al., 2008b). In the New Policy Alternative several additional 
stimulations and restrictions were added to the Current Policy Alternative, partly 
replacing initial policies: 

• a spatial redefinition of the ecological main structure was incorporated that 
includes more detailed spatial demarcation and some areas, important from a 
water quality perspective; 

• the current locations of greenhouses were maintained, although locally some 
expansion is possible; 

• new developments in commercial transportation were stimulated on sites that are 
well-connected to international water, rail and road networks; 

• the compact urbanisation strategy introduced in the spatial exploration project 
and discussed in preceding section was chosen in the New Policy alternative, 
thus concentrating residential developments in the larger towns; 

• recreational developments were stimulated in attractive rural areas, nearby the 
ecological main structure or the national landscapes. 

The differences between the Current and New Policy alternatives are 
considerable, especially for the location of new urban areas (highlighted in Fig. 7.3). 




Fig. 7.3 Increase in built-up area by 2040 under current and new policy alternatives 



7 Simulation of Future Land Use for Developing a Regional Spatial Strategy 



125 



7.3 J Assessing Environmental Impacts 

To what degree are policy aims regarding sustainability met in the Current and 
New Policy alternatives? In the SEA, a list totalling 28 indicators for the domains 
of People, Planet and Profit was used to assess whether the current policies are 
sufficient to meet the sustainability aims. Environmental impacts were calculated 
using several approaches, depending on the availability of established spatial data 
and methods of analysis that could be applied within the time frame of the SEA 
process. Table 7.1 presents a selection of 12 indicators that, to some extent, relied 
on land-use information from Land Use Scanner. These indicators link to policy 
themes for which the spatial distribution of land use is important. 

In most SEAs, impacts are described in a qualitative manner based on expert 
judgement (see, for example, VROM, 2008). This makes the assessment process 
more difficult to reproduce, less transparent and potentially sensitive to subjectivity. 
In the SEA of Overijssel's regional strategy we limited these problems as much 
as possible, striving to maximise reproducibility, transparency and objectivity in the 
impact assessment. Land Use Scanner played a crucial role in this respect, providing 
a quantitative basis for the assessment of many sustainability impacts. The degree 
to which the model could be used in the assessment process is indicated in Fig. 7.4. 

When these were available, we preferred using Land Use Scanner's internal 
modules for the impact assessment. As these internal modules were designed 
to utilise simulation outputs, no additional data transformations were necessary. 
However, for most sustainability indicators no internal modules in Land Use 
Scanner are available as yet. In these cases, the land-use configurations can often 
be used as inputs for an external assessment model or for a quantitative comparison. 



Table 7.1 Overview of sustainability indicators related to land-use information from Land Use 
Scanner 



Sustainability theme/ Aspect/Indicator Relation to land use scanner 



People/Landscape/Landscape openness 
People/Landscape/Preservation national 

landscapes 
People/Landscape/ldentity and diversity 
People/Safety and health/Flood risk 
People/Safety and health/Excess water damage 
Planet/Nature/Condition Natura 2000 areas 
Planet/Nature/Realisation EHS 
Profit/Economy/Zero-grazing livestock area 
Profit/Accessibility /Car accessibility urban 

areas 

Profit/ Accessibility /Public transport 

accessibility urban areas 
Profit/Accessibility /Private motor vehicle 

accessibility rural areas 
Profit/Accessibility /Public transport 

accessibility rural areas 



Internal land use scanner assessment model 
Quantitative comparison 

Quantitative comparison 

Internal land use scanner assessment model 

Quantitative comparison 

Input for external assessment model 

Quantitative comparison 

Quantitative comparison 

Input for external assessment model 

Input for external assessment model 

Input for external assessment model 

Input for external assessment model 



126 



A. Koekoek et al. 



Land use not 
used as input 



2040 Current and New Policy alternative land-use configurations 



e.g. realization of 
sustainable 
energy aims 



Quantitative 
comparison 

e.g. preservation 
National 
Landscapes 



Input for 
external model 

e.g. abiotic 
conditions 
Natura2000 



Input for 
internal module 

e.g. flood risk 



More qualitative 



< Assessment methodology > 



More quantitative 



Impact assessment 



Fig. 7.4 Different applications of Land Use Scanner output for impact assessment in the SEA 



To illustrate the process of assessing sustainability indicators in the SEA, we provide 
below an example of each of the three types of impact assessment that were 
performed in relation to the land-use simulation results. 

An example of an internal assessment module that is available in Land 
Use Scanner is the analysis of flood risk. This module was developed through 
the cooperative efforts of a number of parties, including the Environmental 
Assessment Agency, Delft Hydraulics and the VU University Amsterdam, and 
made available for this project through the Climate changes Spatial Planning 
Programme (www.klimaatvoorruimte.nl). Future land-use configurations were used 
in this module to assess the potential economic damage and number of casualties 
that may result from flooding (see Van der Hoeven et al., 2009). The results of 
the assessment reveal that the differences between the two policy alternatives are 
negligible. A more complete overview of all the land-use-related indicators that 
can be calculated within the Land Use Scanner model is available in the literature 
(Bubeck & Koomen, 2008). 

When no internal assessment model is available, land-use patterns can be input 
in external analysis tools or models. This is, for example, the case for the analysis 
of coherence and abiotic conditions in Natura 2000 areas; achieving favourable 
conditions in nature areas is an important policy goal. Geodan Next developed a 
systematic approach to assess the impact of various spatial developments on basic 
conditions in Natura 2000 areas. Through a predefined set of spatial operations, 
made operational in an ArcGIS script, several aspects of Natura 2000 areas were 
analysed. These relate to the coherence of existing and newly created nature areas; 
differences in nitrogen deposits in these areas; drought effects; and adverse building 
and other developments in those areas that have a hydrological impact on Natura 
2000 areas. The simulations of land use provided information on the expected future 
configuration of nature and new urban areas that was used as one of the input 
variables for this assessment approach. 



7 Simulation of Future Land Use for Developing a Regional Spatial Strategy 



127 



If land-use changes are deemed relevant for a sustainability policy issue, but 
internal and external assessment models are lacking, Land Use Scanner output 
can be used for quantitative comparison. This is, for example, the case for the 
indicator 'preservation of national landscapes'. National and regional policies aim 
to preserve the 'core qualities' of these landscapes by, amongst other things, limiting 
large-scale urbanisation. Therefore, new urban areas in the two national landscapes 
of Overijssel were isolated from the land-use simulations and compared for both 
the trend and policy alternatives. This spatial analysis highlighted differences in the 
amount and location of projected urban development. The New Policy alternative 
contained slightly more building development in the national landscapes than the 
Current Policy. Nevertheless, this difference was considered too marginal to be 
included in the final assessment. 

In cases for which land-use changes are not important for achieving policy 
objectives, simulation output was not used in the assessment process. This is the 
case for issues such as the realisation of sustainable energy aims. 

7.4 Discussion and Conclusion 

Land Use Scanner proved to be a useful tool for supporting the regional planning 
process in the Dutch Province of Overijssel completed in 2009. Different capacities 
of Land Use Scanner were used in two types of study. The spatial exploration study 
focused on the optimisation of land-use patterns according to rather hypothetical 
sets of policy objectives, whereas the Environmental Impact Assessment relied on 
the model's ability to provide probable land-use patterns related to an integrated 
policy alternative that could be used as input in various impact assessments. Both 
studies contributed to the acceptance and formulation of a Regional Spatial Strategy 
for the province. This section briefly discusses experiences with using the model in 
the two studies and concludes with some comments on its general applicability in 
spatial planning processes. 

In the spatial exploration study, Land Use Scanner was used to frame current 
spatial policy objectives and simulate land-use patterns according to coherent sets 
of policy objectives. In cooperation with the Province, regional policy maps were 
used to improve the quality of data input and create regional future land-use 
configurations linked to two trend-based scenarios. These offered a solid framework 
for developing and comparing the sets of spatial policy alternatives. As the same 
economic, demographic and other scenario assumptions - and thus the same 
quantities of land-use change - were used in both simulations, they offer a clear view 
of the specific implications of the spatial policy alternatives. Indeed, comparison 
revealed potential (unwanted) spatial impacts directly related to the proposed policy 
sets. The use of two trend-based reference scenarios allowed the assessment of these 
impacts under different levels of spatial pressure. The inclusion of more diverse 
reference scenarios would have provided an even more robust assessment of the 
impacts of the sets of policy alternatives. But this benefit would have come at the 
cost of obscuring the specific impacts associated with these alternatives since more 



128 



A. Koekoek et al. 



variation in the spatial developments would be introduced. An underestimated, but 
significant, benefit of the modelling process lies in its iterative, open character. 
The application of Land Use Scanner and the presentation of intermediate results 
in workshop sessions provided a forum that encouraged people from a variety of 
professional backgrounds (e.g. water and nature management, urban development) 
to discuss potential developments, propose alternative policies and, above all, clarify 
their suggested planning interventions in a shared spatial environment. In the 
process of constructing a coherent Regional Spatial Strategy, Land Use Scanner 
provided a means for exploring and integrating sector-specific interests and thus 
aided the policy formulation process. 

The SEA of the Regional Spatial Strategy for Overijssel is unique since it is 
the first in the Netherlands in which land-use modelling results were used in the 
sustainability assessment process. SEAs are often fairly qualitative and thus tend 
to be non-transparent in nature. The use of Land Use Scanner made part of the 
sustainability assessment process more transparent and reproducible. The results 
of the SEA and the role Land Use Scanner played in the assessment process 
were greatly appreciated by the provincial policy -makers involved. They judged 
the sustainability impacts to be presented in a clear manner, despite the complex 
assessment process. It is interesting to note that the policy-makers preferred to 
use only one trend-based reference scenario as they found the addition of an extra 
scenario to be confusing . 

The studies also revealed two more general modelling issues that have also been 
encountered in subsequent studies performed for other provinces (Atzema et al., 
2008; Kuijpers-Linde et al., 2008). These relate to the impact that can be expected 
of planning initiatives and the need for integrated data infrastructures . Each of these 
issues is briefly explained below. 

On the issue of impact, a prominent feature of the land-use simulations we 
present in this chapter is the assumption that the proposed spatial policies will 
be actually applied and fully effective. Land Use Scanner thus shows what could 
happen if a certain policy is put in place. In practice, restrictive or stimulative 
spatial policies are not often as effective as intended. Even spatial policies that 
are generally considered successful, such as the national buffer zones intended to 
keep green zones between major cities free from urban development, do not fully 
prevent urbanisation (Koomen et al., 2008a). This issue of effectiveness is especially 
relevant when the aim of the study is to assess the impact of a specific policy. The 
model has the flexibility to account for partial effectiveness of implemented policies, 
but quantitative assessments of the degree of success of different types of spatial 
policies are rarely found in planning literature. This indicates a more generally felt 
lack of ex -post evaluations of spatial plans and concepts. The dynamic, interrelated 
and often vague nature of most spatial policies and plans makes this a difficult issue, 
of course, and establishing the effectiveness of new planning concepts that have 
no historic parallel is obviously close to being impossible. Yet, more attempts in 
this direction would greatly enhance planning in general and our type of land-use 
modelling in particular. 



7 Simulation of Future Land Use for Developing a Regional Spatial Strategy 



129 



Furthermore, consistent and up-to-date data input are a prerequisite for 
successful modelling. In our study the data needed were scattered across different 
governmental departments and stored in range of different data formats. Some 
departments would even use different socio-economic scenarios for various future 
sector-specific outlooks. Comparable studies performed for other provinces have 
provided similar experiences. We hope that the emphasis regional authorities 
currently put on integrated (spatial) data-infrastructures will improve data 
consistency and availability, thus enhancing the potential for spatial analysis in 
general and land-use modelling in particular. 

To conclude, we propose that Land Use Scanner be developed in such a way 
that it becomes a platform for integrated spatial-impact assessment that makes 
sustainability assessment more transparent and robust. For many sustainability 
indicators with a spatial component, model output still plays a modest role, while 
we observe a need for a more quantitative assessment base: only a couple of internal 
Land Use Scanner assessment modules were available for the SEA conducted. 
Development of new modules for, for example, accessibility, safety and health (e.g. 
air pollution and external safety) would significantly increase the value of Land Use 
Scanner as a platform for sustainability assessment. 



References 

Atzema, O., Van Egmond, K., Mommaas, H., Wenting, R., & Kuijpers-Linde , M. (2008). Utrecht 
2040; Strategische notifies in het kader van het traject 'Samen op Weg naar 2040' van de 
Provincie Utrecht. Utrecht: Universiteit Utrecht Faculteit Geowetenschappen. 

Borsboom-van Beurden, J. A. M., Bakema, A., & Tijbosch, H. (2007). A land-use modelling 
system for environmental impact assessment; Recent applications of the LUMOS toolbox. 
Chapter 16. In E. Koomen, J. Stillwell, A. Bakema, & H.J. Scholten (Eds .), Modelling land-use 
change; Progress and applications (pp. 281-296). Dordrecht: Springer. 

Borsboom-van Beurden, J. A. M., Boersma, W. T., Bouwman, A. A., Crommentuijn, L. E. M., 
Dekkers, J. E. C, & Koomen, E. (2005). Ruimtelijke Beelden; Visualisatie van een veranderd 
Nederland in 2030. RIVM report 550016003. Bilthoven: Milieu- en Natuurplanbureau. 

Bouwman, A.A., Kuiper, R., & Tijbosch, H. (2006). Ruimtelijke beelden voor Zuid-Holland. 
Rapportnummer 500074002.2006. Bilthoven: Milieu- en Natuurplanbureau. 

Bubeck, P., & Koomen, E. (2008). The use of quantitative evaluation measures in land-use change 
projections; An inventory of indicators available in the land use scanner. Spinlab Research 
Memorandum SL-07. Amsterdam: Vrije Universiteit Amsterdam/SPINlab. 

CPB, MNP and RPB (2006). Welvaart en Leefomgeving. Een scenariostudie voor Nederland in 
2040. Den Haag: Centraal Planbureau , Milieu- en Natuurplanbureau en Ruimtelijk Planbureau. 

Elkington, J. (1994). Towards the sustainable corporation: Win-win-win business strategies for 
sustainable development. California Management Review, 36(2), 90-100. 

Hermans, R, & Dagevos, J. (2006). De duurzaamheidsbalans van Brabant 2006. Tilburg: Telos. 

Koomen, E. (2008). Spatial analysis in support of physical planning. VU University: Amsterdam. 

Koomen, E., Dekkers, J., & Van Dijk, T. (2008a). Open space preservation in the Netherlands: 
Planning, practice and prospects. Land Use Policy, 25(3), 361-377. 

Koomen, E., Loonen, W., & Koekoek, A. (2008b). Beschrijving en uitwisseling regionale 
Ruimtescanner toepassingen. Amsterdam: Geodan Next. 

Kuijpers-Linde, M., Koekoek, A., & Loonen, W. (2008). Uitwerking ruimtelijke beelden voor het 
nieuwe omgevingsbeleid van Drenthe. Amsterdam: Werknotitie . Geodan Next. 



130 



A. Koekoek et al. 



MNP (2001). Who is afraid of red, green and blue? Toets van de Vijfde Nota Ruimtelijke Ordening 

op ecologische effecten. RIVM-rapport 71 1931005. Bilthoven: RIVM. 
MNP (2007). Nederland Later; Tweede Duurzaamheidsverkenning deel fysieke leefomgeving 

Nederland. MNP-publicatienr.500 12700 1/2007. Bilthoven: Milieu- en Natuurplanbureau. 
Stillwell, J. C. H., Geertman, S., & Openshaw, S. (1999). Geographical information and planning. 

Advances in spatial science. Berlin/Heidelberg/New York: Springer. 
Van der Hoeven, E., Aerts, J., Van der Klis, H., & Koomen, E. (2009). An integrated 

discussion support system for new Dutch flood risk management strategies. In S. Geertman & 

J. C. H. Stillwell (Eds.), Planning support systems: Best practices and new methods 

(pp. 159-174). Berlin: Springer. 
Vonk, G., Geertman, S., & Schot, P. (2005). Bottlenecks blocking widespread usage of planning 

support systems. Environment and Planning A, 37, 909-924. 
VROM (2008). PlanMER Structuurvisie Randstad 2040: naar een duurzame en concurrerende 

Europese topregio. Den Haag: Ministerie van Volkshuis vesting, Ruimtelijke Ordening en 

Milieubeheer (VROM). 



Chapter 8 

Lessons Learned from Using Land-Use 
Simulation in Regional Planning 

Chris Jacobs, Arno Bouwman, Eric Koomen, and Arjen van der Burg 



8.1 Introduction 

What happens if we maintain current spatial planning policies? How will new 
planning concepts turn out? Where is potential for urban expansion and where 
are valued landscapes under threat? These are questions a planner might often 
ask. Models that simulate future land use, such as Land Use Scanner, may 
provide answers to such questions. Until recently, such land-use simulations in the 
Netherlands were mainly performed on the national level, as described in Chapters 4 
and 5 of this book. Studies such as these were often done in preparation for or as 
an evaluation o/national spatial-planning decisions. 

The revised Land Use Scanner model, with its more detailed 100-m grid 
resolution, is now more suitable for regional planning applications. The model's 
potential to simulate regional developments was first explored by PBL in 2006 
in cooperation with the province of Zuid-Holland (see Borsboom-van Beurden, 
Bakema & Tijbosch, 2007; Bouwman, Kuiper & Tijbosch, 2006). Several other 
regional studies followed in 2008. This chapter briefly describes three types of 
regional applications of Land Use Scanner. The first type was performed by Geodan 
Next in conjunction with PBL and was used to provide input for regional strategic 
visions of three Dutch provinces (see Section 8.2). A second type, performed by 
the Dutch Ministry of Housing, Spatial Planning and the Environment (VROM), 
was aimed at exploring spatial planning concepts in the preparatory phase of the 
'Randstad 2040' project (see Section 8.3). In the third type, the VU University 
Amsterdam used Land Use Scanner to integrate sector-specific climate adaptation 
measures for the Province of Groningen (see Section 8.4). Although each of the 
three types of projects had different objectives, all made use of Land Use Scanner 
and these experiences are discussed in this chapter with reference to: 



C.Jacobs (El) 

Department of Spatial Economics/SPINlab, VU University Amsterdam, 
DeBoelelaan 1105, 1081 HV Amsterdam, The Netherlands 
e-mail: c.g.w.jacobs@vu.nl 

E. Koomen, J. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 131 
Practice, The GeoJournal Library 101, DOI 10.1007/978-94-007-1822-7_8, 
© Springer Science+Business Media B.V. 201 1 



132 



C. Jacobs et al. 



• the purpose of the project; 

• the kind of spatial developments that were depicted; 

• the policy alternatives that were considered; and 

• the modelling- and planning- related lessons learned. 

8.2 Providing Input for Regional Strategic Visions 

In 2008, a new Dutch spatial planning act came into force. In this new act, 
municipal, provincial and national administrations are obliged to create a strategic 
vision as a guiding 'master plan' for legally binding spatial plans and other policy 
measures (e.g. providing subsidies). Furthermore, within the boundaries of national 
policies that are formalised in a spatial by-law, the act gave regional authorities 
greater responsibilities in the planning process. The new spatial planning act 
demands the definition of strategic regional and local visions that are new in legal 
status and in the magnitude of effect on local policies. In 2008, a number of 
provinces (Dutch regional administrations) began defining these strategic visions, 
based on almost a century of regional planning tradition. Geodan Next supported 
the development of these visions by providing simulations of future land use. 

Spatially explicit depictions of future spatial developments in a province, 
based on current socio-economic trends and spatial policies, offered a useful 
and thought-provoking starting point for the development of strategic visions. 
These depictions were based on assumptions and quantified explorations of 
trends taken from the 'Second Sustainability Outlook on the future of the 
Netherlands' study of PBL (MNP, 2007), which is also discussed in Chapter 4. 
In the provincial explorations, national spatial depictions were complemented 
with province-specific information. Such information included regional policies on 
restrictions and incentives, more detailed figures on the demand for land per capita 
and region-specific knowledge on regional spatial developments. 

In all studies, simulated autonomous developments were shown on maps of 
future land-use patterns related to two trend-based scenarios. Both scenarios attach 
the same importance to current spatial policies, but they differ in the demand for 
space. The simulations show, therefore, potential autonomous developments under 
conditions in scenarios of moderate development pressure (Baseline scenario) and 
high development pressure (High Development Pressure scenario) previously used 
in the 'Second Sustainability Outlook' study. Besides the land-use maps, many 
other maps were produced to depict possible impacts of the projected spatial 
developments. These maps focus on themes such as urban sprawl, deterioration of 
landscape values and flood risk. For a more elaborate description of these maps 
studies, see relevant publications (Atzema, Van Egmond, Mommaas, Wenting & 
Kuijpers-Linde, 2008; Koomen, Kuijpers-Linde & Loonen, 2008b; Kuijpers-Linde, 
Koekoek & Loonen, 2008) and Chapter 7 by Arjen Koekoek, Eric Koomen, Willem 
Loonen, and Egbert Dijk, this volume. 

For the provinces of Overijssel and Drenthe, the depictions of autonomous spatial 
developments were supplemented with studies on the possible spatial consequences 



8 Lessons Learned from Using Land-Use Simulation in Regional Planning 



133 




Fig. 8.1 Land-use simulations for the province of Drenthe in 2040 according to employment (left) 
and climate adaptation policy alternatives (Source: Kuijpers-Linde et al., 2008) 

of new policy alternatives. Together with this supplementary work, the studies show 
not only the tasks at hand for these provinces but also explore possible solutions. 
The simulations of alternative policies for Drenthe and Overijssel were only done 
for the High Development Pressure scenario as this was thought to more clearly 
depict potential developments: the simulations show a kind of worst-case scenario 
with regard to environmental impact. 

By way of example, Fig. 8.1 shows two policy alternatives that were simulated 
for the province of Drenthe. The maps show the potential spatial patterns associated 
with a strong focus on employment (stimulating, for example, a concentration of 
industrial/commercial activities and competitive large-scale farming) and climate 
adaptation (e.g. promoting robust nature areas and pastures in small river valleys). 
(Other policy alternatives stressed the recreational and natural values of the 
province.) Land-use simulations were also developed for the province of Overijssel 
that stressed specific policy ambitions. Elements of these, often rather extreme, 
policy alternatives were eventually included in the final policy alternative of the 
strategic vision. In the compulsory environmental impact assessment (EIA) of the 
regional plan, this policy alternative was assessed on sustainability aspects, which 
are briefly described in Chapter 7. 

8.2.1 Lessons Learned 

Trend-based simulations were used to understand which issues should be on 
policy -makers' agendas. In addition, simulations of policy alternatives were 
developed to explore possible planning alternatives. The simulation results 
were highly praised by the provincial authorities. According to the policy 
makers involved, the simulations clearly showed the extent of expected spatial 
developments, making the simulations useful for estimating potential impacts 



134 



C. Jacobs et al. 



of spatial policies. But not only the final land-use maps proved to be useful. 
Especially discussions about intermediate results were highly valued by the 
provincial policy-makers: these helped them understand and prioritise issues on 
their spatial planning agenda. For example, the environmental effects of light 
pollution initially seemed important for the regional planners of Overijssel. 
However, during discussions about the land-use modelling results, the spatial 
impacts of light pollution proved hard to define and were, in general, relatively 
small compared to the impacts of urbanisation and urban sprawl. Subsequently, light 
pollution lost priority to other issues in the Overijssel modelling process. 

The simulations raised a number of research questions and led to suggestions 
for model improvements that deserve more attention in following projects. One 
important question relates to the demand for space that is specified in the 
trend-based scenarios used in the simulations. These figures for demand were 
initially taken from the 'Second Sustainability Outlook' study (MNP, 2007). 
However, the figures in that study are for expected demand for residential land at the 
national level, not at regional level. Furthermore, the evaluation of local suitability 
values for specific types of land use might be improved if the balance between 
factors such as accessibility, quality of the neighbourhood and policy restrictions 
is thoroughly founded on empirical findings. This understanding may help identify 
more precisely future developments in urbanisation. 

Experiences from the provincial studies point to a number of improvements of 
the modelling tools used. Ideally, a direct link between Land Use Scanner and the 
models that produce the sector-specific demand for land should be established. By 
that means, direct feedbacks could be applied between consequences of spatial 
developments and regional demands for land. This would enhance Land Use 
Scanner's capacity to match developments between, for example, the housing 
market and markets for industrial and commercial land. 

Currently, Land Use Scanner only simulates mono-functional types of land 
use. Discussions with policy-makers have made it clear that the combination 
of different types of use on the same piece of land is an increasingly popular 
approach for reconciling the many different demands for space. Incorporation of 
multifunctional land-use types in the model is thus desirable to be able to simulate 
such novel approaches relating to, for example, nature management on agricultural 
land or water management in nature areas. Furthermore, in a number of simulated 
policy alternatives, accessibility measures and related developments were deemed 
important. To facilitate simulations that focus on the impact of accessibility on 
transport-related policy alternatives, further integration of Land Use Scanner and 
a transport model ought to be considered. 

8.3 Exploring Spatial Planning Concepts 
for the Randstad Conurbation 

The urban core in the Netherlands, known as the 'Randstad', presents very specific 
spatial planning challenges. The special status of the Randstad conurbation has been 
acknowledged at a national level by Dutch policy-makers, resulting in the need for 



8 Lessons Learned from Using Land-Use Simulation in Regional Planning 



135 



an integrated strategic plan for spatial developments there. Consequently, the Dutch 
Ministry of Housing, Spatial Planning and the Environment (VROM) launched the 
so-called 'Randstad 2040' study in 2008. The VU University Amsterdam used Land 
Use Scanner for an initial depiction of various regional planning concepts, whereas 
VROM applied Land Use Scanner to perform a more elaborate exploration of the 
regional possibilities for urban densification and expansion. Both simulations were 
used as input for the 'Randstad 2040' study. 

8.3.1 Depicting Regional Planning Concepts 

At the very start of the 'Randstad 2040' study, three spatial concepts for the 
Randstad conurbation were publicised by the Dutch minister responsible for spatial 
planning (Cramer, 2008). These concepts were seen as possible solutions for spatial 
problems and were to serve as outlines for the further spatial development of the 
Randstad conurbation. The concepts were: 

• extending current city boundaries; 

• concentrating urban development along major transport corridors; and 

• intensifying current land use within existing city boundaries. 

The SPINlab of the VU University Amsterdam simulated future land use while 
taking these planning concepts into account. The simulations were adaptations taken 
from the 'Second Sustainability Outlook' study (MNP, 2007). SPINlab's goal in 
doing so was to depict these concepts, taking into account the amount of land 
needed for urban use in the Randstad conurbation. This enabled policy-makers 
to understand the extent of demand for urban land in the area and helped create 
a sense of urgency for finding solutions to the problems that accompany that 
demand. 

Such a sense of urgency can be created when the severity of the situation is 
shown, so a worst-case scenario was used in simulating the effects of the spatial 
policy concepts listed above. Many problems in the Randstad stem from lack of 
space for urban land use and, therefore, the worst-case scenario is one in which 
demand for urban land is highest. Consequently, the simulations performed assumed 
the same High Development Pressure scenario used in the 'Second Sustainability 
Outlook' study. In this scenario, large portions of land in the Randstad conurbation 
are presumed to be needed for new residential, industrial and commercial use. 

In the simulations, the impacts of putting the three concepts into practice had 
to be translated into logic rules and values to be used in Land Use Scanner. 
This was done in cooperation with an expert from VROM, the ministry involved 
in the 'Randstad 2040' study. Expert judgement was used instead of academic 
exploration of these logic rules and values because the planning-process dynamics 
demanded relatively fast results. A rapid and clear depiction of planning concepts 
and the regional demand for urban land was deemed more important than a 
thorough analysis and assessment of all the location factors and spatial processes 
involved. 



136 



C. Jacobs et al. 



8.3.2 Lessons Learned 

Use of Land Use Scanner in this initial phase showed that the model is a powerful 
tool for quickly visualising planning concepts with specific restrictions . Especially 
depictions in which the planning concept and land-use simulation results were 
combined (see Fig. 8.2) turned out to be very communicative. It should be noted, 
however, that the validity of such applications is limited as they are strongly 
influenced by the relative values the modeller applies to the various components 
of the suitability values used when producing visualisations of planning concepts. 
They do not show the likely consequences of planning concepts, but rather visualise 
planning concepts that would otherwise remain rather vague, qualitative notions. 
While the land-use simulations were being done, the preference of VROM turned 
towards one of the policy concepts explored here: intensifying land use within 
current city boundaries. This brought with it the need to incorporate more detail on 
the impact of land-use densification in the model. The resulting simulations, which 
were performed by VROM itself, were used as input in the 'Randstad 2040' study 
(described in the following subsection). 




Fig. 8.2 Depiction of new urban areas in the transport axes corridor concept. Orange indicates the 
corridors in which urban development was simulated 



8 Lessons Learned from Using Land-Use Simulation in Regional Planning 



137 



8.33 Exploring Regional Possibilities for Urban Densification 
and Expansion 

For the development of a structural vision for the 'Randstad 2040' study, the 
Ministers of Housing, Spatial Planning and the Environment (VROM), Economic 
Affairs (EZ), Transport, Public Works and Water Management (V & W) and 
Agriculture, Nature and Food Quality (LNV) agreed upon a number of main 
assumptions relating to the long-term spatial development and policy priorities of 
the area: 

1 . A large permanent demand for land is expected; 

2. The quality of the demand for land is becoming increasingly important; 

3. Specialisation, and (spatial) differentiation are important in realising economic 
growth objectives; 

4. Internal and external accessibility must be substantially improved; 

5. Improvement of public transport is very important to improve accessibility; 

6. More space is needed for safety (flooding risks near the coast and rivers) because 
of the future effects of climate change; 

7. Urbanisation should be concentrated, but land is needed to create spacious green 
living and working environments; 

8. A focus on major restructuring and transformation of the existing built-up area. 

These priorities are described in the 'guiding principles' and the Government's 
'spatial choices' of the structural vision 'Randstad 2040' study (VROM, 2008). 
For some choices, more detailed information was needed. This was the case with 
priority number 7, dealing with concentration of urbanisation. The aim of this 
policy objective was to make maximum use of existing built-up areas in order to 
preserve open landscapes for agriculture, nature and recreation. VROM used Land 
Use Scanner to explore the limits of densification and expansion of urban land. 

This application encountered two methodological issues. The first issue concerns 
the limits to incorporate densification in the model, specifically the possibility of 
adding residences and workplaces to an existing built-up area. Land Use Scanner 
cannot directly analyse the possibilities for densification because it only includes 
a limited number of urban land-use classes, each of which has a fixed density of 
use and occupation. The projected densification would, nevertheless, take place 
mainly within the same land-use types: single- or multi-story office buildings are, for 
example, both classified as commercial land use. The model can, however, show the 
effects of urban densification and avoided urbanisation on the preservation of open 
space. In comparison with a Baseline scenario it can show the lack of expansion of 
new residential and commercial areas. 

The second issue concerns the difficulty of determining limits to expansion 
that are socially acceptable. This is important because the demand for (relatively) 
spacious green urban housing is large, while the supply of space is scarce. Typically, 
such green urban housing environments have densities of about 17 dwellings per 



138 



C. Jacobs et al. 



hectare, in contrast with 39-56 dwellings per hectare for innercity urban land use 
(VROM, 2001). Usually, to preserve open space, various restrictions are imposed 
that lead to higher dwelling densities and the protection of specific open spaces 
(Koomen, Dekkers & Van Dijk, 2008a). These limits are incorporated in the model 
as part of the demand for residential land (through the included dwelling densities) 
and by specifying (non)suitable locations (through the inclusion of spatial plans and 
restrictions). The model parameters related to these issues reflect the expected future 
importance of preserving open space in spatial planning. They imply, therefore, an 
outcome of the public debate on balancing the degree of individual freedom allowed 
in choosing a house in a low-density, green urban residential environment and the 
societal importance of having open spaces. The assumptions related to the dwelling 
densities included in the model are underpinned with a statistical analysis that is 
explained further on. 

The configuration of the Land Use Scanner based analysis of limits to 
densification and expansion is similar to the configuration used in 'Second 
Sustainability Outlook' study carried out by PBL (MNP, 2007; and Chapter 4 
by Rienk Kuiper, Marianne Kuijpers-Linde, and Arno Bouwman, this volume). 
However, the configuration had to be revised by VROM to meet the specific 
objectives of the 'Randstad 2040' study. To do so, first a 'Randstad 2040' specific 
set of valuations of factors for the model suitability maps were derived from the 
'Second Sustainability Outlook' study. The so-called Combination Map in that latter 
study optimises land-use patterns according to a number of policy objectives and 
describes a potentially sustainable design for spatial development under moderate 
and high economic and demographic growth rates. The weights assigned to various 
spatial planning concepts in the Combination Map corresponded well with the 
policy priorities developed in the 'Randstad 2040' study and were thus maintained. 
The weights of the suitability maps reflect, for example, a societal preference for 
concentrated urbanisation. The implementation of this study's policy alternatives 
focused, therefore, on introducing variations in the demand for residential land to 
simulate scenario-specific preferences for various residential land-use types. 

To explore the spatial impact of the uncertainty in the demand for housing, 
the Baseline and High Development Pressure scenarios of 'Second Sustainability 
Outlook' study were used as they represent a considerable bandwidth of potential 
urban development. The demand for housing in these growth scenarios is based on 
demographic projections and assumptions related to, for example, household size, 
preferred dwelling types, dwelling densities and current densification rates. 

The latter two are especially relevant in relation to the 'Randstad 2040' study 
and their treatment in the model is discussed here. Dwelling density denotes the 
number of residences per hectare. In the 'Randstad 2040' study, the demand for new 
residential areas of administrative regions (COROP or NUTS-3) is based on current 
dwelling densities, derived from clusters of regions with approximately common 
dwelling densities. These clusters do not necessarily form contiguous regions; 
however, dwelling densities in the west of the Netherlands (generally included in 
Cluster 1) are higher on average. See Table 8.1 for an overview of these results. 

The table clearly shows the differences between densities used in the 'Second 
Sustainability Outlook' study and those obtained for the 'Randstad 2040' project. In 



8 Lessons Learned from Using Land-Use Simulation in Regional Planning 



139 



Table 8.1 Results of the 'Randstad 2040' analysis of dwelling density (residences per hectare) 
compared to the 'Second Sustainability Outlook' study. The types of residential areas listed here 
are defined according to Oskamp, Poulus, and Van Til (2002) 

Second sustainability 

outlook Results density analysis 



National National Average Average Average 

Type of residential area average average cluster 1 cluster 2 cluster 3 



Central urban area 43 41 65 40 24 

Non-central urban area 38 40 53 37 29 

Green urban area 25 23 32 23 18 

Village central area 24 19 26 19 16 

Countryside 24 13 18 14 11 



particular, the new national average dwelling densities for housing in the countryside 
and villages obtained from Land Use Scanner in the Randstad project are lower 
than those used previously. This implies that the demand for land for these types of 
housing increases . 

The densification rate is considered to be completely dependent on the share 
within existing urban areas taken by new residences. This share indicates which 
proportion of new houses built will not claim additional open space. Between 2002 
and 2005, around 33% of new dwellings were built within existing urban areas 
(Snellen, Farjon, Kuiper & Pieterse, 2006) in the Netherlands. In the Randstad 
conurbation that densification rate was lower (26%). Especially in the provinces of 
Flevoland (which only contains new settlements that were built on reclaimed land 
since the 1940s) and Zuid-Holland (where, excluding possibilities for densification 
in defunct harbour areas in the Rotterdam area, densification may have reached 
its limits), the share of new dwellings built within existing urban areas was 
lower. Based on existing, approved municipal zoning plans (NIROV, 2005), the 
'Second Sustainability Outlook' study assumes that about 20% of new residential 
construction will take place in the existing (as of 2000) built-up area for the entire 
period 2002-2010; much lower rates are expected in subsequent years. Currently, 
the expected densification is approximately 10% for the whole period 2002-2040. 
In addition to a scenario that simulates land use change with these densification 
expectations, in the 'Randstad 2040' project a much more ambitious densification 
rate of 50% of residential land claims has been simulated. An additional measure 
for saving space was explored by shifting 50% of the demand for green urban and 
countryside housing to the types central and non-central urban areas. Together we 
call it the 'ambitious policy' variant in the 'Randstad 2040' study. This ambitious 
policy variant was applied to the moderate and high growth scenarios and compared 
to two current policy-based variants, thus generating a total of four scenarios 
(see Fig. 8.3). The trend-based scenarios are similar to those in the 'Second 
Sustainability Outlook' study, only differing in the regional division and dwelling 
densities used. 



140 



C. Jacobs et al. 



Development 





A: Medium scenario (TM) 
with 'current policy' 
spatial development 






B: High scenario (GE) 
with 'current policy' 
spatial development 




Pressure on space 














C 


C: Medium scenario (TM) 
with 'ambitious policy' 
spatial development * 






D: High scenario (GE) 
with 'ambitious policy' 
spatial development * 


D 



Spatial 



Fig. 8.3 The scenario framework for the 'Randstad 2040' study containing two axes: One 
indicates variations in ambition level of implemented spatial policies; the other indicates pressure 
on space. * and 50% intensification and 50% of the demand for green urban and rural housing 
moved to urban housing 



In the 'Randstad 2040' study, the demand for residential land for all 
four scenarios was specified for an aggregation level that overall distinguishes 
between 31 regions in the country. This regional aggregation comes from the 
so-called 'Socrates' model (Poulus and Heida, 2005). This regional division was 
considered more appropriate than the larger residential regions applied in 'Second 
Sustainability Outlook' study because the latter study allowed for considerable 
spatial spread of allocated residential land use. Simplified forms of two of the 
resulting land-use maps are shown in Fig. 8.4. 



8.3.4 Lessons Learned 

Land Use Scanner proved to be useful as a research tool for exploring extreme policy 
scenarios. The densification of 50% specified in the 'ambitious policy' variant is an 
example of such an extreme scenario; the feasibility of this degree of densification 
has not been thoroughly investigated, however. The authorities of the major urban 
areas in the Randstad conurbation, i.e. the environs of Amsterdam (the so-called 
'North wing' of the Randstad) and the environs of Den Haag and Rotterdam (the 
so-called 'South wing'), also have high ambition levels for densification. The city 
of Rotterdam even has a target of 80% densification for the period 2010-2030. 
These 'limits of densification are thus not easy to assign. The opportunities and 
possibilities for densification depend on many other things. For example, the 
architectural design of residential areas has some influence on densification options . 
Achieving maximum densification is theoretically possible, but the conditions for 
maximum densification are unknown, and it is doubtful whether such a degree of 
densification would be accepted by society. Nevertheless, as the High Development 
Pressure scenario proves, it is clear that if the trend of limited densification observed 
in recent years continues, a large throng of houses will once again be sited on scarce 
and valuable open space. This has been reality since the end of the Second World 



8 Lessons Learned from Using Land-Use Simulation in Regional Planning 



141 




Fig. 8.4 Outcomes of simulations with scenarios C and B of Fig. 8.3 for the 'Randstad 2040' 
study 



War, with the exception of only a few periods when a higher degree of densification 
was actively pursued . 

The maps that resulted from simulating urban densification policies are based 
on assumptions about the proportion of densification versus expansion. They 
show the long-term difference between development under the current policy and 
development under an ambitious densification policy. The analysis needs to produce 
the maps facilitates understanding of the playing field in which other relevant 
interests must be considered. The playing field is presented here in its spatial 
dimension: a map and the underlying figures on the demand for land. More insight 
into the limits of densification in relation to expansion contributes to making 
well-founded decisions about a desirable urbanisation strategy for the Randstad 
conurbation. 



8.4 Integrating Climate Adaptation Measures 

The Province of Groningen has supplemented their strategic regional plan with an 
exploration of spatial adaptations to ameliorate the effects of climate change. During 
a number of workshops the ideas for these spatial adaptations were shared by experts 
on, for example, energy, ecology, agriculture and climate. Two kinds of workshops 
were organised to define a coherent set of adaptation measures. Sector-specific 
sessions focussed on generating solutions for problems in a specific domain, while 




more general sessions focussed on establishing an integrated vision for adapting to 
climate change in Groningen. 

In the sector-specific sessions, experts proposed measures for adaptations in 
societal domains such as energy generation, ecology, coastal management and water 
management. The proposals generated a large number of maps showing threats 
and proposed counter-measures for a number of specific themes. In a total of four 
sessions, 28 separate maps were produced, each with a (partially) different set of 
adaptation measures for Groningen province (For examples of these types of maps, 
see Fig. 8.5.) These digital maps were subsequently incorporated in a Geographic 
Information System (GIS). 

The general sessions used a backcasting approach (Robinson, 1982) as their 
starting point. In the first session, the participants outlined two visions of 
climate-change-driven adaptations for Groningen in the year 2100, taking into 
account different strategies for adapting to a rise in sea level. In the second session, 
the same groups decided which policies were necessary to establish their visions, 
thereby creating a third vision. This implementation-oriented session also enabled 
the participants to rethink and, if necessary, adjust their visions. In the end, three 
visualisations of a climate-change-proof Groningen were produced and their related 
policy measures proposed (Van 't Klooster, Pauw & Roggema, 2008). 

A second round of general sessions focussed on integrating the results of 
all preceding sessions. The integration step was initially aimed at clarifying 
and integrating the sector-specific adaptation measures that differed substantially 
in spatial and thematic resolution: some maps showed realistic, detailed, 
local adaptation measures, whereas others only sketched concepts of potential 
adaptations . To merge these differing views on adaptation into one coherent set of 
adaptation measures and thereby create a 'climate-proof Groningen, a GIS-based 
spatial integration approach was followed (Jacobs, Koomen & Roggema, 2009). 
In a concluding integration session, participants from the previous workshops 



8 Lessons Learned from Using Land-Use Simulation in Regional Planning 



143 



confronted the integrated set of climate adaptations with the more visionary 
backcasting maps to analyse their robustness under different sea-level adaptation 
strategies (Roggema, 2009). 

VU University Amsterdam supported this GIS -based integration process by 
providing tools, information and analyses of results to the participants of these 
sessions. GIS makes it easy to share, compare, create and process large quantities 
of information. All information that was produced during the sector-specific and 
backcasting sessions was made available to the participants. Furthermore, the results 
of previous scenario-based Land Use Scanner simulations (documented in Riedijk, 
Van Wilgenburg, Koomen & Borsboom-van Beurden, 2007) were included in the 
application. Important steps in the integration approach are: (1) the selection of 
spatially explicit adaptation measures; (2) a more exact definition of their nature 
and location; and (3) an analysis of possibilities for combining different adaptation 
measures. 

The first two steps involve a careful analysis of the proposed measures and a 
conversion into spatial demand for a certain area on the map - a process similar 
to the configuring a Land Use Scanner simulation. To facilitate the third step, 
special overlay analysis was performed to assess where demand for land for different 
adaptation measures potentially overlapped. This was done by dividing the province 
into grid cells of various resolutions and then counting the number of sector-specific 
land claims per cell of a specific size. The analysis then showed how many sectors 
claim land for adaptation measures in a given area (Fig. 8.6). A resolution of 5 km x 
5 km was considered to be most effective for providing this information on potential 
spatial pressure. 

In a next step, potential conflicts were identified with these spatial pressure maps. 
These conflicts were resolved with the help of local and sector-specific experts by 
either moving or removing specific land claims. This resulted in a map of integrated 
land claims for adaptation to climate change, which was then overlaid with the three 
adaptation strategies from the backcasting sessions in the concluding integration 
session described above. 

Simulation results from Land Use Scanner were used to sketch the constraints 
on adaptation measures posed by current and projected patterns of urbanisation. 
Human occupation clearly limits the potential for climate adaptation measures at a 
location: policies that aim to move people, or reclaim built-up land for economically 
less-intensive land uses are unlikely to be successful. Knowing where people will 
live and work is, furthermore, necessary to make clear which areas will need strong 
measures for protection from flooding. Therefore the location of existing and likely 
future urban areas was added as an additional layer in the overlay analysis of 
adaptation measures. 

A so-called tangible user interface or TouchTable was used in two integration 
sessions to facilitate interaction between a limited number of participants and the 
spatial data stored in a computer (Scotta, Pleizier & Scholten, 2006). The interface 
displays the collected spatial information in a GIS application on a table surface 
and allows participants to sketch and comment on the projected maps. Anytime 
a participant touched the table, it registered the specifics of that touch (e.g. place 



144 



C. Jacobs et al. 



Number of sector specific land claims 




Fig. 8.6 The number of land claims per 5 km x 5 km grid cell following from sector-specific 
measures for climate-change adaptation and projected future urban areas in Groningen province 

on the table, movement, number of elements that were touched). The software on 
a computer connected to the table then translated the participants' touches into 
commands to the GIS application. 

The organisers found that using a TouchTable in combination with a GIS had a 
number of advantages over using printed maps or a regular computer screen: 

• The table setup stimulated participants to stand around and form a group. It was 
fun to use and had some 'gadget-value', which enticed people to join in and 
participate. 

• Using a GIS enabled the sharing and viewing of a wide array of spatial 
information. Furthermore, the GIS facilitated the easy comparison and editing 
of information and it allowed changes to be reversed and additions made. 

8.4.1 Lessons Learned 

Several lessons for facilitating multidisciplinary planning are evident from the 
climate adaptation study for the province of Groningen. First and foremost, the 
overlay analysis used calls for spatially explicit and non-ambiguous adaptation 
proposals. This forced participants to reflect on the sector-specific measures 
proposed in previous sessions. Some of the adaptation measures that were initially 



8 Lessons Learned from Using Land-Use Simulation in Regional Planning 



145 



proposed were too ambiguous for subsequent analysis. Not all proposals were, 
for example, explicit in where their adaptation measures required land, or in how 
much land was needed. A number of groups proposed multiple scenarios for 
climate change adaptations, while the proposals of other groups entailed system 
demands instead of reservations and adaptations of land . Still other groups proposed 
adaptation measures that took into account all land in the province. Reflection on 
the previously created sector-specific proposals stimulated participants to make new 
proposals for various adaptation measures. The integration process thus encouraged 
reflection on some proposals and eventually led to better definitions of sectoral 
claims and ideas. 

The climate adaptation process for the province of Groningen first focused 
on stimulating creativity and only later demanded explicit and definite proposals 
from the many sector-specific groups. This benefit of maximum creativity and 
flexibility in the sector-specific sessions came at the cost of relatively intensive and 
time-consuming integration sessions. The process could have been more efficient if 
the same technology had been used in all sessions. 

Using a GIS in combination with a TouchTable and employing overlay analysis 
appears to be useful for multidisciplinary planning sessions. In the integration 
sessions, the table was effective in actively involving people in the sessions and 
the GIS gave relatively easy access to a large number of datasets. The results of the 
spatial overlay analysis were effective in guiding the participants through possible 
conflict areas and were very useful in explaining to users the need for being spatially 
explicit and definite . 

Land Use Scanner results were only used to sketch some of the boundary 
conditions for the development of adaptation measures . A stronger role for Land Use 
Scanner was considered, for example by using the model to explore the conditions 
necessary for developing the proposed adaptation measures. This role was not 
pursued as maximum creativity was preferred in the development of adaptation 
options. The structured approach of Land Use Scanner in, for example, defining 
spatial demands and analysing whether they coexist or compete was, however, 
helpful in structuring the integration sessions. Furthermore, the overlay analysis 
indicating potential spatial pressure, the extreme flexibility in defining the (location 
of) adaptation measures, and the application of the TouchTable suggest that the 
current modelling framework of Land Use Scanner might need to be extended. 
This project made clear, once again, that quantitative modelling and structured, 
interactive design can have similar objectives and follow a comparable approach. 
For providing creative outlooks for the mid-term future, they offer complementary 
tools, something already demonstrated in an earlier comparative study on Land Use 
Scanner and design methods (Groen, Koomen, Ritsema & Piek, 2004). 

8.5 Parallels 

In 2008, Land Use Scanner was used to support a number of regional-planning- 
related projects, of which several are presented in this chapter. Although the projects 
described in this chapter varied in their goals and the way the model was used, 



146 



C. Jacobs et al. 



some parallels in their modelling characteristics, as well as their main purpose, 
can be observed. A description of these follows in this section. With reference to 
these parallels, we also discuss here several more general modelling issues related 
to uncertainty, the supposed success of planning concepts and enhancing the current 
model layout. 

8.5.1 Modelling Characteristics and Main Purpose 

An obvious parallel among the modelling characteristics can be observed in the 
spatial and temporal resolution used while modelling. All projects used a 100 m x 
100 m resolution, and it seems that this scale is sufficient for performing regional 
studies in a way that is meaningful to planners. The same temporal resolution 
was also used for all projects. The year 2040 is the furthest into the future in the 
strategic planning studies discussed in this chapter. Only the Groningen case, with 
its strong emphasis on climate-driven changes that typically become prominent with 
a longer time-scale, attempts to look further into the future: as far as 2100. As 
socio-economic developments are highly uncertain over such long periods of time, 
land-use simulations were not deemed feasible. Yet, notions of socio-economic 
change can play an important role in studies on the potential impact of climate 
change. Initial attempts to create such long-term simulations with Land Use Scanner 
have therefore been made (Van der Hoeven, Jacobs & Koomen, 2008), but these 
should first and foremost be considered as philosophising. 

Another common characteristic of the land-use simulations we have presented in 
this chapter is their strong focus on urban development: in all studies, new urban 
areas are clearly the most important derivative of land-use simulations. Most of the 
discussions on land-use simulation with the relevant regional authorities centred on 
the strikingly large demand for residential, commercial and industrial areas. 

The main purpose of modelling is normally not to depict the most likely future 
land-use changes according to observed trends or socio-economic scenarios. Rather, 
the emphasis is on 'what-if?' types of studies. These studies aim to show what could 
happen if certain adjustments were to take place. Communicating planning concepts 
rather than depicting the most likely developments seems, therefore, most important. 
Furthermore, the intermediate results that were generated in this process proved to 
be useful ingredients for stimulating discussion and provoking thought about the 
implications of policy interventions. The process of depiction did, however, appear 
to be useful for helping to structure and filter some of the issues and interests 
policy -makers have to deal with. 

8.5.2 Modelling Issues 

Uncertainty is hardly addressed in the simulations described in this chapter. 
Apparently, elaborating on a range of possible futures is not desirable in the context 
of policy-making. Policy-makers seem, in fact, averse to being confronted with 



8 Lessons Learned from Using Land-Use Simulation in Regional Planning 



147 



uncertainty. They prefer a clear, singular reference point against which policy 
choices can be tested. Great confidence is therefore placed in the baseline conditions 
used for 'what-if?' simulations. Several flaws can, however, be observed in these 
trend-based land-use simulations that, for example, underlie the studies presented in 
Sections 8 .2 and 8 .3 . Extrapolation of current trends and policies is not necessarily 
likely to produce plausible results since past changes offer no guarantee for the 
nature of future changes . Using only one trend obscures the fact that any projection 
of future socio-economic conditions has a high degree of uncertainty. In fact, the 
so-called trends are to some extent based on scenario assumptions (e.g. the High 
Development Pressure trend line that is used to model demand for residential land is 
related to the so-called Global Economy scenario). To combine these scenario-based 
assumptions with other trends that do not follow the same scenario conditions can 
be considered inconsistent. 

Such issues are not very worrisome when simulations are only used to 
illustrate policy concepts as we have discussed above. However, when land-use 
simulations have a more substantial role in the formulation process (for example, 
to highlight potential controversial spatial developments or the impact of policies in 
Environmental Impact Assessments), the definition of more solid reference points 
becomes essential. For long-term projections (e.g. 30 years or more) we advocate 
using a small set of, for example, two to four scenarios covering most of the 
uncertainty related to key socio-economic conditions to address the full bandwidth 
of potential land-use changes. To meet policy -makers' need to limit the uncertainty 
in studies related to concrete planning initiatives for the short term (e.g. 15-20 
years), it would be useful to create a well-founded and validated reference scenario 
that incorporates current trends and policies . Explicit attention should then be paid 
to underpinning the expected degree of success of the policies included in the 
scenario. 

The what-if type of applications we have presented share a common faith in the 
supposed success of planning concepts . Planning concepts were generally thought 
to be fully successful in the land-use simulations we have presented here. Restrictive 
policies were, for example, quantified as a substantial decrease in suitability of 
a location for restricted types of land-use, which in most cases prevented the 
allocation of these land-use types on those locations . The policy goal of intensifying 
the dwelling densities was also strictly enforced in the simulations of the Randstad 
project. From rare ex-post evaluations of spatial policies, it is known, however, that 
such spatial policies are never fully successful. Even the (relatively successful) 
longstanding Buffer Zones and Green Heart restricted development zones have 
only slowed down the pace of urbanisation - to about half that of surrounding, 
non-restricted areas (Koomen et al., 2008a). One should, thus, be very careful in 
communicating the value of the obtained simulation results. They show what could 
happen when a certain policy is put in place, but they cannot be used directly to 
show the likely degree of impact associated with a policy. When simulations are 
used in the latter type of impact assessments, their implementation in the model 
(e.g. as a restriction on urban development) should be based on empirical evidence 
of the degree of success of similar policies in the past. Alternatively, varying degrees 



148 



C. Jacobs et al. 



of the success of proposed planning concepts can be connected with differing views 
on the rigidity of planning regimes in different scenarios. 

If policy options are not fully restrictive or stimulatory, but rather influence the 
intensity of land use, their inclusion would benefit from enhancing the current 
model layout. Densification policies strive to increase the density of occupation, 
while other policies may strive to limit or reduce the intensity of land use. In 
the current version of Land Use Scanner, the impacts of such policies cannot be 
simulated directly. To be able to better assess, for example, the potential impact of 
varying degrees of prescribed dwelling densities and densification rates on land-use 
changes, as was done in the Randstad study, ideally the model should endogenously 
incorporate a notion of land-use intensity that varies with differing degrees of 
pressure on space. This would make it possible to assess the indirect influence of 
spatial restrictions on the intensity of urban land use in certain areas. The concept 
of land-use intensity is closely related to land prices. As the model uses a kind of 
price mechanism to allocate land, endogenous inclusion of land-use intensity seems 
a feasible option, albeit a complex one that calls for careful coordination with the 
external sector-specific models that currently deliver the demands for various types 
of land. 



References 

Atzema, O., Van Egmond, K., Mommaas, H., Wenting, R., & Kuijpers-Linde , M. (2008). Utrecht 
2040; Strategische notifies in het kader van het traject 'Samen op Weg naar 2040' van de 
Provincie Utrecht. Utrecht: Universiteit Utrecht Faculteit Geowetenschappen. 

Borsboom-van Beurden, J. A. M., Bakema, A., & Tijbosch, H. (2007). A land-use modelling 
system for environmental impact assessment; Recent applications of the LUMOS toolbox. 
Chapter 16. In E. Koomen, J. Stillwell, A. Bakema, & H.J. Scholten (Eds .), Modelling land-use 
change; Progress and applications (pp. 281-296). Dordrecht: Springer. 

Bouwman, A. A., Kuiper, R., & Tijbosch, H. (2006) Ruimtelijke beelden voor Zuid-Holland. 
Rapportnummer 500074002.2006. Bilthoven: Milieu- en Natuurplanbureau. 

Cramer, J. (2008). Nieuwe maten van de Randstad. Nova Terra [Januari 2008], 4-7. NIROV: The 
Hague. 

Oroen, J., Koomen, E., Ritsema, van E. J., & Piek, M. (2004) Scenario's in kaart; 

model- en ontwerpbenaderingen voor toekomstig ruimtegebruik. Rotterdam/Den Haag: NAi 

Uitgevers/Ruimtelijk Planbureau. 
Jacobs, C. G. W., Koomen, E., & Roggema, R. (2009). Towards an integrated vision of a climate 

proof Groningen (forthcoming). Groningen: Provincie Groningen. 
Koomen, E., Dekkers, J., & Van Dijk, T. (2008a). Open space preservation in the Netherlands: 

Planning, practice and prospects. Land Use Policy, 25(3), 361-377. 
Koomen, E., Kuijpers-Linde, M., & Loonen, W. (2008b). Ruimtelijke verkenning Overijssel 2040. 

Amsterdam: Geodan Next. 
Kuijpers-Linde, M., Koekoek, A., & Loonen, W. (2008). Uitwerking ruimtelijke beelden voor het 

nieuwe omgevingsbeleid van Drenthe. Amsterdam: Werknotitie . Geodan Next. 
MNP (2007). Nederland Later; Tweede Duurzaamheidsverkenning deel fysieke leefomgeving 

Nederland. MNP-publicatienr.500 12700 1/2007. Milieu- en Natuurplanbureau, Bilthoven. 
NIROV (2005). Nieuwe Kaart, Nieuwe Ruimte: Plannen voor Nederland in 2015. Den Haag: 

Nirov. 



8 Lessons Learned from Using Land-Use Simulation in Regional Planning 



149 



Oskamp, A., Poulus, C, & Van Til, R.-J. (2002). Spatial concentration and deconcentration of 

household types in the Amsterdam region; Effects of three scenarios of new construction. 

Journal of Housing and the Built Environment, 17, 321-335. 
Poulus, C, & Heida, H. (2005). Methodiek en toelichting Socratesmodel . Delft: ABF Research. 
Riedijk, A., Van Wilgenburg, R., Koomen, E., & Borsboom-van Beurden, J. (2007). Integrated 

scenarios of socio-economic change. SPINlab research memorandum SL-06. Amsterdam: Vrije 

Universiteit Amsterdam. 
Robinson, J. (1982). Energy backcasting: A proposed method of policy analysis. Energy Policy, 

70,233-337. 

Roggema, R. (2009). Hotspot klimaatbestendig Groningen: Eindrapportage ontwerp. Groningen: 
Provincie Groningen. 

Scotta, A., Pleizier, I. D., & Scholten, H. J. (2006, May 15-17) Tangible user interfaces in order 

to improve collaborative interactions and decision making. In E. Fendel & R. Rumor (Eds.), 

Proceedings of 25th Urban Data Management Symposium (UDMS 2006). Aalborg, Denmark: 

Delft, Urban Data Management Society. 
Snellen, D., Farjon, H., Kuiper, R., & Pieterse, N. (2006) . Monitor nota ruimte. De opgave in beeld. 

Den Haag/Bilthoven: Ruimtelijk Planbureau/Milieu- en Natuurplanbureau. 
Van der Hoeven, E., Jacobs, C. G. W., & Koomen, E. (2008). Beknopte beschrijving van 

sociaaleconomische scenario's voor het jaar 2100. Amsterdam: Vrije Universiteit. 
Van 't Klooster, S., Pauw, P., & Roggema, R. (2008). Backcasting analyse van een 

klimaatbestending Groningen. Groningen: Provincie Groningen. 
VROM (2001). Ruimte maken, ruimte delen. Vijfde nota over de ruimtelijke ordening 2000/2020. 

Den Haag: Ministerie van Volkshuisvesting Ruimtelijke Ordening en Milieubeheer. 
VROM (2008). Structuurvisie Randstad 2040. Naar een duurzame en concurrerende Europese 

topregio. Den Haag: Ministerie van Volkshuisvesting Ruimtelijke Ordening en Milieubeheer. 



Part III 

Future Developments 



Chapter 9 

Explaining Land-Use Transition 
in a Segmented Land Market 

Potential Input for Land Use Scanner 

Jasper Dekkers and Piet Rietveld 



9.1 Introduction 

Spatial planning policies can create segmented sub-markets for land, leading to an 
artificial scarcity for certain types of land use (usually urban land use) and higher 
land prices in that submarket, which can eventually spillover to other sub-markets. 
In a spatial context, these spillover effects are mainly observed in the urban fringes, 
where rural land owners often ask a higher price for their rural land because of 
expectations that this will be converted into urban land. The price for rural parcels 
in the urban fringe is thus related to the probability that these parcels will be put to 
urban uses in the (near) future, which is referred to as the 'transition probability'. 
Figure 9.1 describes this relation in more detail. Land prices decline as the distance 
from the city centre increases. In situations where there are no spatial planning 
constraints (Fig. 9.1a), there is no gap between land prices at the urban fringe. If 
there is a spatial planning policy being implemented that restricts building on the 
fringe of a city (Fig. 9.1b), a sudden change in land prices can be observed for rural 
parcels immediately bordering the city. When it is expected that these limitations 
along the fringe may be lifted, the market price for those rural parcels will be 
somewhere between the urban and the agricultural land price. If the market price 
is closer to the agricultural land price, there is a low transition probability, and if the 
market price is closer to the urban land price, there is a high transition probability. 

Various approaches exist for analysing land markets. One approach is to study 
land market processes in a descriptive way - this is commonly referred to as 
institutional analysis (e.g. Segeren, Needham & Groen, 2005; Segeren, 2007). 
Another, very different approach is to study land market outcomes in a quantitative 
way. In this chapter we use this second approach, since previous studies (e.g. 
Buurman, 2003; Dekkers, Rietveld, Van den Brink & Scholten, 2004) have already 
confirmed that quantitative models such as the Hedonic Pricing Method (Rosen, 
1974) can effectively incorporate multiple sub-markets at the same time. We do 



J. Dekkers (Kl) 

Department of Spatial Economics/SPIMab, VU University Amsterdam, De Boelelaan 1 105, 
1081 HV Amsterdam, The Netherlands 
e-mail: j .dekkers@vu.nl 

E. Koomen, J. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 153 
Practice, The GeoJournal Library 101, DOI 10.1007/978-94-007-1822-7_9, 
© Springer Science+Business Media B.V. 201 1 



154 



J . Dekkers and P. Rietveld 



(a) 



Land price 




Agricultural land use 



Distance from city centre 



(b) 



Land price 




Spatial planning constraint 



This transaction price suggests a 50% 
probability of transition from agricultural to urban land use 



Agricultural land use 



Distance from city centre 
► 



Fig. 9.1 Relation between land price, land use and distance from city centre: (a) with no spatial 
planning constraints; (b) with spatial planning constraints (Source: Dekkers, 2010) 



think, however, that an institutional analysis of the land market could be done to 
complement the approach we use here, since it can improve the understanding of 
the driving forces and motivations of (groups of) actors on various sub-markets and 
provide insights into the processes in these markets that lead to the market outcomes 
we are studying. 

In this chapter we present a novel analytical method for explaining how 
probable locations for future urban development can be identified from information 



9 Explaining Land-Use Transition in a Segmented Land Market 



155 



on actual land prices. In Section 9.2, we describe our approach for explaining 
the relation between transition probabilities and land prices using a single-step, 
multiple-equation approach. In Section 9.3, we apply this model to the land market 
in the Province of Noord-Holland and discuss the results. In Section 9.4, we use the 
results from the model to derive a transition probability map for the entire province, 
thus including land that has not been traded. There we also discuss how our method 
can be used in combination with land-use models, such as the Land Use Scanner 
model, to support spatial planning processes in The Netherlands. Finally, in Section 
9.5 we offer some concluding remarks. 

9.2 A New Explanatory Model for Land-Use Transitions 

Our new method for modelling land-use transitions integrates the strong points 
of two existing quantitative models for explaining land prices in The Netherlands 
(Buurman, 2003; Luijt, Kuhlman & Pilkes, 2003) into a linear probability model, 
thus combining Buurman's large set of explanatory variables and the probabilistic 
approach of Luijt et al. 

Buurman (2003) and Luijt et al. (2003) both use a single-equation logarithmic 
regression model for explaining land prices. Their models are revealed-preference 
models that use the Hedonic Pricing Method (HPM). The HPM determines the 
implicit value of non-tradable characteristics of goods by analysing the observed 
values of tradable goods that incorporate all or part of those non-tradable 
characteristics. Rosen (1974) was the first to publish a generally accepted article 
about this method. For a more in-depth summary of the hedonic pricing technique, 
we refer interested readers to Griliches (1971, Chapter 1) and Gordon (1990). 

The models of Buurman and Luijt et al. both use cadastral data on transactions 
for parcels of land outside urban areas. And both include numerous factors - based 
on extensive literature research - that are most likely to affect transaction prices 
of land. These factors can be divided into transaction characteristics (e.g. price, 
date of sale, type of buyer, type of seller), parcel characteristics (e.g. parcel size, 
soil quality) and spatial characteristics (e.g. accessibility measures, environmental 
measures, zoning designations). 

Both models have the drawback that not all characteristics are equally important 
for all buyers. For instance, project developers most likely do not value soil quality 
as much as farmers do. Therefore, Luijt et al. (2003) proposed an alternative for the 
single-equation model in the form of a two-step, multiple equation model (Fig. 9.2), 
using separate explanatory models for 'red'(ie. urban) and 'green' (i.e. agricultural) 
buyers. In order to determine which equation to use for a transaction, the transition 
probability must first be estimated using parcel and buyer characteristics. Luijt et al. 
(2003) discerned building constructors, municipalities and project developers as red 
buyers and farmers as green buyers. On the vertical axis of Fig. 9.2, price g i is the 
green transaction price and price „• is the red transaction price. The horizontal axis 
shows the transition probability. Logically, the higher the transition probability from 
green to red of a certain parcel, the higher the price of that parcel. 



156 



J . Dekkers and P. Rietveld 



Price/ha 




Fig. 9.2 Transition probability model of Luijt et al. (2003) using a two-step multiple equation 
(Source: Dekkers, 2010) 

A major drawback of using an artificial cut-off value at prob rl — 0.5 is the 
uncertainty for land transactions with probabilities around this value of being 
allocated to the correct equation, thus reducing the explanatory power of the model 
around this point. In the case study of Luijt et al. (2003), the results of the 
estimations show that the two functions certainly do not connect. 

Drawing on the two-step, probabilistic approach of Luijt et al. and the richness of 
the explanatory factors in Buurman's approach, we have developed a new modelling 
approach for: 

• estimating hedonic price models for the green and the red sub-markets separately, 
in order to do justice to the segmented nature of land markets; 

• computing transition probabilities from green to red in order to describe market 
expectations concerning observed transitions; 

• determining the contribution of explanatory variables to these transition 
probabilities; 

• using this model to predict transition probabilities for zones where no 
transactions have taken place. 

The analytical steps necessary to obtain the desired results with this approach are 
summarised in Fig. 9.3. 

A more detailed description of our approach now follows. Suppose a potential 
buyer foresees a transition probability from green to red of prob rl . Suppose, also, 
that we know the prices of the parcels of land concerned if they belong in the green 
market segment (price g i) and their prices if they belong in the red segment (price „•). 
How much is the buyer willing to offer? This model can be depicted graphically 
as in Fig. 9.4. Note that in this model the market price is equal to the green price 
when the transition probability equals 0. The market price equals the red price when 
the transition probability is 1 . Intermediate market prices are related to intermediate 
probabilities. 



9 Explaining Land-Use Transition in a Segmented Land Market 



157 



Step 1a Find subset of transactions for which it is clear on an a priori basis that the new use of 

the parcels concerned is red. These are called "red transactions". 
Step 1b Same for green transactions. This leads to a set of "green transactions". 
Step 1c The remaining transactions are called "uncertain transactions". 

Step 2a Estimate hedonic price model for subset of red transactions. 
Step 2b Same for green transactions. 

Step 3a Use the red price model estimated in Step 2a to predict the value that all parcels (red, 

green, uncertain) would have according to red market conditions. This leads to predicted 
red prices for all parcels. 

Step 3b Same for green price model. This leads to predicted green prices for all parcels. 

Step 4 Compute for all transactions the transition (from green to red) probability as probri as the 
ratio of the difference between the actual price minus the predicted green price divided 
by the difference between the predicted red price and the predicted green price. 

Step 5 Use a linear probability model to explain the values of the transition probability defined 
in Step 4 by means of a set of explanatory variables. 

Step 6 Compute the transition probability of all zones in the region concerned, also including 
the zones where no transactions took place. 



Fig. 9.3 Summary of steps in linear probability approach (Source: Dekkers, 2010) 



Price/ha 



Price ri 



Price n 




0 1 

Transition probability (prob ri ) 

Fig. 9.4 Single-equation transition probability model (Source: Dekkers, 2010) 



Step la - Find red subset of transactions 

In order to apply this method we need estimates of red and green prices. To do 
this, we can select a subsample of transactions for which we know that the parcels 
concerned will almost certainly be developed, i.e. prob,i — 1 . In this case, we choose 
to take a sample of 10% of the transactions with the highest price per square metre. 

Step lb - Find green subset of transactions 

Similarly to Step la, consider a sample of transactions where, according to expert 
judgement, prob rl is very close to zero: for example, a parcel lying within the 



158 



J . Dekkers and P. Rietveld 



National Ecological Network and for which, in that area, another policy applies 
that restricts urbanisation. As with the red subsample, we choose to take a sample 
of 10% of the transactions with the lowest price per square metre. 

Step lc - Remaining transactions 

The remaining 80% of the transactions are only used from Step 3 onwards. 

Steps 2a & 2b - Estimate hedonic model for red and green transactions 

Since for the red sample selected in Step la we assume that prob rl — 1, this 
means we can estimate for the red transactions: 

price, — a r + fJ r ■ x r ; + e,- (9.1) 

where a r is a constant and is a vector of coefficients for the related red matrix of 
independent variables (x r f). This leads to estimates <i,-and j3 r . 

In the case of the green transactions, we assume that prob r \ — 0, so we can again 
simplify the basic equation and estimate: 

price i — OLg + /3 g ■ x g i + £; (9.2) 

where a ? is a constant and /3 ? is a vector of coefficients for the related green matrix 
of independent variables (x g i). This leads to estimates a^and fi g . 

Steps 3a & 3b - Compute red and green transaction prices 

Once we know a, and /3 r and a g and fi g , we can compute red and green prices for 
all transactions: 

price r \ — a r + ft,- ■ x,i and price g i — a g + j3 g ■ x g i (9.3) 

In this way, we can obtain predictions of land prices for all parcels concerned, 
albeit conditional on whether they fall under the red or the green regime. 

Step 4 - Compute transition probability 

We can now compute the transition probability for all transactions based on the 
model represented in Fig. 9.4: 

price; — (1 — prob r i) ■ price g \ + prob r i ■ price rii (9.4) 

After rewriting this formula, we can compute prob r i using price-, and the 
predicted red and green land prices price r j and price gl : 

prob ri = P^-P^ (9 .5) 
pricey — price r i 



9 Explaining Land-Use Transition in a Segmented Land Market 



159 



Step 5 - Explain transition probability 

Next, we can estimate the factors that have an impact on prob,i . Let these factors 
be denoted as z, (mainly overlapping with Xj, only the surface and time dummy 
variables are excluded). We have to take into account that there will probably be 
observations for which the market price lies either below the estimated green price 
or above the estimated red price. Plausible causes for prices dropping below the 
green price could be, for instance, unobserved features such as soil contamination. 
Unobserved features could also explain why an observed price is higher than the 
red, urban price. 

A linear probability model is used to explain variations in the transition 
probability prob rl : 

prob ri — y + & ■ Zi + Pi (9.6) 
where y and <5 are parameters to be estimated, and is an error term. 

Step 6 - Compute transition probability for all zones 

In this final step we can use the coefficients of the linear probability model in 
combination with the underlying spatial data sets of the explanatory variables to 
calculate the transition probability for all zones in the region concerned. This can be 
done by rasterising all data sets on, for instance, a 25 m grid. 

9.3 A Case-Study in the Province of Noord-Holland 

9.3.1 Description of the Study Area 

The land market in Noord-Holland (Fig. 9.5) consists of many sub-markets. There 
are many agricultural areas for which various nature protection policies and/or 
nature development plans apply. Also, a great deal of land is used for horticulture 
and flower bulb cultivation, which is relatively expensive. Then there is the national 
airport, Schiphol Amsterdam airport, which is assumed to have a large impact on 
land use and prices in its surrounding areas. In our opinion, the very heterogeneity 
of land uses and prices was an interesting setting in which to apply our model (as 
described in Section 9.2). 

The surface area of the Province of Noord-Holland covers 4,059 km 2 , of which 
2,657 km 2 is dry land. This equals 7.8% of the land surface of the Netherlands. 
In 2000, 2.5 million people were living in Noord-Holland, making it the province 
with the second highest number of inhabitants in the Netherlands. Approximately 
19% of the national Gross Domestic Product (GDP) is generated in Noord-Holland 
(CBS, 2000). Table 9.1 shows that the 'Commercial services' sector is relatively 
important in Noord-Holland and that the 'Manufacturing' and Agriculture, forestry 
and fisheries' sectors are relatively less important. As in the rest of the Netherlands, 
agricultural land use is decreasing in Noord-Holland, while urban and natural land 
use are increasing. Compared to the Netherlands as a whole, Noord-Holland has 
less forest and nature and more built-up areas. 



160 



J . Dekkers and P. Rietveld 




\/\ Randstad area 
Noord-Holland 
Other Provinces 



Fig. 9.5 Location of the Province of Noord-Holland within The Netherlands (Source: Dekkers, 
2010) 



Table 9.1 Relative sectoral contributions to the economies of Noord-Holland and The Netherlands 
(Source: CBS, 2000) 





Gross added value 1999 (%) 




Sector 


Noord-Holland 


The Netherlands 


Agriculture, forestry and fisheries 


1.6 


2.8 


Industry 


17.0 


25.5 


Commercial services 


59.3 


49.0 


Government and healthcare 


22.2 


22.8 


Total 


100.0 


100.0 



9.3.2 Description of the Land Transaction Data 

For our case-study, we used the cadastral InfoGroMa database of DLG (Government 
Service for Land and Water Management, part of the Ministry of Agriculture, 
Nature and Food Quality). This data set is a subset of the Dutch cadastral database 
containing all transactions for parcels outside urban areas in the Netherlands. 



9 Explaining Land-Use Transition in a Segmented Land Market 



161 



When we disregarded transactions within families, the database registered for 
Noord-Holland 2,685 parcels in 1,625 transactions for the period 1998-2002. 
We also disregard transactions containing immobile property, as we do not have 
information on the characteristics of the immobile property and the influence of 
these characteristics on the transaction price. 

Table 9.2 gives some descriptive statistics for both the red and green subsamples 
(Steps la and lb). We see large differences in the transaction price per square metre 



Table 9.2 Summary statistics for the red and the green subsamples related to Step la and lb 

R (red) 



Variable 


G (green) Min. 


Max. 


Mean 


Std. dev. 


Transaction characteristics 












Transaction price ( x 1 ,000 euros) 


R 


15.24 


29,743 


857.31 


2,583 






1 53 


267.69 


30.84 


38.45 


Transaction price per square metre 


R 


5.15 


153.15 


19.83 


20.14 


(euros/m 2 ) 


G 


0.16 


1.07 


0.55 


0.19 


Leased land (0/ 1 ) 


R 


0.00 


0.00 


0.00 


0.00 




G 


0.00 


1.00 


0.54 


0.50 


Structural characteristics 












Surface area (x 1 ,000 m 2 ) 


R 


2.50 


592.34 


45.01 


80.94 




G 


2.51 


569.20 


56.25 


73.82 


Spatial characteristics 












Share of a transaction that is located in 


R 


0.00 


1.00 


0.33 


0.47 


an urban development zone (New Map 


G 


0.00 


1.00 


0.01 


0.08 


of The Netherlands) 












Distance to the Randstad (km) 


R 


0.00 


77.29 


15.00 


17.56 




G 


0.00 


80.09 


23.54 


19.99 


Distance to the nearest built-up area (km) 


R 


0.00 


1 .81 


0.29 


0.28 




G 


0.00 


2.23 


0.57 


0.42 


Distance to main road < 200 m (0/ 1 ) 


R 


0.00 


1.00 


0.13 


0.34 




G 


0.00 


1.00 


0.07 


0.26 


Share of a transaction's land use that is 


R 


0.00 


1.00 


0.06 


0.24 


urban green 


G 


0.00 


1.00 


0.01 


0.11 


Share of a transaction's land use that is 


R 


0.00 


1.00 


0.05 


0.21 


greenhouse horticulture 


G 


0.00 


0.00 


0.00 


0.00 


Land development project (BBL/DLG) 


R 


0.00 


1.00 


0.23 


0.42 


(0/1) 


G 


0.00 


1.00 


0.40 


0.49 


Share of a transaction that is located in 


R 


0.00 


1.00 


0.25 


0.43 


the provincial ecological main structure 


G 


0.00 


1.00 


0.65 


0.47 


(PEHS) 












Share of a transaction that is located in a 


R 


0.00 


1.00 


0.10 


0.30 


Buffer Zone 


G 


0.00 


1.00 


0.20 


0.40 


Share of a transaction that is located in a 


R 


0.00 


1.00 


0.07 


0.26 


belvedere policy zone (UNESCO world 


G 


0.00 


1.00 


0.08 


0.26 


heritage area) 












Share of a transaction's land use that is 


R 


0.00 


1.00 


0.01 


0.11 


nature 


G 


0.00 


1.00 


0.06 


0.24 


Note: Transactions in Noord-Holland (period 1998-2002; N = 


1,625; AW 


= 163; 7V g 


reen — 162) 



162 



J . Dekkers and P. Rietveld 



between the two subsamples. The mean price per square metre of the red subsample 
is about 36 times higher than that of the green subsample. This underlines the 
strong segmentation between the green and the red land markets in the Netherlands. 
No leased land was found in the red subsample and, on average, transactions in 
the red subsample lay closer to the Randstad and to urban areas, often within 
200 m of a main road. Further, in the red subsample, on average a larger share 
of the transactions lay within and were subject to urban development plans or 
were classified as 'urban green' or 'greenhouse horticulture' land-use types. 
Similar to the distinction between land uses as made in Koomen, Dekkers, and 
Van Dijk (2008), the seemingly contradictory urban green category contains sparse 
vegetation (bare soil or grassland) in or nearby urban areas. It contains land with a 
predominantly green appearance and a functional relation to the neighbouring urban 
area, such as sports-fields, parks and land set aside for construction. Greenhouses 
are distinguished separately from other agriculture because of their distinct urban 
appearance. 

In contrast, in the green subsample, on average a larger share of the transactions 
lay within the National Ecological Network. The same applies for Buffer Zones 
(green corridors no less than 4 km wide that must be left open between major 
agglomerations; RNP, 1958) and for the land-use type 'nature'. For Belvedere 
policy zones (UNESCO world heritage areas; OCW, LNV, VROM & V&W, 1999), 
there is virtually no difference between the two subsamples. Figure 9.6 shows 
four maps containing the locations of the sold parcels for the two subsamples, 
family transactions, and the remaining parcels; recall that we disregarded family 
transactions in our analysis. Amsterdam is located at the bottom of the maps (the 
harbour area is just visible). What is clearly visible is that the red parcels (Fig. 9.6b) 
are in general concentrated near urban areas, while the green parcels (Fig. 9.6c) are 
generally further away from urban areas. 



9.3 3 Selecting Model Variables 

The variables we used overlap to some extent with the variables we used in our 
explanatory Hedonic Pricing model for the land market in the Province of Noord- 
Holland, which is described in Dekkers et al. (2004). That model is the same as 
the one Buurman (2003) used. The difference is that our current analysis does not 
exclusively focus on the explanation of rural land prices but rather on rural-urban 
transition probability. For that reason, we included factors that we expected would 
have an influence on this transition probability. First, we included transport noise as 
a factor in the red explanatory Hedonic Pricing model by defining a dummy variable 
that had the value 1 when a parcel is located less than 200 m from a main road. 
We also included information on land use: built-up, urban green and greenhouse 
horticulture in the red model; and nature in the green model. Finally, in the green 
explanatory model we took into account whether or not parcels lay in a Buffer Zone 
or a Belvedere/UNESCO Zone. We expected these policies to negatively influence 



9 Explaining Land-Use Transition in a Segmented Land Market 
(a) 



163 




***. i* • V 




Fig. 9.6 Locations of all geocoded rural land transactions without immobile property in a part of 
the Province of Noord-Holland in the period 1998-2002 and land use as in 2000 (Source: Dekkers, 
2010) 



164 



J . Dekkers and P. Rietveld 



land prices because they restrict land use to some extent. After testing various 
model and variable specifications, we chose a semi-logarithmic model specification 
since this is the most widely used form. We also tested other variables related to, 
for instance, soil quality, accessibility and spatial policies aimed at nature and/or 
recreation development, but these factors were not significant. 

9.4 Results of the Model Parameter Estimations 

The estimation results (Table 9.3) give us confidence in both the red and the green 
models: the coefficients show the expected signs and most of them are significant. 
First, in the red model we see that parcels that were closer to urban areas had a 
higher price. Next, parcels that were less than 200 m from a main road had a lower 
price than parcels located further away, the most probable reason being the nuisance 
experienced from transport noise in this area. Further, when the land use was urban 
green or greenhouse horticulture, this positively influenced parcel prices. 



Table 9.3 Results for the subsample estimates in the formulas related to Step 2a and 2b 

Red (N= 162) Green (N = 1 62) 



Variable (LN(price-deflated) = dependent variable) Coeff. Sign. Coeff. Sign. 



Constant 


2.426 


j 289 *** 


LN(surface) 


0.972 


1.021 *** 


Share of a transaction that is located in an urban 


0.189 




development zone (New Map of The Netherlands) 






Distance to the Randstad (km) 


-0.002 




Distance to the nearest built-up area (km) 


-0.255 




Distance to main road < 200 m 


-0.256 




Share of a transaction's land use that is urban green 


0.397 




Share of a transaction's land use that is greenhouse 


-0.085 




horticulture 






Leased land 




-0.116 


Land development project (BBL/DLG) 




-0.045 


Share of a transaction that is located in the provincial 




-0.078 


ecological main structure (PEHS) 






Share of a transaction that is located in a buffer zone 




-0.125 


Share of a transaction that is located in a Belvedere 




-0.028 


policy zone (UNESCO world heritage area) 






Share of a transaction's land use that is nature 




-0.097 


s.e. of regression coefficient 


0.589 


0.303 


R 2 


0.797 


0.928 


Adjusted R 2 


0.788 


0.924 



Note: *** = significant at 0.01; ** = significant at 0.05; * = significant at 0.10 



9 Explaining Land-Use Transition in a Segmented Land Market 



165 



In the green model, we see that leased land was cheaper, as was land that was 
in a land development project area of DLG or in the provincial concretisation of 
the National Ecological Network. Two major reasons why DLG buys land are its 
use in re-allotment projects and to create new nature areas. In both cases, the land 
use remains green: agriculture or nature, which certainly does not have a positive 
impact on prices. For parcels in Buffer Zones, we do indeed see a negative influence 
on the land price, confirming our expectation. The Buffer Zone policy has a strong 
protective value (Koomen et al., 2008; Van Rij, Dekkers & Koomen, 2008). When 
Buffer Zones were established in the 1960s, the original intent was that agriculture 
would be a driving force in these areas, in combination with nature and recreational 
developments. Over the years, the national government has changed its strategy 
more and more toward actively acquiring land for nature development. Apart from 
that, Buffer Zones are, in general, not very large areas, and they are to be found in 
areas between major urban agglomerations where urbanisation pressures are already 
high. For all these reasons, scale increases for agricultural businesses are hard to 
realise, so that these areas have less potential for modern farming than elsewhere. 
The same kind of reasoning applies to the Belvedere/UNESCO areas: this policy 
also has a highly protective value, as it is difficult to expand farming activities 
here and multiple restrictions apply. The model coefficient of the Belvedere policy 
zone confirmed our expectations as it had a negative sign, but this variable is not 
significant. 

We subsequently used the coefficients of the red and the green model to 
estimate red and green transaction prices and to compute the rural-urban transition 
probability. Figure 9.7a and b show the estimated prices (x-axis) plotted against 



70 -, 
60 - 
50 - 
40 - 
30 - 
20 - 
10 - 



Real Price/m 2 (i) 



J, 



0,8 -, 



0,6 - 



0,4 ■ 



0,2 



Real Price/m 2 (i) 



#5. 



0 10 20 30 40 

Estimated Price/m 2 (i) 
Red sample (162 obs.) 



I 

50 



0,0 -f- 
0,0 



0,2 0,4 0,6 
Estimated Price/m 2 (i) 
Green sample (162 obs.) 



0,8 



Fig. 9.7 Plot of real versus estimated transaction price per m 2 for the red and green sub-samples, 
respectively (some outliers are not shown) (Source: Dekkers, 2010) 



166 



J . Dekkers and P. Rietveld 



Number of observations % 




| Frequency 
Cumulative 



0 2 4 6 8 10 12 14 16 



Probability of becoming developed (red) 

Fig. 9.8 Histogram of calculated rural-urban transition probabilities of the total sample (Source: 
Dekkers, 2010) 




Fig. 9.9 Map of the computed transition probabilities for all 1 ,625 observations (Source: Dekkers, 
2010) 



9 Explaining Land-Use Transition in a Segmented Land Market 



167 



the transaction prices (y-axis) for red and green transactions, respectively. Both 
models somewhat underestimated land prices and in particular the red model does 
not capture the full variation in land prices. 

As Fig. 9.8 shows, more than 90% of the calculated transition probabilities lie 
between 0 and 1, meaning that the Ordinary Least Squares (OLS) estimation does 
not have to correct for a large number of probabilities outside this range: there 
are no probabilities below 0 and only 8.8% of the observation's probabilities are 
above 1. Theoretically, probabilities cannot have values outside the range [0:1]. 
In practice, however, these values can occur due to unobserved features in the 
model (see also Step 5 in Section 9.2). The mean probability is 0.40 and the range 
is [0.01:15.05]. As can be seen, there are some observations with a very high 
transition probability (see also Fig. 9.9). The explanation is that in the areas where 
these transactions have taken place, a lot of urban development has occurred, with 
numerous 'red' factors/variables simultaneously having a positive influence on the 
transition probability. 

The results of the OLS analysis shown in Table 9.4, with the rural-urban 
transition probability as the dependent variable, indicate that the coefficients of 
the 'red variables' explain these variables well: when a parcel lies in an area with 



Table 9.4 Results for the ordinary least squares estimation on all 1 ,625 observations according to 
Eq.(9.6) 



Variable 


OLS results 
Coeff. 


Sign. 


Constant 


0.773 




Urban development plans (New Map of The Netherlands) 


0.724 




Distance to the Randstad (km) 


-0.008 




Distance to the nearest built-up area (km) 


-0.104 


** 


Distance to main road < 200 m 


0.044 




Share of a transaction's land use that is urban green 


0.051 




Share of a transaction's land use that is greenhouse horticulture 


0.949 


*** 


Leased land 


-0.257 


*** 


Land development project (BBL/DLG) 


-0.137 


*** 


Share of a transaction that is located in the provincial ecological main 


-0.093 




structure (PEHS) 






Share of a transaction that is located in a buffer zone 


-0.226 




Share of a transaction that is located in a Belvedere policy zone 


-0.262 


*** 


(UNESCO world heritage area) 






Share of a transaction's land use that is nature 


-0.125 




s.e. of regression coefficient 


0.766 




R 2 


0.142 




Adjusted R 2 


0.135 





Note: *** = significant at 0.01; ** = significant at 0.05; * = significant at 0.10 



168 



J . Dekkers and P. Rietveld 



urban development plans, the probability of a green parcel becoming urbanised 
(prob,j) is positively influenced; the further away from the Randstad, the lower 
prob,j is. The same goes for the distance to the nearest built-up area, although 
this effect is not significant. Further, when a transaction takes place in an area that 
is predominantly built-up, urban green or greenhouse horticulture, the transition 
probability is influenced positively. The 'green variables' tell us that the probability 
of a transition to red land use is lower when land is leased and/or when parcels are in 
an area where DLG has land development plans. Also, parcels lying in the provincial 
ecological main structure have a lower probability of becoming developed for urban 
land use than parcels not lying in this protective nature policy zone. Next, the 
influence of a buffer zone on prob rl is negative. Finally, being located in a Belvedere 
policy zone also influences the probability negatively. 

9.5 Application of Model Results 

9.5.1 Calculation of a Transition Probability Map 

The model results can be used to calculate a transition probability map for all 
parcels - sold and not sold - in the entire Province of Noord-Holland (see Step 6 in 
Section 9.2). The map legend in Fig. 9.10 shows that the range of probability values 
is concentrated around the range [0:1]: from -0.46 to a maximum of 2.5, to be more 
precise, with the mean value at 0.53. There is no substantial difference between 
the mean values of this transition probability map and the computed transition 
probabilities for the observed transactions (0.40). An explanation for the probability 
values less than zero could be the presence of unobserved variables and, for instance, 
the fact that we are not able to include the variable 'leased land' in this final step as 
we do not know for the entire province which parcels are leased and those which are 
not. Another possible explanation is that our model does not account for the large 
impact of uncertainty resulting from land-use change over time: a buyer buys a 
parcel now and then has to wait for, for example, 10 years before a decision is taken 
concerning whether or not the land use of that parcel is allowed to change from rural 
to urban. So, in addition to uncertainty about whether the change will occur, the fact 
that the buyer determines his offer using a discounted rate also lowers his bid price 
and, thus, the transition probability values in our model. 

To gain an insight into how the factors uncertainty and time play a role in our 
current model, consider the following: suppose a buyer has to decide whether or 
not to buy a parcel now, while knowing that after 10 years a decision will be 
taken whether or not the rural parcel may be developed. In our model, we assume 
for the sake of convenience that the developed or undeveloped parcel prices are 
offered at the moment just before this decision is taken. Suppose these prices are 
equal to the average transaction prices per hectare for the two subsamples: P re d = 
198,300 euros/ha and P green = 5,500 euros/ha. Now assume that the buyer estimates 



9 Explaining Land-Use Transition in a Segmented Land Market 



169 



0 



Transition probability 
| -O.46-O.25 

□ O.26-0.55 
I I 0.56-0.94 

□ 0.95-1.3 
■ 1-4-1-9 

I I Water 
] Built-up 
I Infrastructure 



Fig. 9.10 Rural-urban transition probabilities in the Province of Noord-Holland (Source: Dekkers , 
2010) 



that he will have a 50% chance of being allowed to develop the parcel. Taking 
only this uncertainty into account, his bid price will be 0.50 x (P re d+ Pgreen) = 
101,900 euros/ha. Suppose we now modify the model to allow for the decision 
whether or not he is allowed to develop the parcel is taken not now but rather in, 
say, 10 years time. In this case the buyer will incorporate an annual interest rate of, 
for example, 7% on his investment, to compute his bid price: 

385 385 385 

Pgreen = 385 H 1 =- + ... H -= 5,500 (9.7) 

green 1 +0.07 (1 +0.07 2 ) (l+0.07) n V ; 

where 385 is the annual revenues in euros per hectare (0.07 x 5,500) related to the 
use of this land for green purposes . 



170 



J . Dekkers and P. Rietveld 



385 385 385 

P red - 385 H 1 =r + ... H „ + 

1 +0.07 (1 +0.07) 2 (1 +0.07) 9 

13,881 13,881 13,881 

103,500 



(9.8) 



(1 +0.07) 10 (1 +0.07) 11 (1 +0.07) 11 



where 385 is the annual revenues in euros per hectare for the first 10 years in which 
the parcel is still green, and 13,881 are the annual revenues in euros per hectare from 
year ten onwards, when the parcel will be used for 'red' purposes. This situation 
leads to a bid price of 0.50 x (P re d + Pgreen) = 54,500 euros/ha. The result is, thus, 
a considerably lower bid price than if this delay in development is not taken into 
account. If this lower bid price were to be used as input in our model it would lead 
to a transition probability that is considerably lower than 0.50. Thus the transition 
probabilities we measure reflect a combination of possible delay of the transition 
and the probability of such a transition in a narrow sense. 



9.5.2 Relevance of Results for Land Use Scanner 

The understanding of how the land market works and the occurrence of current 
land use, especially in the urban-rural transition zone, can help improve our ability 
to model future land-use change. The quantification of land-use change with the 
use of land-use models is very important for evaluating the effects of spatial policy 
(MNP, 2004; Borsboom-van Beurden et al., 2005). There are plenty of land-use 
models available for simulating land-use change. Most of them only simulate urban 
or rural land-use. However, because of an increasing overlap of policies in the 
Netherlands and the importance of rural-urban transitions in land-use change, it is 
desirable to have models that are suitable for integrated scenario analyses. For the 
Netherlands, several of these integrated models are available, one of them being 
Land Use Scanner, which is described at length in this book. This economics- 
oriented probabilistic model uses a logit-function in an iterative process to simulate 
demand for, and supply of, land (Hilferink & Rietveld, 1999). The way current and 
future developments, spatial policies and spatial pressure on land are modelled is 
subject to scientific debate. Over the past 2 years, ample attention has been paid 
to calibration and validation issues in the form of autologistic regression analyses, 
for example in the study by Loonen and Koomen (2009), which focuses on the 
probability of land-use change based on proximity to other land-use types. Further 
improvements can be made with regard to the sensitivity analysis of the results 
to the scale or resolution of modelling. A more fundamental improvement can be 
the restructuring, or enhancement, of the theoretical-economics foundations of the 
model, i.e. using information on land prices and land market processes in a more 
direct way in the model. Because of the big difference between processes in the 
land market and factors that affect land-use change, strengthening this theoretical- 
economics link is quite a challenge. However, we think that economics-based 
land-market models can provide the necessary theoretical foundations for these 
models. Consider the bid-rent theory and how, based on Alonso's urban land-market 



9 Explaining Land-Use Transition in a Segmented Land Market 



171 



model (Alonso, 1964), McFadden (1978) developed an empirically more practical 
stochastic maximisation model that lies at the heart of Land Use Scanner. 

Based upon our work, we propose three improvements to the way Land Use 
Scanner is currently used: 

1. Land Use Scanner's algorithm computes probabilities for all land-use types 
in all cells separately and compares them as a kind of utility, where the 
probabilities/utilities are static. By using the results of multinomial logistic 
regression analysis (documented in Loonen & Koomen, 2009), we can improve 
the relative scaling of the different suitability factors for each land use's 
suitability map. For now, we have used only a limited set of factors in this 
analysis, which points to additional factors that help explain the transition to 
urban types of land use. This means that the results of the analysis can be used 
to select what other factors to include in the suitability maps for the multinomial 
logistic regression analysis. For example, in addition to location factors that 
are already included, such as the National Ecological Network and various 
accessibility measures, other factors that could be included are: distance to the 
Randstad, the presence of the land-use types 'urban green' and 'greenhouse 
horticulture', land development projects of DLG, Belvedere/UNESCO zones 
and/or Buffer Zones. 

2. Since our analysis is based on actual land transactions in real euros, the results 
can also be used to rescale the suitability values for the different land uses in 
Land Use Scanner in real euros/m 2 . Currently, the range of suitability values for 
the various types of land use are still based on expert judgement and vary from 
5 euro/m 2 for nature up to 35 euro/m 2 for residential land. Further, being based 
on real euros also means that our transition probability map can be used as an 
initial bid-price map for urban land use (i.e. undeveloped land). 

3. Finally, on a more fundamental level, we can strengthen the theoretical- 
economics link by examining processes in the land market using interview 
techniques and descriptive analysis (in line with the complementary role of 
institutional analysis mentioned in the introduction) to include information that 
can be derived from these analyses of bid prices for various land uses. For 
instance, we can examine how profits per hectare for different agricultural uses 
are translated by actors into actual bid prices in euros/m 2 and what differences 
there are between the average bid prices for different land uses so as to improve 
their relative scaling. An advantage of this approach is that it can also be applied 
to land uses for which no historical data (for the multinomial logistic regression 
analysis) are available (e.g. for biofuel production). This line of research is 
clearly linked to the work described by Kuhlman and others in this book. 

The effect of directly including information from our analyses can be examined 
by defining a model configuration with only two types of land use: urban or 
rural. We further include the spatial data layers that are used in our 'red' and 
'green' hedonic pricing models in Land Use Scanner. We then directly insert our 
HPM model coefficients of the red and the green models in the red and the green 



172 



J . Dekkers and P. Rietveld 



land-use suitability maps, respectively. If this approach is used, these suitability 
maps represent the initial bid of land-use type j for a parcel in cell c, thus improving 
the comparison of suitabilities, both between land-use types and for cells used by a 
single land-use type (Buurman, 2003). An additional advantage is that the suitability 
values can be directly related to market transactions in terms of euros per square 
metre, whereas the values currently used, although expressed in euros per square 
metre, are not based on such transaction data. Of course, in this example we only 
estimate one red and one green hedonic pricing model. With 10 modelled land-use 
types, this would mean constructing 10 hedonic pricing models. 

On the basis of the above analysis, we conclude that there is scope to implement 
this approach in the current version of Land Use Scanner. This could be based 
on market transaction data, initially for, for example, only two aggregate land-use 
types. This would allow comparison of the original probabilistic algorithm of 
Land Use Scanner with the novel probabilistic approach we propose here. More 
specifically, it would allow us to examine what is the best way to transfer the 
knowledge derived from our approach into Land Use Scanner's allocation model: 
through replacing the model algorithm with our probabilistic approach; through 
using the probability maps derived as initial bid-price maps; or through a translation 
of the relative scaling of suitability maps into real euros. Finally, working with more 
realistic monetary values in Land Use Scanner would potentially shed more light 
on the derivation and actual values of the shadow prices that are generated in the 
iterative modelling process . 

9.6 Concluding Remarks 

In this chapter we have described our efforts to model the possible occurrence of 
spillover effects between segmented markets in the urban fringe. These sub-markets 
are created by spatial planning policies that restrict urban land development. 

Earlier attempts to model the Dutch land market using hedonic pricing techniques 
only tried to explain what (spatial and non-spatial) factors contributed to land 
prices and/or tried to explain price differences between rural parcels with either 
an agricultural or urban designation. As described, we have taken these approaches 
one step further by developing a model that uses the hedonic pricing method in 
combination with a linear probability model. The resulting model uses the factors 
that explain differences in land prices to analyse the probability of rural land 
becoming urban land. 

The results show that our approach satisfactorily answers a number of 
technical-methodological issues that occur in the models that originally gave us the 
idea for our approach: by not directly including different buyers in the analysis, 
we avoid the problem of assuming that all characteristics are equally important for 
all buyers. Furthermore, although the level of explained variance is not that high, 
when we try to explain what factors influence the probability of a rural parcel 
becoming urbanised, the factors in the Ordinary Least Squares estimation have 
plausible signs, are in general significant and explain their respective influence on 
rural-urban transition probabilities well. 



9 Explaining Land-Use Transition in a Segmented Land Market 



173 



Several recommendations for improving our model can be made. First, we 
expect that its explanatory power can be improved by, for instance, adding more 
explanatory variables, such as distance to railway stations, soil quality data and the 
share of a transaction that is located within areas defined by the European Bird and 
Habitat Directive. Second, the New Map of the Netherlands variables that describe 
where land-use change will occur should be replaced with variables that include 
information on the driving forces behind the land-use changes that occur. This 
could be done through literature study (e.g. De Nijs, De Niet & Crommentuijn, 
2004; Verburg et al., 2004). Third, and perhaps most importantly, a new version 
of the model should include the timing of a transition, that is, how long does a 
buyer probably have to wait before the decision is taken whether or not the expected 
transition will take place? Procedures for changing the designation of a parcel can 
take many years, depending on, among other things, whether or not stakeholders 
submit formal protests against such a change. Explicitly including the factor time in 
the model is empirically relevant. 

From an economics perspective, our analysis points to an important consequence 
of current spatial policy: the price gap at the urban fringe indicates that rural 
land just outside the urban fringe (observed from within a city) is used very 
inefficiently - alternative uses (i.e. urban uses) would yield higher returns on 
investment. Figure 9.11, in which the red and the green prices per m 2 of a parcel 
at the urban fringe are shown to differ by a factor of almost 24, illustrates just how 
large this price gap actually is . 



Land price 




0.33 €/m 2 



Spatial planning constraint 



The computed red and green land price differ 
by a factor of about 24 at the urban fringe 



Urban land use Agricultural land use 



Distance from city centre 



Fig. 9.11 Relation between red and green land prices at the urban fringe in the case of a spatial 
planning constraint (Source: Dekkers, 2010) 



174 



J . Dekkers and P. Rietveld 



Acknowledgements We would like to thank the BSIK programmes 'Ruimte voor 
Geo-lnformatie' (www.rgi.nl) and Habiforum (www.habiforum.nl) for partially funding this 
research. We extend, furthermore, our thanks to the Government Service for Land and Water 
Management of the Netherlands (DLG), a department of the Netherlands Ministry for Agriculture, 
Nature and Food Quality (LNV), for supplying the land transaction data. Thanks are also due 
to Jan Luijt and Tom Kuhlman of the Agricultural Economics Research Institute (LEI) of 
Wageningen University and Research Centre for discussions about land sub-markets, modelling 
and data issues. And finally, we thank Gerard Kooman, Edwin Bleijinga, Rik Heskes and Ton van 
Bart of the Province of Noord-Holland for discussions about the land market in their province. 



References 

Alonso, W. A. (1964). Location and land use: Toward a general theory of land rent. Cambridge: 

Harvard University Press. 
Borsboom-van Beurden, J. A. M., Boersma, W. T, Bouwman, A. A., Crommentuijn, L. E. M., 

Dekkers, J. E. C, & Koomen, E. (2005). Rtiimtelijke Beelden - Visualisatie van een veranderd 

Nederland in 2030. Bilthoven: Milieu- en Natuurplanbureau. 
Buurman, J. J. G. (2003). Rural land markets: A spatial explanatory model. PhD Dissertation, 

Vrije Universiteit, Amsterdam. 
CBS (2000). Statline, Centraal Bureau voor de Statistiek. www.statline.nl. Retrieved 15 June 2005. 
Dekkers, J. E. C. (2010). Externalities, land use planning and urban expansion. PhD Dissertation, 

Vrije Universiteit, Amsterdam. 
Dekkers, J. E. C, Rietveld, P., Van den Brink, A., & Scholten, H. J. (2004). Exploring the land 

market in the province of Noord-Holland using a spatial regression model. Paper for the 44th 

ERSA-congress, Porto, Portugal, August 25-29. 
De Nijs, T. C. M., De Niet, R., & Crommentuijn, L. E. M. (2004). Constructing land-use maps of 

The Netherlands in 2030. Journal of Environmental Management, 72(1), 35^42. 
Gordon, R. J. (1990). The measurement of durable goods prices. Chicago, IL: University of 

Chicago Press. 

Griliches,Z. (Ed.). (1971). Price indexes and quality change. Cambridge, MA: Harvard University 
Press. 

Hilferink, M., & Rietveld, P. (1999). Land use scanner: An integrated GIS based model for long 
term projections of land use in urban and rural areas. Journal of Geographical Systems, 1(2), 
155-177. 

Koomen, E., Dekkers, J. E. C.,& Van Dijk,T. (2008). Open space preservation in The Netherlands: 

Planning, practice and prospects. Land Use Policy, 25(3), 361-377. 
Loonen, W., & Koomen, E. (2009). Calibration and validation of the land use scanner allocation 

algorithms. PBL Report No. 550026002. Bilthoven: PBL. 
Luijt, J., Kuhlman, J. W., & Pilkes, J. (2003). Agrarische grondprijzen onder stedelijke 

druk - Stedelijke optiewaarde en agrarische gebruikswaarde afhankelijk van ligging. NPB 

Werkdocument 2003/15. Den Haag: LEI. 
McFadden, D. (1978). Modelling the choice of residential location. In Spatial Interaction Theory 

and Planning Models, In A. Karlqvist, L. Lundqvist, F. Snickars & J. Weibull (Eds.), 

(pp. 75-96). Amsterdam: North Holland Publ. 
MNP (2004). Milieu- en nahiureffecten Nota Ruimte. Bilthoven: RIVM. 

OCW, LNV, VROM, & V&W (1999). Nota Belvedere, Ministerie van Onderwijs, Cultuur 
en Wetenschap, Ministerie van Landbouw, Natuurbeheer en Visserij, Ministerie van 
Volkshuisvesting, Ruimtelijke Ordening en Milieu and Ministerie van Verkeer en Waterstaat, 
SdU, Den Haag. 

RNP (1958). ontwikkeling van het Westen des lands. Werkcommissie Westen des Lands van de 
Rijksdienst voor het Nationale Plan (RNP), Staatsdrukkerij, The Hague. 

Rosen, S. (1974). Hedonic prices and implicit markets: Product differentiation in pure competition. 
Journal of Political Economy, 82(1), 34—55. 



9 Explaining Land-Use Transition in a Segmented Land Market 



175 



Segeren A. (2007). De grondmarkt voor woningbouwlocaties . Belangen en strategieen van 

grondeigenaren. The Hague/Rotterdam: Ruimtelijk Planbureau/NAi Uitgevers. 
Segeren, A., Needham,B.,& Groen, J. (2005). De markt doorgrond . Een institutionele analyse van 

de grondmarkten in Nederland. The Hague/Rotterdam: Ruimtelijk Planbureau/NAi Uitgevers. 
Van Rij, E., Dekkers, J. E. C, & Koomen, E. (2008). Analysing the success of open space 

preservation in The Netherlands: The midden-Delfland case. Tijdschrift voor Economische en 

Sociale Geografie, 99(1), 115-124. 
Verburg, P. H., Ritsema van Eck, J. R., De Nijs, T. C. M., Dijst, M. J., & Schot, P. (2004). 

Determinants of land-use change patterns in the Netherlands. Environment and Planning B: 

Planning and Design, 31 , 125-150. 



Chapter 10 

A Policy Perspective of the Development 
of Dutch Land-Use Models 

Marianne Kuijpers-Linde 



10.1 Introduction 

Preceding chapters show how land-use models have been used successfully in 
recent years to support decision-making processes for the formulation of new 
spatial policy strategies. In several of those chapters (see Chapters 4, 6 and 8), 
studies are described that explore the potential spatial patterns that may result 
from different policy alternatives related to, for example, climate change, the 
stimulation of bioenergy production and various strategies of urbanisation . This 
chapter focuses on future policy-related research questions that need to be addressed 
by the LUMOS toolbox. It describes, therefore, the LUMOS research agenda from 
a policy perspective. 

To be able to define future policy-related land-use modelling questions, first of 
all insight is needed into current policy themes and the spatial planning relationships 
that are now being addressed in policy discussions. For example, the theme of 
depopulation is now receiving more and more attention. Until 2000, the expansion 
of urban areas was modelled and depopulation was seen as a liveability issue 
in peripheral rural areas, which was not included in spatial models. No explicit 
calculation rules were included in recent applications of Land Use Scanner for 
the depopulation of urban areas. Recently, this issue has also been linked to other 
issues, for example to urban renewal: is a concentration policy in urban areas the 
correct spatial planning strategy? A question that is now being asked as part of the 
depopulation discussion is: 'what are the advantages and disadvantages of building 
in an existing urban area to meet a need for expansion?' To answer such a question, 
planners must not only model an expansion of the urban area, but also changes in 
the urban area. This means that the spatial characteristics of changes in the housing 
market must be included much more explicitly in spatial models. 

Second, to determine the modelling questions for the future, insight is needed 
into how policy is used to guide development. The points of leverage for policy 



M. Kuijpers-Linde (0) 

TNO Urban Development, PO Box 49, 2600 AA Delft, The Netherlands 
e-mail: marianne.kuijpers@tno.nl 

E. Koomen, J. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 177 
Practice, The GeoJournal Library 101, DOI 10.1007/978-94-007-1822-7_10, 
© Springer Science+Business Media B.V. 201 1 



178 



M. Kuijpers-Linde 



in the social and spatial system must be clearly distinguished in order to make a 
policy-relevant schematisation of reality - i.e. a model. The way in which spatial 
planning guides development changes over time and depends on the role that 
government plays in society. The Dutch philosophy of governance in land-use 
planning has, for example, changed drastically over the past few decades. Strategy 
embracing a centralised, development-control policy has been replaced by a one 
embracing a more localised, development-oriented policy. The different roles 
played by government can be illustrated with the policy themes of sustainable 
energy and environmental protection. To stimulate sustainable energy, government 
chooses primarily financial measures, whereas in environmental policy, government 
influences society primarily through the development of legislation and regulations. 
This implies that financial characteristics must be modelled when sustainable energy 
policy is the topic of research. However, if environmental policy is concerned then 
environmental standards based on spatial relationships between various types of 
land use should be included in the land-use model, for example as maps describing 
restrictions on various types of development. 

From an inventory of current spatial planning themes and an analysis of the new 
role and instruments of government, it is possible to define the requirements the 
LUMOS toolbox must meet. To do so, one needs to look at the possibility of deriving 
meaningful indicators from land-use simulation results. Indicators are quantitative 
evaluation measures that summarise results, preferably in relation to specific policy 
themes (Bubeck & Koomen, 2008). Policy variables (points of leverage for policy 
in the spatial and societal system) and indicators are selected based on knowledge 
about topical policy themes and policy instruments. The flood-risk assessment that 
was performed using Land Use Scanner (Van der Hoeven, Aerts, Van der Klis & 
Koomen, 2008) offers an example of a useful indicator. This indicator relates to the 
current theme of climate change policy in spatial planning. One of the major aims of 
this policy theme is to limit the risk of flooding (expressed as economic damage or 
potential casualties). Flood risk is, thus, an important evaluation criterion (indicator) 
for spatial plans: policy can focus on preventing building on flood-prone locations. 
This means that the model must take into account the relationship between urban 
development (a policy-related variable) and flood risk at specific locations. Together 
with the organisation of knowledge input into spatial planning, such indicators and 
policy variables form the basis of the research agenda for the LUMOS toolbox. 

The relationships between the LUMOS toolbox and policy are illustrated in 
Fig. 10.1 . The development of a policy-support model begins by surveying the aims 
and policy instruments to be found in the policy for which the model will be used . 
Researchers transpose the policy aims into indicators (model output) and the policy 
instruments into policy variables (model input). 

The structure of this chapter is as follows. The main, current spatial 
planning themes are introduced in Section 10.2. Section 10.3 then describes new 
developments in the policy instruments that are used to support the planning 
objectives. Finally, Section 10.4 discusses how the LUMOS toolbox should be 
adapted to help prepare new policies for current planning themes and related 
new instruments. It, thus, presents a number of modelling requirements that offer 



10 A Policy Perspective of the Development of Dutch Land-Use Models 



179 



Fig. 10.1 Method for 
portraying relationships 
between policy variables, 
targets , instruments and 
indicators (source: Zondag & 
Borsboom-van Beurden, 
2008) 



Indicators 



Policy variables 




► Targets 



Instruments 



directions for the further development of the LUMOS toolbox, which is discussed 
in the final chapter of this book. 



10.2 Substantive Spatial Planning Themes 

Themes in spatial planning typically follow developments in the economy, society 
and culture and are affected by dominant value orientation. Table 10.1 (Van 
der Cammen & De Klerk (2003) provides a clear overview of the history of 
land-use planning in the Netherlands, describing the most important spatial issues 
in relation to societal developments. The dominant land-use planning doctrines 
(values) described by Van der Cammen and De Klerk have been supplemented 
with topical policy dossiers of the departments involved. A striking feature of 
this summary is that since the 1980s a consistent choice has been made in the 
Netherlands for compact urbanisation and maintaining the urban-rural dichotomy 
(Faludi & Van der Valk, 1994; Koomen, Dekkers & Van Dijk, 2008). 

Relationship Between Societal Issues and Spatial Planning 

The issues on the spatial planning agenda have changed over the years and the 
overview presented in Table 10.1 clearly shows a succession of themes in the spatial 
planning dossier. For example, the expansion of the road network and the fanning 
out (or sprawl) of urban functions across the rural area in the 1970s, combined with 
the poor condition of inner cities, led to new attention for cities in the 1980s. In 
addition to the succession of themes on the agenda, interactions between societal 
developments and the way they are included in spatial planning are also apparent. 
Spatial issues are always linked to topical issues in society. The dossiers in the 
prevailing coalition agreement between the governing parties in the Netherlands for 
the period 2007-2010 are reflected in the spatial planning domain. For example, 
the financial crisis of 2009 has manifested itself in spatial projects as an investment 
issue. 



180 



M. Kuijpers-Linde 



.B ^ 8 E 

b b Oh 

U > ° 

> s 13 13 

- 1 ill 

| § S 3 -a 

5 1 S M § 

w — o 3 
bo 'u B "C g 
.9 »,2s 



I -S I 2 

E M a 3 

O Z OS OS 



{r 



O W 



S a 

i 

o -9 

8 g 
5 



a » s 
§ I -e 



oo 2 ■=. 
3 oo 3 

pa S u 



■ & S 



■a 2 



s 

c 
a. 



>. o 



B " 



w t» pq a 



> j2 a 

"JO aj 

s o £ 

1 c £ 



1 1 



3 OJ 

£•"8 13 

« -a 1 



o a 

n T3 



— o 



llll! 



a « 



1 ° I I | 



IT} IT} JO 



5 
93 



s 

3 

C 



1 8 -e 3 ■g 

2 3 U 00 
n 2 8 £ 

— « S3 



3 X) 



3 U .2 Th KS Oh 

feZQDZO 



S3 P 

& S3 



O T3 $ 2 




S 

3 5 '^3 



> -7! .a a 



> JS s 



■a 
fa 



JO cj O -3 !° 
£ £ BJ JO 



8- * 

Q 



5 <2 

II 

U 



■c 



A 



Policy 



Perspective of the Development of Dutch Land-Use Models 



SO O 

a a 

■C S 



_ C bo u 
CO 03 < OS 



.a ~ 



s 
g- 



a, 3 

0 10 13 

5 — >, > 

1 a 1 1 -s 1 
■1 ■£ § I -s i ' 

™ '35 ,3 w 9 q ■ 

6 o "3 S s 

O Oh > U ^ Oh 

a y 3 

5 < Ml 



J I S B IS 



3 S P 3 . 

•5 ■= a | q 

■3 «J la d "S 

S 3 g « S a 

3 u 5 ^ S d 

u> M d -E3 .S 



■3 "3 



.2 a s 



*3 I 

oo 5 c 



<u S bo b C 
d -=! <u a u 
W UQ2 > W 



to 
d 

d 4-1 «3 

1 S 3 I? 

S -2 



P 3 r- «1 



.2 S3 



u T3 & ^ 

a v b "g a $ 

oj t3 - - 



c/5 

■a 1 1 I 
I i 1 °f i 

z 



? 2 



3 ^3 "O "O 



"3 -a 

&o d 

I & 
a g 



§ .2 

U M U 
a <d u 

2 »h d 

M 2 J 



° S .2 | S a 2 o 3 » 



Oh 02 GO 



GO 



Q & 



B | 

o IS 



= o - 
3 Dji 



3 B 



g 1 



s 

§■ 



-c -a 

u , 

3 ~ 
2 s 

D..2 

9- eo • 



3 2 I 

! o | I 

3 3 a 1=0 

M ^ rt d 



1-i ^ w = , 



3 5 £ 



^3 Cti m 

Oh rt u g 



c ft d 



c g g g 
5 D- fo .23 



-Q 2 5 « 
d JJ ^ 



d u 33 



|| el 

Q — 



u u u 
.4S -a bo > w 



o as s 



■2 » S> a 

3 SS| 

E 1 -S * 

§ 111 

D- bo £ 

5 W 



182 



M. Kuijpers-Linde 



Relationship Between Sectoral Policy and Spatial Planning 

Policy developments in sectors other than spatial planning determine the topics 
on the spatial planning agenda, the underlying governance philosophy of that agenda 
and the set of instruments that are chosen. A good example of this is the external 
integration of environmental policy that has been implemented since the 1990s. 
The formal statutory link between land-use planning and various environmental 
dossiers (such as air quality and noise) has resulted in these issues remaining high on 
the policy agenda, despite major improvements in environmental quality. Another 
example is the strategic spatial vision for the Randstad (VROM, 2008), which is 
primarily linked to climate policy and traffic and transport policy. 

Role of Public and Private Actors 

Besides these interactions between various policy fields, the policy agenda in 
spatial development is determined by shifts in the balance between private and 
public actors and the distribution of tasks and competencies between tiers of 
government. Currently, the division of tasks between public and private parties is 
becoming less clear. Through their land policy, some municipalities are starting 
to act as entrepreneurs, while, in contrast, some private parties consider collective 
values in their investments . This implies that land-use models should pay attention to 
the changing agreements between public and private parties. In addition, collective 
values may receive an economic value in modelling applications. 

Role of National Government, Provinces and Municipalities 

Regarding the responsibility of the national government, provinces and 
municipalities, the National Spatial Strategy is very explicit: 'local if possible, 
central if necessary'. Based on current developments in many policy dossiers, it 
can be expected that the role of government in spatial planning will once again 
increase (due to the failure of market forces and the need for financial investment in 
collective values). This applies not only for the national tier, but also the regional tier 
(provinces and water boards). By means of a regional approach, a transition is taking 
place from diverse sectoral policy within the European and national restrictions 
to an integrated spatial planning strategy. At the national level, issues concerning 
security, climate and energy, water, nature, agriculture and landscape will require 
considerable centralised intervention. 

To support further development of the LUMOS toolbox, publications of PBL's 
departments were surveyed and interviews were conducted (Kuijpers-Linde & 
Koomen, 2009). This resulted in a list of important and topical themes in the spatial 
planning domain (Table 10.2). In the future, spatial models will be used to support 
policy that will be developed for these themes. This application of spatial models 
partly relates to questions about future developments (exploring the future) and 
partly to questions about potential options for guiding spatial development (ex-ante 
policy assessments). 

In the future, spatial planning will focus on the compatible and conflicting 
aspects of these issues. The LUMOS toolbox could be used in policy development if 
potential spatial developments are included in relevant model relationships, as will 
be illustrated in the following sections. 



10 A Policy Perspective of the Development of Dutch Land-Use Models 183 

Table 10.2 Most important policy themes in the spatial planning domain 

1 . Coping with climate change and energy supply. Climate change requires strong interventions 
focusing on mitigation and adaptation, mainly related to water management. What measures 
are necessary? And when should these measures be implemented? Mitigation measures are 
clearly linked to the energy theme 

2. Integrating sustainability in spatial planning processes. What does this mean and how should 
this be implemented? 

3 . Accommodating the need for growth in the Randstad and managing stabilisation and 
depopulation elsewhere. How can this trend be given shape in the new, general administrative 
orders and other instruments (investment plans, programmes and projects)? 

4. Intensifying the use of existing built-up areas, improving the accessibility of cities and 
strengthening main ports. This issue relates to the demand for urban space and the positioning 
of the Netherlands in European and international networks. What space is available in rural 
areas and what does this mean for the liveability in cities and the countryside? 

5 . Stimulating the agricultural economy. How can agriculture be strengthened while taking into 
account new European policy and possibilities for using agriculture to conserve landscape 
qualities? 

6. Basing nature conservation on dynamic biodiversity targets. In the context of climate change it 
may be necessary to adjust nature conservation policies to better fit the changing conditions. 
How can this be put in practice? 

7. Strengthening regional identity (landscape , culture-historical values). This issue is closely 
related to the Ministry of Housing, Spatial Planning and the Environment's theme of 
emphasising regionally oriented and differentiated policy. Increasingly, planning issues require 
an integrated regional approach. For example, improving the spatial quality of the rural-urban 
fringe and strengthening agriculture as a carrier of rural qualities . Who will steer these 
processes and which division of tasks among different governmental layers is needed? 



10.3 Policy Instruments for Dutch Spatial Planning 

In the spatial planning domain, there has not only been a shift in substantive 
focus, but also in the way in which government tries to achieve these aims. The 
development of policy instruments has been strongly influenced by changes in 
values and standards. Government is no longer seen as the primary force that guides 
spatial development, but more as one of the actors who works together to influence 
spatial development in the Netherlands. The role of the various tiers of government 
and the relationship between government and society have been set down in the 
Spatial Planning Act (WRO). This legislation went into force in January 2008, 
while the policy changes resulting from it are being implemented in phases. Tasks 
and competencies have been more clearly defined and ambitions and strategy have 
been separated. The various tiers of government lay down their indicative regional 
plans in their spatial strategy, as well as establishing the legal framework for zoning 
plans in their general administrative orders (national government) or regulations 
(provinces and municipalities). In this new Spatial Planning Act the procedures 
have been significantly simplified. The new instruments it provides not only make 
it possible to act more quickly, but also allow the complex relationships between 
various policy tasks - based on the interests of the various parties - to be accounted 
for in policy. This is illustrated by the fact that the majority of the topics in the 
general administrative orders arise from sectoral policy. 



184 



M. Kuijpers-Linde 



A distinction can be made between the various roles played by government: 

• organising; 

• developing; 

• researching; 

• agenda setting . 

Depending on the role taken by government, land-use models can be deployed in 
different ways, as is briefly explained below. 

Organising Role 

The organising role of government is specified in the general administrative 
order on spatial planning and in the regulations. Besides these spatial planning 
instruments, sectoral legislation also plays an important role. For example, the 
restrictions for spatial development are established in environmental legislation. 

The Dutch government has taken a strong position on the protection of 
collective values. Examples include protection of valuable landscapes, protection 
of biodiversity and protection of the population against flooding and environmental 
risks. This protection is provided by: 

• designating special areas (National Landscapes, National Ecological Network, 
flood defences); 

• setting limits on spatial development, for example by defining the contours of 
urbanisation; 

• guiding development by means of relative locations (spatial decision ladders, 
node policy) . 

In any case, the modelling instruments must take into account these procedures, 
which are set down in general administrative orders and regulations. This means 
that the calculation rules in land-use models can now differ more between regions: 
after all, each province emphasises different aspects when assessing spatial issues. 
It is important to survey the applicable laws and regulations . 

Developing Role 

Regional developments are used to realise ambitions for specific areas. Important 
characteristics of regional development are: 

• an integral approach to issues, based on a vision that is shared by all actors 
involved; 

• Public-Private Partnership (PPP); 

• complementary cooperation and subsidiarity with other processes: all parties do 
what they are good at doing, and overlap with other processes is limited; 

• a business case results from the characteristics specified above that specifies 
not only the targets (indicative plan), but also the means (a balanced budget or 



10 A Policy Perspective of the Development of Dutch Land-Use Models 



185 



operational plan): the new Land Servicing Act is an important instrument in this 
process; 

• surplus profit (the amount remaining after the agreed profit allocation to the 
parties involved) is a central theme: the surplus is to be invested in the collective 
values of the region (landscape, security, nature, etc.). 

Land-use models are very suitable for regional development because they can 
break down the complex relationships between many aspects of the region into 
components. As a result, it is possible to improve understanding of a problem, 
and therefore the understanding of tasks as well. The argumentation in a discussion 
between the parties can be supported by land-use models. For example, the relative 
effect of making changes in water resources management for agriculture, landscape 
and urbanisation can be clearly illustrated by map images and indicators, whereby 
existing agreements can also be taken into account. All other things being equal, 
alternatives can be evaluated as well. Land-use models can be used to map the 
effects of specific choices on the interests of the parties involved (possibly expressed 
as costs and benefits). During this process, a distinction can be made between 
short-term and long-term time horizons. 

However, such applications require the following: 

• the existing modelling instruments must be suitable for application at a regional 
scale; 

• the effect of land ownership must be included in the model (linked to the budget) 
and the costs and benefits must be independent variables; 

• it must be possible to model costs; and 

• it must be possible to specify the tasks for each land use-function in both 
quantitative and qualitative terms. 

Researching and Agenda-Setting Roles 

Spatial models can be used in exploratory studies to work out conceivable 
scenarios. One example of this is the use of Land Use Scanner in the 'Second 
Sustainability Outlook on the future of the Netherlands' study (MNP, 2007, see also 
Chapter 4). For each policy target, modelling instruments were used to literally map 
out storylines about future development. By comparing the map images, it became 
clear where possible spatial conflicts could occur. In this way, the details of a policy 
agenda can be assessed for consistency and completeness. 

Another example of the use of models for research and agenda setting is offered 
by the design workshops that were held as part of the 'Vision of the Randstad 2040' 
study (VROM, 2008 and Chapter 8 this volume). To support this creative process, 
explorations from the 'Prosperity and Living Environment' scenario study (CPB, 
MNP and RPB, 2006) were converted into map images. These provided support 
for developing new spatial planning strategies. The results of explorations such as 
these can be used to improve the model schematics and parameters of the LUMOS 
toolbox. 



186 



M. Kuijpers-Linde 



Besides providing these rules and the corresponding instruments, government 
can influence spatial development other ways too. For example, government can 
influence land use by means of financing (investment programmes, subsidies etc.), 
and by employing soft instruments such as design competitions, design workshops, 
regional process managers and public information campaigns. 



10.4 Policy-Based Land-Use Modelling Requirements 

From the discussion of policy instruments in Section 10.3, it is clear that policy 
developments present many difficulties for the further development of the LUMOS 
toolbox. Modellers face new challenges, especially relating to some of the new 
themes listed in Table 10.2, such as water management, urban restructuring and 
intensification, and new developments in the agricultural sector. In addition, new 
roles of government and related new instruments should receive attention in future 
model development. The knowledge requirements related to land-use modelling of 
these issues are discussed in the rest of this section. 

Water Management 

In the Netherlands, water is an aspect that must be included in many policy 
dossiers. For example, the nature policy dossier includes water quality and falling 
water-tables; the agricultural policy dossier includes water quality, water shortage 
and damage by waterlogging and flooding; the economics policy dossier includes 
risk of flooding. By means of spatial planning, the government can intervene in the 
hydrological system and in water supply and demand. The current version of Land 
Use Scanner does not explicitly model the characteristics of either of these factors. 
As a result, the effect of spatial development on risk of flooding, water nuisance, 
water shortage and falling water-tables has not been mapped out properly. Including 
the hydrological system in land-use modelling is an option that can be explored. 

Urban Restructuring and Intensification 

According to many observers, the era of expansion locations for big cities and the 
development of large business estates is over. Attention has shifted to intensifying, 
improving and transforming existing urban areas. Even without large expansion 
locations and new business estates, planners must still think about the desired 
spatial structure of the urban area and its interaction with the surrounding rural 
area. Networks play an important role in this process. Creating smart links between 
networks and coordinating these links (nodes) is a second theme that is central to 
many policy issues. Policy leverage points include the accessibility of locations, the 
qualities of the residential environment and the qualities of businesses estates. The 
mutual relationship between network development and other spatial developments 
must be included in new land-use model applications, from which greater insight 
can be acquired about the role of effective policy on the accessibility of various 
locations. 



10 A Policy Perspective of the Development of Dutch Land-Use Models 



187 



Developments in the Agricultural Sector, Partly in Relation to Water, Nature and 
Landscape 

Model schematics are useful for dealing with themes relating to rural areas 
for which urbanisation is an exogenous variable. The central focus is the 
demand for 'smart' combinations of agriculture with societal functions such as 
water management, nature management, cultural history and recreation such that 
agriculture continues to develop within the given economic conditions and new 
EU agreements. Important policy leverage points are the economic structure of 
agricultural areas, the cultural-historical values of the rural area, the recreation 
potential of specific areas, the qualitative and quantitative characteristics of the 
hydrological system, and the changes in these abiotic conditions for nature. 

The research agenda for land-use modelling is determined not only by the 
substantive planning themes discussed above, but also by the characteristics of 
policy and the role of knowledge in policy. Points of attention include more 
emphasis on regional policy and dynamic modelling and a clear specification of 
the role of land-use models in formal planning assessment procedures. These issues 
are discussed below. 

More Emphasis on Regional Policy and Dynamic Modelling 

Many complex spatial planning tasks require regional specification of policy. 
To support choices in regional specifications, it is important to picture the effects 
of alternatives. A land-use model can help depict these effects if relevant spatial 
developments are modelled in detail, the factor of time is made explicit, and spatial 
interaction (such as agglomeration and competition effects) and actor behaviour are 
included in the model. 

Specifying the Role of Land-Use Models in Formal Planning Assessment Procedures 
During the development of policy, the application of land-use models in 
formal assessment procedures must also be taken into account. Instruments that 
will become increasingly important in the future are: strategic environmental 
assessments (SEAs and sustainability assessments); social cost-benefit analyses 
(including collective values such as biodiversity, security, health and cultural 
history); and business cases for spatial investments. The use of these process 
instruments will lead to a convergence in the assessment criteria that are being 
used in policy discussions. It is important that investments in the LUMOS toolbox 
anticipate this development. 



10.5 Closing Remarks 

New challenges in spatial planning require modifications to the available LUMOS 
toolbox. Policy requires further specification of applications such as those for the 
various themes described in this section. It is unlikely that all new policy questions 
can be answered by a single integrated model. Yet, a start has already been made on 



188 



M. Kuijpers-Linde 



substantial improvements to Land Use Scanner to make it better equipped to answer 
(part of) the questions discussed above. For example, in the 'Sustainability outlook 
on the Netherlands' (MNP, 2007) an attempt was made to make components of the 
model more dynamic and link these to a transport model (see also Chapters 4 and 5). 
The model was, furthermore, linked to a flood-risk model (Van der Hoeven et al., 
2008) and was used in several SEAs of Regional Strategic Visions (see Chapters 7 
and 8). All these applications show that Land Use Scanner can be further improved 
in the context of specific policy -related applications. The greatest challenges for the 
near future are to expand this model based on these new requirements, to renew 
these and other components of the LUMOS toolbox and, at the same time, continue 
to apply it in policy -related research. If expansion of the toolbox is primarily focused 
on the policy questions related to water management and urban intensification and 
restructuring, new elements to be incorporated are: 

1 . information on the location of activities and actors, and the objects in the physical 
environment they use for these activities, for example, land and real estate; 

2. information on actor behaviour and its consequences for land use and the 
transport system, for example, location choices of companies, substitution in the 
housing market, and responses to transport policy measures; 

3. information on the dynamics of use of geographical space; 

4. a structural link between land use and the transport system; 

5. improved linkage to hydrological models for the assessment of water 
management measures in relation to changes in land use. 

During further development of the LUMOS toolbox, the organisation of the 
knowledge input for spatial planning must also be taken into account. In spatial 
planning, there is continuing formalisation of this knowledge input and ongoing 
professionalisation of knowledge management. For example, in the Netherlands 
the method for estimating the costs and benefits of spatial investments has been 
systematically described. When designing a spatially explicit land-use model, it 
is important to determine which role the method should play, which land-use 
categories must be distinguished and which spatial relationships are relevant. 



References 

Bubeck, P., & Koomen, E. (2008). The use of quantitative evaluation measures in land-use change 
projections; An inventory of indicators available in the land use scanner. Amsterdam: Spinlab 
Research Memorandum SL-07. Vrije Universiteit Amsterdam/SPINlab. 

CPB, MNP and RPB (2006). Welvaart en Leefomgeving. Een scenariostudie voor Nederland in 
2040. Den Haag: Centraal Planbureau, Milieu- en Natuurplanbureau en Ruimtelijk Planbureau. 

Faludi, A., & Van der Valk, A. (1994). Rule and order: Dutch planning doctrine in the twentieth 
century. Dordrecht: Kluwer. 

Koomen, E., Dekkers, J., & Van Dijk, T. (2008). Open space preservation in The Netherlands: 
Planning, practice and prospects. Land Use Policy, 25(3), 361-377. 



10 A Policy Perspective of the Development of Dutch Land-Use Models 



189 



Kuijpers-Linde, M., & Koomen, E. (2009). Beleidsvragen en indicatoren voor een nieuw 

ruimtegehruiksmodel. Amsterdam: Geodan Next. 
MNP (2007). Nederland Later; Tweede Duurzaamheidsverkenning deel fysieke leefomgeving 

Nederland. MNP-publicatienr.500 12700 1/2007. Milieu- en Natuurplanbureau, Bilthoven. 
Van der Cammen, H., & De Klerk, L. (2003). Ruimtelijke ordering van grachtengordel tot 

Vinex-wijk. Utrecht: Het Spectrum. 
Van der Hoeven, E., Aerts, J., Van der Klis, H., & Koomen, E. (2008). An integrated 

discussion support system for new Dutch flood risk management strategies. In S. Geertman & 

J. C. H. Stillwell (Eds.), Planning support systems: Best practices and new methods, 159-174. 

Berlin: Springer. 

VROM (2008). Nota Randstad 2040. Ministerie van Volkshuisvesting. Den Haag: Ruimtelijke 

Ordening en Milieubeheer. 
Zondag, B., & Borsboom-van Beurden, J. A. M. (2008). Uitwerking project LUMOS 1 .0 in 2008. 

Interne notitie. Bilthoven: Planbureau voor de Leefomgeving. 



Chapter 11 

Developing a New, Market-Based 
Land-Use Model 

Judith Borsboom-van Beurden and Barry Zondag 



11.1 Introduction 

This book describes the extensive experience that PBL Netherlands Environmental 
Assessment Agency and its partners have built up in more than a decade of 
using land-use models to support policy-making. The land-use models of PBL - 
Land Use Scanner and Environment Explorer - have in several large studies 
contributed substantially to the research findings and policy recommendations, 
making them a standard component of the analytical framework of outlooks where 
spatial dynamics are essential for sustainability and environmental quality in future. 
The performance of the models has been evaluated both by PBL and an audit 
committee of international experts; the findings of the committee are summarised 
in Chapter 2. The overall conclusion is that the models represent the state of the art 
for their current practice, but that to enlarge their potential contribution to policy 
questions and to address new policy challenges a substantial model redesign, or the 
development of a new model, is needed. Important challenges in doing so would 
be the inclusion of the behaviour of key actors in the model chain, the inclusion 
of essential feedback loops between different sectors and geographical scales, the 
structural inclusion of transport within the land-use model, and a better integration 
of water management and land use to be able to evaluate spatial adaptation strategies 
related to climate change. The insights from the different subject-specific research 
findings have been highlighted and processed to formulate several basic features 
to be included in the design of a new land-use modelling framework. In this final 
chapter, we describe the way forward for the development of such a land-use model. 
After an evaluation of achievements and drawbacks of the current model chain for 
the support of strategic policy questions, we discuss the ambitions and options for 
a new land-use model. Subsequently, the various activities carried out to arrive at 
the general specifications for the new model and the results achieved so far are 
highlighted . 



J. Borsboom-van Beurden (H) 

TNO Behavioural and Societal Sciences, PO Box 49, 2600 AA Delft, The Netherlands 
e-mail: judith.borsboom@tno.nl 

E. Koomen, J. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 191 
Practice,The GeoJournal Library 101, DOI 10.1007/978-94-007-1822-7_ll, 
© Springer Science+Business Media B.V. 201 1 



192 



J. Borsboom-van Beurden and B. Zondag 



11.2 Evaluation of Land Use Scanner Applications 

Since the first version of the Land Use Scanner was developed in 1997, the 
model has been applied in a wide range of policy assessment studies at PBL (see 
Borsboom-van Beurden et al., 2005; Koomen, Kuhlman, Groen & Bouwman, 2005; 
MNP, 2004a, 2004b; RIVM, 1997; RIVM & Stichting DLO, 2001, 2002; Van de 
Velde et al., 1997). The most recent applications of the Land Use Scanner model by 
PBL and its partners have been discussed in several of the previous chapters of this 
book, including the comprehensive 'The Netherlands in the future' study that is a 
part of the Second Sustainability Outlook (MNP, 2007; PBL, 2010). 

The diverse applications of land-use models in policy assessment studies have 
been evaluated internally at PBL. In addition, an international audit committee 
reviewed both Land Use Scanner and Environment Explorer in 2007 (see Chapter 2) . 
Both models have been judged by the international audit committee in 2007 as 
'among the best in their field, and very useful for assessing policy impacts relating 
to the spatial distribution of land use at the national level' (See Chapter 2 by Harry 
Timmermans, Michael Batty, Helen Couclelis, and Michael Wegener, this volume). 
Nevertheless, the audit committee concluded at the same time that a substantial 
redesign of the model or the development of a new activity-based model would 
be required to overcome the drawbacks of the current models. In addition, from 
the overview of current policy questions in Chapter 10, it follows that reliable, 
more detailed model outcomes are needed at a sub-national, or regional, level, 
which cannot be fully provided at this time. This section summarises the main 
achievements and drawbacks of the current set up of the land-use modelling chain, 
including that present in Land Use Scanner. 

A major achievement of the current Land Use Scanner model is that it 
incorporates an extensive knowledge base, developed by PBL and its partners, 
containing information about the drivers for and progress of land-use changes. 
This knowledge base is actually formalised in the data underlying the model, 
such as figures on future demand for land, allocation rules and about 400 geo- 
datasets representing the suitability of particular land uses. The knowledge base 
also comprises detailed information on the impact of current and potential policy 
measures, not only for spatial planning measures such as urban containment and 
designated development sites, but also for sector policies such as water management 
and nature development. As such, this knowledge base enables the model to chart the 
spatial consequences of current trends and the potential impact of policy measures 
in the form of 'what-if reasonings. 

This is a truly integral knowledge base: all types of land use have been worked 
out in detail. Many models claiming to be integral are in fact only directed at 
urban functions. 'Integral' means then the integration of housing, employment and 
transport, but does not include more rural functions. One category of examples of 
this partial integration are the so-called LUTI models (Land Use and Transport 
Interaction models), as developed by Miller, Kriger, and Hunt (1998), Simmonds 
and Enchenique (1999), Zondag and Pieters (2001), and Wegener (2004). The 
category of 'Spatial Computed General Equilibrium' models, which focus on 



1 1 Developing a New, Market-Based Land-Use Model 



193 



the regional economic developments from a systems perspective, as developed 
by Oosterhaven, Knaap, Ruigrok, and Tavasszy (2001) and Brocker (2004), also 
exemplifies this partial integration. Other families of models are mainly directed at 
rural functions so that urban functions are somewhat neglected; the CLUE-S model 
(Verburg and Overmars, 2009) is one such model. 

Another achievement of Land Use Scanner is that the spatial elaboration of 
diverse trends and policy measures using a land-use model has really contributed to 
the preparation of strategic policies by 'framing' possible futures. Different aspects 
can be highlighted here. It appears that the application of a land-use model often 
turns out to work as an integrating framework: it visualises the spatial consequences 
of various trends and policies which might have otherwise remained unknown. 
Land-use models show where contradictions and frictions may occur, and where 
more coherence between policies could be pursued. Especially when policies are 
developed separately in specific sectors without considering the impact on other 
sectors, such integration can be very valuable. This support of the interactive process 
of policy preparation and policy-making by 'raising awareness' has proven to be as 
important for policy makers as the actual model outcomes (Guhathakurta, 2003). 
Further, results of land-use modelling give insights into the pressure that is placed 
on weaker functions such as agricultural land, valuable landscapes, nature areas, 
and flood-prone areas. This pressure might be contrary to the spatial, environmental 
and ecological policies pursued for these areas. What is more, land-use models can 
show where different land uses might be combined or optimised. By mapping the 
best options under certain conditions ('what-if'), the possibilities for optimisation 
of spatial structure and land use become clearer. Both conflicting and optimising 
outcomes can set the agenda for strategic policy-making by signalling desired or 
undesired spatial changes. 

The most comprehensive example of such a spatial elaboration of trends is the 
'Second Sustainability Outlook' study (PBL, 2010; see also Chapters 4 and 5), 
which contributed to the preparation of the Randstad Vision 2040 of the Dutch 
Ministry of Housing, Spatial Planning and the Environment (VROM, 2008; see 
also Chapter 8). By investigating the demand for land in the congested Randstad 
conurbation for several decades to come, and to show which land-use changes 
could be expected and how land use could be optimised according to various 
sustainability goals, the development of a new spatial strategy for the Randstad area 
was supported. Another important contribution was an extensive investigation of 
adequate responses to the risks posed by climate change that was commissioned by 
the Dutch Cabinet. The land-use changes simulated in the 'Second Sustainability 
Outlook' study clearly showed the concentration of demographic and economic 
growth in the western part of the Netherlands, and the increasing potential damage 
in cases of flooding (Delta Commission, 2008). 

These examples of the contributions of land-use model outcomes in key policy 
documents demonstrate that Land Use Scanner has been successfully applied in 
the preparation of strategic policies in the Netherlands; in many other countries 
land-use modelling has remained a predominantly academic activity with a strong 
methodological-technical focus. Despite these successes, internal evaluations and 



194 



J. Borsboom-van Beurden and B. Zondag 



an international audit have brought a number of drawbacks to light (Borsboom & 
Zondag, 2009): 

1. Lack of information about actors, activities and spatial objects apart from their 
land consumption 

In a number of studies, it appeared that not all environmental and spatial 
problems could be addressed adequately because the current model disregards 
the distribution of households and persons, businesses, employment, farms, etc., 
when it simulates land use in the future. Often this information is available at an 
earlier stage in the model chain, for example, in sector models at a sub-regional 
scale. However, only their translation into future claims for land is processed 
in Land Use Scanner, usually assigning it to coarser spatial units. As a result, 
various indicators that require information on the future spatial organisation of 
households and businesses, such as exposure to air pollution, congestion arising 
from transport measures and the potential casualties and economic damage 
resulting from flooding, cannot be properly assessed. Although a workaround 
was developed for the 'Second Sustainability Outlook' study that assigned data 
on population distribution after land used was allocated, this solution was not 
ideal. Besides, under the current set-up the preferences and behaviour of actors 
are simulated indirectly by means of suitability maps for a specific type of land 
use, instead of being directly modelled via the location choices of households 
or firms. A more direct simulation of the behaviour of key actors, such as 
households, firms or farms, is needed to reflect the variety in responses of 
different types of households, firms or farms, to changes in circumstances, such 
as the physical environment or higher energy costs. 

2. Lack of dynamics in land-use simulation 

The static approach of Land Use Scanner, which follows on from equilibrium 
modelling, is not in line with the usually gradual, path-dependent, changes that 
occur in land use. What is more, equilibrium modelling is mainly motivated 
by economic theory: the existence of an equilibrium in practice has not yet 
been proven. It can be argued that such a general equilibrium does not exist 
at any moment in time, due to various time lags and differences in the pace of 
change for various parts of the spatial system. This has been demonstrated by 
Wegener (2004), who gave an overview of the different time spans of change in 
the nine elements of the urban/regional system. A dynamic approach, allowing 
for partial equilibriums in specific markets, is therefore to be preferred as it is 
much more capable of addressing the relatively small changes in land use and 
the path dependency of these changes (Simmonds, 1999; Zondag, 2007). 

3. Lack of consistency between sector models mutually and between sector models 
and the land-use model 

As we have noted, sector-specific models providing input for land-use models 
were developed for a particular purpose: for example a housing market model 
for making forecasts of future housing needs, or an agro-economic model 
for analysing agricultural profitability. Therefore it is not surprising that the 
coherence and consistency of assumptions, spatial units, definitions and lines of 



1 1 Developing a New, Market-Based Land-Use Model 



195 



reasoning in all these models can be improved significantly (Dekkers & Koomen, 
2006). A major point of improvement is the inclusion of the mutual relation 
between employment, housing and transport, as Timmermans (2007) identified 
for the Welfare, Prosperity and Quality of the Living Environment study (CPB, 
MNP & RPB, 2006). An additional problem to the limited coherence and 
consistency between models in the model chain is the unidirectional character 
of the model chain in which Land Use Scanner is applied, not allowing the 
modelling of feedback mechanisms which exist in reality. As a result, changes in 
the availability of land and restrictions for specific land-use types are not dealt 
with in the sector models. Essential feedback mechanisms are lacking, such as 
substitution in the housing market for other types of housing or other locations, 
in response to housing shortages in particular cities. 
4. Insufficient elaboration and quantification of indicators 

Many indicators cannot process Land Use Scanner's output adequately. Causes 
are highly divergent spatial units for analysis, incompatible definitions and 
the lack of information on actors or activities needed for assessing exposure 
to risks. 

Considering the drawbacks it was decided to fundamentally redesign the Land 
Use Scanner model. 

Prerequisites for this new LUMOS model are: 

1 . The model must simulate for the whole of the Netherlands, so the model has to be 
geared to national and regional analyses and is not designed for local analyses; 

2. The model must be fully multi-sectoral, meaning that all relevant land uses, urban 
as well as rural, have to be covered; 

3. The model must clearly distinguish trends from scenarios, by extrapolation of 
current developments or well-supported assumptions for important variables 
such as housing need or economic growth. The model has to be fitted for both 
types of applications. 

4. The model must simulate simultaneously land use, actors/activities and the 
objects needed for these activities; these layers have to be integrally addressed 
in the model design, while essential feedback, such as transport in relation to 
housing and employment have to be included. 

5. The model must be dynamic and simulate intervals of one or 2 years, which 
shifts the emphasis from modelling of equilibriums to the modelling of changes 
in time, allowing path-dependency and temporary impacts - for example, the 
impact of policy measures between 2015 and 2025. 

6. The model must be developed by PBL jointly with preferred external partners. 

Work on the requirements of the new model has entailed the charting of strategic 
policy questions, identification of the most important spatial processes and drivers, 
exploration of various options for the model design, and improvement of the 
calculation of various indicators, especially those for flooding risks. 



196 



J. Borsboom-van Beurden and B. Zondag 



11.3 Requirements of the Model: Policy Questions 
and Indicators 

At request of PBL, Geodan Next explored the policy questions that were expected 
to potentially need outcomes of integrated land-use modelling (Kuijpers-Linde & 
Koomen, 2009) in the coming 5 years. This investigation has been discussed in detail 
in Chapter 2. The relevant policy questions found, which are closely interrelated, can 
be summarised in three coherent blocks: 

1 . Urban restructuring, transformation and liveability 

a. What are the possibilities and limitations for densification, urban 
restructuring and urban containment? 

b. How can more sustainable cities be created, for example in regard to their 
space-efficiency and energy-efficiency? 

c. How to respond to the impact of demographic processes on urbanisation 
patterns, such as population decline and immigration, from the viewpoint of 
segregation and liveability? 

d. How to reduce congestion and environmental pressure in the cities? 

2. Water management and biodiversity 

a. What are the risks resulting from climate change and what needs to be done 
to further mitigate for and adapt to climate change? 

b. What will be the impact of changes in the energy system on spatial 
organisation and land use, and what are the spatial consequences of 
alternative energy sources? 

c. How to maintain a water management system in which both the quality and 
quantity are rated good? 

d. How to prevent further losses of biodiversity, also taking into account the 
consequences of climate change, and how can biodiversity policy aims be 
made more flexible so that it is able to respond to climate change? 

3. Retaining the identity and aesthetic value of Dutch landscapes 

a. How to prevent further deterioration in Dutch landscapes? 

b. How to reinforce the vitality of Dutch agriculture without harming nature and 
cultural landscapes? 

From this overview, it can be concluded that since the middle of the 1990s there 
has been a shift in the policy questions being dealt with: from the accommodation 
of a growing and competing demand for land to a better management and utilisation 
of land already in use by a particular function. Besides, it appears that to be able 
to answer many of these policy questions, reliable model outcomes are needed at a 
regional or even sub-regional scale, in addition to the national level. These outcomes 
should also include more information than just land use, namely information on 



1 1 Developing a New, Market-Based Land-Use Model 



197 



actors/activities and objects, to be able to produce answers for these policy questions 
(Borsboom & Zondag, 2009). 

11.4 Requirements of the Model: Driving Forces 
of Land-Use Change 

In order to create an overview of the main driving forces behind changes in spatial 
organisation and land use, a number of experts were invited to participate in working 
groups for the themes of housing, businesses, agriculture, water management and 
nature management. In addition, PBL experts carried out research to identify 
overarching themes such as transport, leisure and energy (De Niet and Van 
Middelkoop, 2009; Jacobs, 2010; Rijken, 2009). Topics addressed were the major 
driving forces of spatial changes, their spatial materialisation, changes in spatial 
distribution and intensity of land use, and international aspects such as the influence 
of European and global developments and policies on spatial organisation and land 
use. All working groups and experts concluded their investigations with a paper 
describing the main spatial dynamics for their sector and the most important drivers 
(see Atzema, Korteweg, Lambooy & Van Oort, 2009; Priemus & Hoekstra, 2009; 
Van Bruchem & Silvis, 2009; Groot et al., 2009). Zondag and Borsboom (2009) 
summarised their main findings, which are presented in Table 11.1. 

This summary of drivers was used to identify the following requirements that the 
new land-use model should meet (see Zondag & Borsboom, 2009 for a complete 
overview): 

• A shift in focus from modelling quantity to modelling quality. Such a shift includes 
more attention for the modelling of characteristics of existing spatial structures 
and the processes of transformation or renewal of these structures. Traditionally, 
land-use models at PBL focus on changing land-use types, such as urbanisation 
on agricultural land. More attributes of the existing spatial structure and built- 
up area will have to be included to be able to model its relevant qualitative 
aspects, for example land and property ownership, social composition and value 
of biodiversity. 

• Increased possibilities for simulation of regional diversification. It is expected 
that in the coming decades the Netherlands will face patterns of regional growth 
and shrinkage in terms of population and employment. These processes will also 
have an impact on the functionality of areas, since they affect the potential for 
large-scale agriculture, nature development, leisure (day recreation and holiday 
homes), public transport and other services. Traditionally, land-use models focus 
on growth: the demand for land from various sectors is allocated to scarce land 
resources. 

• Integrated and simultaneous modelling of land use (for example agriculture), 
objects (for example farms) and actors (for example farmers). In particular, the 
inclusion of some characteristics of the main actors, allowing for some detail 
in the modelling of their responses to the driving forces of land use change, 



198 



J. Borsboom-van Beurden and B. Zondag 



Table 11.1 Overview of the main drivers of land-use change 



Housing • Demographic drivers remain important; a new phenomenon for the 

Netherlands is population shrinkage in more peripheral regions. Other issues 
are an increasing variety in household types, migration and the behaviour of 
migrants 

• Increasingly important are economic and financial drivers , such as household 
income and capital, and financial instruments of the government 

• On the supply side, the existing spatial patterns (e.g. houses, employment and 
infrastructure) and policies are dominant. Changes occur more and more 
within the existing housing stock 

• Relationship between supply and demand is becoming more and more 
complex (definition of a house, second homes) 

Businesses • Drivers and then' impact differ strongly by sector (industry, transport and 

distribution, consumer or business services) - economic structure effects play 
role too in the demand for land 

• Besides traditional location factors (labour cost, transport cost and land 
prices), soft location factors are becoming more important (skills of 
employees, access to knowledge, residential amenities, image etc.) 

• Strongly interact with demographic developments in regions and transport 
facilities 

• In the real estate market there is a shift in focus on quantitative growth towards 
qualitative improvements, as the renewal and restructuring of existing sites and 
buildings . Other dominant trends are concentration and densification 

Agriculture • Conversion from agricultural land to urban land or nature areas strongly driven 
by developments exogenous to the agricultural sector 

• Changes in type of agricultural production (crop, cattle and horticulture) are 
mainly influenced by developments on world markets , technological 
developments and environmental restrictions 

• Multi-functional land use is being driven by changes in income support by 
European Common Agricultural Policy, stressing more and more nature and 
landscape management instead of further rationalisation and rise of 
production, but also by societal changes as the increasing importance of 
recreational activities in the countryside and a rising number of hobby farmers 

Nature and • Societal and economic drivers such as outdoor leisure activities or short stay 
water vacations and housing at the water's edge influence quantity and quality of 

management both water and nature 

• Climate change influences land use via its impact on physical drivers, such as 
intensification of rainfall and drought, rise of the sea level and salt intrusion, 
and via adaptation strategies (additional water storage and development 
restrictions) 

• Quantity and quality aspects of nature and water management interact strongly 
with developments in the agriculture sector, leading to scale increases and 
regulation of water levels and emissions 

• EU policies are becoming more and more important, e.g. EU water framework, 
Natura2000 (and the national ecological network) and the nitrate guideline 



is needed. It is known that within the same class of actors (such as households, 
companies and farmers), people respond differently to the same drivers, 
depending on their characteristics. For example, a classification of households by 
income is needed to adequately model their responses to road pricing strategies. 
Another example is the classification of households by size, since a household 
with children is likely to prefer other types of housing and residential locations 



1 1 Developing a New, Market-Based Land-Use Model 



199 



than a household without children. The future use of geographical space in terms 
of objects and land is highly influenced by relative shifts in the breakdown of 
categories, for example in population composition, socio-economic background, 
and economic structure. Changes in these breakdowns such as more or less 
households with children, more or less high income households, or more offices 
and less manufacturing, all have an impact on the use of geographical space. The 
reason for this is that variables explaining that use, such as accessibility or urban 
green, work out differently in the location choices of each category of actors. 

• More thorough modelling of the impact of climate change. Climate change affects 
physical conditions in the Netherlands as a result of, for example, rising sea 
levels, periods of heavy rainfall and drought, and increasing seepage of salt water. 
As such, it affects the amount of land needed for nature, water management 
and agriculture, but also the way this land is utilised. In addition, flood risk 
for urbanised regions along coasts and rivers is likely to determine future land 
use. The water system plays a crucial role when calculating the impacts of 
climate change or addressing the effectiveness of adaptation strategies. Therefore 
the interactions between water and land use in the broadest sense should be 
included more fundamentally in any new land-use model. In the current model 
chain, however, mutual interactions between water and land use are virtually 
lacking. Although several maps are available that describe locations with a 
reduced or enhanced suitability for specific land-use types, e.g. groundwater 
protection areas that limit the suitability for new housing, water management 
is poorly represented. Surface water is exogenously planned for as a land- 
use category, and information on hydrological relationships is not included at 
all. For the new LUMOS model framework, it will be important to include 
such hydrological relationships to enable the establishment of upstream and 
downstream relationships . Insight in the hydrological relationships could be used 
to restrict upstream developments at certain locations to protect downstream 
functions. Or alternatively, this information could be used to select, for example, 
suitable locations for nature development. 

• Attention for the impact of energy systems. Energy transition, driven by climate- 
change mitigation policies and high energy prices, is an important and sensitive 
scenario variable, affecting the land use of different sectors in diverse ways. Land 
can be used directly to produce energy (windmills, biomass, etc .) and to distribute 
it. The indirect impacts of energy -related changes, for example through mobility 
or housing costs, are potentially large because they affect the spatial organisation 
of production and consumption chains and the costs of construction. 

11.5 Options 

11.5.1 Exploring Options 

One can conclude from Sections 11.3 and 11.4 that the set-up of the current 
model chain no longer suffices for answering current policy questions. When 
the LUMOS project started in 2008, the initial intention was to build, conform 



200 



J. Borsboom-van Beurden and B. Zondag 



the advice of the international audit committee, a 'full-fledged integrated agent 
based simulation framework', which would in the long term replace Land Use 
Scanner. Such a framework could make use of state-of-the-art insights, methods 
and techniques in spatial modelling, such as micro-simulation, and could enable the 
simulation of mutual interaction between actors or processes as firm demography 
(Van Wissen, 2000). A micro-representation of activities in space would also allow 
feedback processes that take place at a local level to be simulated. An example 
of such a feedback process is the influence of traffic-based noise pollution on the 
attractiveness of residential locations. 

A micro model often requires the creation of synthetic data to represent the 
individual level of people, households and firms. In addition, stochastic variation 
is introduced to simulate decisions at the micro level, which introduces the problem 
that multiple model runs based on the same assumptions and causal relations, 
result in different outcomes. This is not desirable when policy questions have to 
be answered: model outcomes usually have to repeatable in such cases. Similarly 
for disaggregate types of models, the coefficients for the explanatory variables in 
micro models often need to be estimated on a larger population in order to have 
enough observations . Further, because of the large data volumes and potentially long 
calculation times for a micro model, complex software will need to be developed to 
ensure efficient data management and acceptable run times . 

The development of a completely new, actor-based integrated model, as preferred 
by the international audit committee in 2007, is no longer considered to be a realistic 
option given the time-span, risks and costs associated with it. Examples of such 
model developments as TLUMIP, ILUTE and ILUMASS (Salvini & Miller, 2005; 
Wagner & Wegener, 2007; Weidner, Donelly, Freedman, Abraham & Hunt, 2006) 
show that, in spite of the large effort in terms of labour and costs, these models are 
not yet fully operational (ILUTE, ILUMASS) or are still scarcely used for policy- 
making (TLUMIP). Therefore the applied methodologies and techniques cannot be 
considered as proven technology. The high expectations for the added value of actor- 
based techniques for the modelling of complex spatial systems have so far failed to 
materialise in actual policy-making (see, for example, Ligtenberg, 2006; Ettema, 
Floor &De Jong, 2009). 

In the meantime, the existing PBL model infrastructure has grown much larger 
due to a merger between MNP and RPB (Netherlands Bureau for Spatial Analysis) 
in 2008, resulting in doubt about the approach to take. In particular, it raises the 
question as to whether LUMOS could also be developed by better utilisation of parts 
already developed within the PBL model infrastructure, such as the demographic 
model PEARL (RPB/CBS, 2005) and the employment model WEBER (De Graaff, 
Van Oort & Florax, 2009). Furthermore, PBL has become co-owner, together with 
the Ministry of Transport, Public Works and Water Management, of the TIGRIS XL 
land-use and transport interaction model. 

Two options for building the LUMOS model framework along these lines 
were explored: building the framework around the LUTI model Tigris XL and 
expanding the Land Use Scanner model. By choosing Tigris XL as the base for 
further development of the LUMOS model framework, nearly all requirements 
of the new model as formulated in Section 11.4 can be fulfilled at lower costs, 



1 1 Developing a New, Market-Based Land-Use Model 



201 



with less development time and at less risk, while the option to expand the Land 
Use Scanner model does not lead to the fundamental improvement of the model 
advised by the audit committee. Although this implies that actor behaviour is not 
represented at a micro-scale but in an aggregated form in Tigris XL, it still allows 
for the representation of heterogeneity in actors and their behaviour, for example 
by distinguishing various types of households exhibiting different responses. This 
set-up also allows for the representation of activities and objects at a detailed 
geographical scale. The structural coupling to the underlying transport model 
enables the assessment of the impact of policy measures such as road pricing and 
investment in public transport. In addition, the Tigris XL model is already dynamic 
because it produces results in time-steps of one year. Moreover, this option opens up 
possibilities for a better connection to sector models such as the demographic model 
PEARL, the employment model WEBER, and the housing model HOMERA. As 
a result, this option is well equipped to answer most policy questions concerning 
urban transformation and restructuring, e.g. possibilities for densification in built- 
up areas, accessibility benefits of urban containment, and the space-efficiency of 
urban sprawl. Borsboom and Zondag (2009) consider the policy questions relevant 
for LUMOS in more detail. 

The Tigris XL option also brings with it the additional advantage that 
elaboration - expansion and deepening - of the model framework can be done 
gradually, by adding modules or meta-data that can be worked out in more detail 
at a later stage. Therefore the development of the new LUMOS model framework 
by elaboration of the Tigris XL model has been proposed as the preferred option 
for developing a new LUMOS model (internal note, Borsboom-van Beurden & 
Zondag, 2009). 



11.5.2 Expanding Tigris XL 

This option does not aim to build a very detailed 'state-of-the-art activity-based 
land-use model', but to build a 'state-of-the-practice market-based land-use model' 
(Miller et al., 1998). In the latter sort of model, the segmentation of the population 
or firms in disaggregated groups should be detailed enough to make it possible 
to simulate the main differences in behaviour. Examples of this are the different 
responses of high- and low-income car owners to the levying of tolls, or the 
different strategies of farmers towards the reform of EU Common Agricultural 
Policy. To develop this option, new model architecture is needed in which land use 
is treated dynamically and actor behaviour is simulated more explicitly and at a 
lower scale than that used so far. For many policy questions, it is precisely the intra- 
regional developments that are of crucial importance: examples are issues such as 
possibilities of urban densification, the impact of agriculture on biodiversity, and the 
consequences of measures for mitigation of climate change on activity patterns and 
transport flows. These developments require much more information on the housing 
market, labour market, transport system and rural functions than is available at the 
moment from Land Use Scanner. 



202 



J. Borsboom-van Beurden and B. Zondag 



The most obvious choice for realising the framework would be to expand the 
existing Tigris XL model, which already includes integral architecture for urban 
functions. The model simulates interactions between demography and housing, 
employment, transport and, to some extent, the availability of land. Other sectors 
such as agriculture, nature and water management first have to be added to the 
framework. The building of the LUMOS model framework can be organised 
using a modular architecture, which allows for a step-wise, gradual extension and 
improvement of the model's framework. To be able to simulate not only land use but 
also the distribution of actors/activities and objects, more knowledge from sector 
models, such as those on demography and housing, needs to be incorporated in the 
modules. 

Then the match between supply and demand for land and objects can take 
place efficiently at the scale where it actually occurs, usually at a regional level, 
corresponding with the spatial unit applied in the sector models, or an aggregation 
of it. This integrated model framework should interact in time with an underlying 
GIS database of grid data containing information on suitabilities, which determine 
the availability of land for specific uses. In this way, factors defining the supply of 
land at grid-cell level, such as designated nature conservation areas or greenhouse 
concentrations, are translated into the higher scale, where supply and demand are 
matched. At the same time, changes in the demand for land, for example as a 
result of the approval of a plan for the construction of housing, are translated to 
this lower level of the individual grid cell (see Fig. 11.1). Such a set-up takes 
advantage of the fact that the sector models are in general strong in the modelling 
of the demand for objects and land, while the supply of objects and land is often 



Scenario values for economy, demography energy, technology, societal values, 
world trade, world agricultural market, climate change (provided by specialised institutions) 



Integrated framework, 
modules for demography, 
housing, labour, transport, 
agriculture, nature and 
water management 
(see figure about sectors 
in model) 



t+1 



Integrated framework, 
modules for demography, 
housing, labour, transport, 
agriculture, nature 
and water management 

demand modelling and 
match 



Supply 
features by 

actor 
(zone level) 



Allocation of 
land by type 



By cell: 
changes in 
land-use, 
suitability, 
actors and 
objects 



t+.. 



Integrated framework, 
modules for demography, 
housing, labour, trans- 
port, agriculture, nature 
and water management 



Supply 
features by 

actor 
(zone level) 



Allocation of 
land by type 



0- 



By cell: 
changes in 
land-use, 
suitability, 
actors and 
objects 



Supply 
features by 

actor 
(zone level) 



By cell: 
changes in 
land-use, 
suitability, 
actors and 
objects 



Fig. 11.1 Model architecture of the new LUMOS model 



1 1 Developing a New, Market-Based Land-Use Model 



203 



Scenario Input 

I I I I I I I I 



Transport 
market 



Demography 



Persons 
households 



Housing 
market 



Firms 
jobs 



Real estate 
market 



Demand for 
product 



Agriculture 
market 



Houses 



gl Office space 
com. land 



Agriculture 
production 



Nature 



Water 



Land-use 



Land market 



Land supply/suitability 



Fig. 11.2 Overview of the sectors incorporated in the new LUMOS model framework 



poorly represented. Land Use Scanner currently contains highly-detailed spatial 
information on the suitability and availability of land, through GIS-data on soil, 
spatial planning, groundwater protection, noise pollution, etc. The modelling of 
supply in sector models can be enhanced by using this information in an aggregated 
form at the regional level . 

To reduce investment costs and risks for the development of this framework, 
where possible, existing models and data will be used. Based on all these elements, 
a global design of the LUMOS model framework has been drafted and is presented 
in Figs. 11.1 and 1 1 .2 although this description is clearly only a very rough sketch, 
which definitely needs to be worked out in more detail before the system can be 
built. Recently, databases and specifications of the demographic model PEARL 
were inserted in TIGRIS XL for the Spatial Outlook project. This could be a first 
building block for the framework. 



11.6 Results So Far and the Road Ahead 

Since 2008, the various research initiatives in the LUMOS project have roughly 
defined the contours of the new LUMOS model. The study by Geodan Next clarified 
the policy perspective (see Section 1 1 .3 and Chapter 10) and the thematic expert 



204 



J. Borsboom-van Beurden and B. Zondag 



groups charted the main drivers behind spatial changes (see Kuijpers-Linde & 
Koomen, 2009; Priemus & Hoekstra, 2009; Atzema et al., 2009; Van Bruchem & 
Silvis, 2009; De Niet & Van Middelkoop, 2009; Jacobs, 2010; Rijken, 2009; 
Groot et al., 2009). In addition, Ettema et al. (2009) explored the architectures of 
comparable systems and their application in practice. Their study confirmed that in a 
policy setting, such as that of PBL, developing a micro- or agent-based model is too 
risky in view of the length of time needed for development, the huge data demands, 
the long computation time and the increasing detail of the behavioural models. Their 
recommendation, therefore, was to tailor the level of detail (or breakdown in classes 
of agents) to the policy scenarios to be evaluated. Furthermore, the ICT aspects 
of the development of such a model or model framework have been explored by 
Hilferink and Grashoff (2009). They drafted a check-list of recommendations for 
the integration of various model components within a framework. 

Now that the option to build the framework around the Tigris XL model has been 
chosen, measures to implement this decision in the next years have to be taken. 
The development of the new LUMOS model will take place in a stepwise fashion. 
Seven essential steps for realising a first version of the new model as presented in 
Figs. 11.1 and 11.2 are described below. The numbering of these developmental 
steps corresponds with the numbers in the figures, indicating which part of the 
framework is improved by that particular step. The following steps are foreseen 
in the short term (2010-201 1): 

1. Preparation of the base year 

First of all, to be able to make simulations using the model prototype, the 
required data on actors/activities, objects and land use have to be collected for a 
specified base year. GIS-data on land use and suitability from Land Use Scanner 
can be used at the grid cell level of 100 m x 100 m. The data on actors/activities 
and objects needed at this level must be derived from geographically very 
detailed datasets, such as business registrations (LISA), address data (CAN), 
and farm data (BIN). After being collected, the data will have to be harmonised 
with and made compatible to the land-use data and other data at the regional 
level. For example the total amount of land used by agriculture according 
to the Agricultural Census is not corresponding to the total amount of land 
used by agriculture according to the Land Use Statistics dataset from Statistics 
Netherlands. 

2. Addition of modules lacking in the framework 

The current Tigris XL model contains modules for demography, housing, 
transport, employment, real estate and land. For carrying out truly integral 
analyses, such as an investigation of the area available for agriculture in the 
light of demand for land for housing, employment, water management and 
nature development, modules for the non-urban domain have to be added in the 
framework. It seems wise to start with the development of a number of modules 
that contain meta-data from the current configuration of Land Use Scanner and 
to gradually improve on this by designing a more satisfactory classification 
of agricultural land use, adding objects such as farms and greenhouses, and 



1 1 Developing a New, Market-Based Land-Use Model 



205 



adding attribute information such as landscape management and environmental 
pressure. The same goes for nature management. In the Spatial Impressions 
project, a very detailed elaboration of nature management using highly detailed 
classes and allocation rules was designed that proved to be difficult to put 
into practice because of a lack of appropriate data (Borsboom-van Beurden 
et al., 2005). However, evaluation of new nature developments in the National 
Ecological Network showed that the possibilities for the realisation of specific 
nature types mainly depends upon local biophysical conditions. For that reason, 
later configurations as used in the Sustainable Netherlands study simplified this 
approach, but their oversimplification hampered an assessment of these future 
land-use changes in terms of biodiversity. Therefore a moderate approach is 
advised when adding a module on nature management to the framework. 

3. Design of a new real-estate and land-market module 

The current Tigris XL model contains a very basic real-estate and land module, 
which computes changes in housing supply or land availability for a specific 
function on the basis of existing land use and a coarse representation of different 
planning regimes. The incorporation of the cell level in the framework enables 
much more information from GIS-data at the grid cell level to be input, at the 
grid-cell level the characteristics of supply of objects and land and characteristics 
of the environment determining the suitability or attractiveness for specific 
functions are known. The GIS-data at grid cell level are then aggregated and 
used to provide information on the availability and suitability of objects and 
land in a zone or municipality. Thus, at the grid-cell level the characteristics of 
supply of objects and land and characteristics of the environment determining the 
suitability or attractiveness for specific functions are known. The characteristics 
of the demand for objects and land, and demand/supply ratios from the past year, 
can be derived from the other modules in Tigris XL. In the real-estate and land 
module, all this information is combined in order to assess the match between 
the supply of and demand for locations, leading to changes in both the use of 
objects and land. 

4. Design of an allocation procedure at the cell level 

An efficient allocation procedure needs to be established at the cell level to 
translate the changes computed at higher levels (municipalities, zones) to specific 
cells. This procedure can be largely based on the current software used in 
Land Use Scanner. Intelligence-based, calibrated, transition rules should be 
capable of delivering a better performance at a higher level than a random- 
based process. Nevertheless, it will be impossible to make precise forecasts for 
a specific cell: the outcomes need to be interpreted at a higher scale. Further, 
the new set-up implies the modelling of changes, not of current land use, in 
contradiction with the current Land Use Scanner. The incremental structure of 
the new modelling framework guarantees that most of the land use, as in reality, 
remains unchanged and that only the relatively small changes are processed. This 
calculation mechanism at the cell level should also ensure consistency between 
the different layers, such as actors, objects and land, in the modelling. The large 
number of cells when a spatial resolution of 100 m x 100 m is used and the 



206 



J. Borsboom-van Beurden and B. Zondag 



maintenance of the links between all data layers representing actors, objects and 
land make computation time and data management at this level important aspects 
in selecting the right approach. 

5. Integration of the characteristics of supply in other modules 

The improved information on building and land supply characteristics as made 
available in the new real-estate and land module described under Point 3 should 
be included in the housing market, labour market and agriculture modules. The 
supply information can be used to improve the quality of existing explanatory 
variables in sector models, to introduce new variables in parameter estimations, 
or to detail the options for actors by assigning more attributes (such as houses 
versus houses by type). The advantage of this integration is that the supply of 
objects and land can play a larger part in the spatial choices in sector models. 
This would imply, then, that a re-estimation of the modules is needed if new 
variables or a more detailed set of options are added. 

6. Link with new (or light) version of the National Transport Model 

The National Transport Model (LMS) of the Ministry of Public Works, Transport 
and Water Management is an integrated part of the TIGRIS XL framework. It is 
the ambition of the Ministry and PBL to maintain this set-up and to integrate 
the new version LMS 2010 in the framework. However, it is expected that the 
transport module will require the most computation time of all modules. For 
applications requiring a less detailed focus on transport, the inclusion of a light 
version of the new LMS is considered as it could be very beneficial for reducing 
total computation time. 

7. Continuation of the work on water indicators 

In 2009, a better connection between the current version of Land Use Scanner 
and DamageScanner (De Bruijn, 2008) was realised by PBL and Deltares. 
This was done to improve the use of land-use modelling outcomes for the 
computation of indicators describing the exposure to flooding risks. Alternative 
spatial developments can be evaluated based on the potential risks they create for 
flood damage. In the coming period, other water management themes of interest 
need to be further integrated with land-use changes. Major issues are the lack 
of water or a local surplus of water and flooding, related to consequences of 
climate change as intensified periods of drought and rainfall. Another major 
issue is the quality of water, in relation to the protection of biodiversity and 
interactions between agriculture and nature. This integration can be done by 
calculating in a post-processing step the impacts of alternative land-use plans 
on water indicators, and by using information on the water themes in the design 
of spatial alternatives. 

After a prototype of the new model - incorporating the new framework - has 
been developed, it is important to test the new system thoroughly by applying it to 
diverse 'real life' policy questions or scenario explorations. The test phase should 
measure the performance of the system and generate a list of priorities for further 
development. 



1 1 Developing a New, Market-Based Land-Use Model 

11.7 Conclusion 



207 



Since 1997, both Land Use Scanner and Environment Explorer have made valuable 
contributions to a wide range of policy assessment studies carried out. However, 
now, nearly 15 years after their development began, progress in computing power, 
in the availability of detailed data (in particular GIS-data) and in the methods and 
techniques available for the modelling of human activities have opened up a wide 
range of new possibilities in this field. These changes have shed new light on the 
desired performance and sophistication of Land Use Scanner and Environment 
Explorer. At the same time, a shift in policy questions since the 1990s, as charted 
by Kuijpers-Linde and Koomen (2009), has prompted the need for the availability 
of reliable outcomes of land-use modelling at lower scales and for more attention to 
be paid to the state and management of the environment in its broadest sense. 

The PBL initiative to include the Tigris XL model in the new LUMOS 
model framework addresses these new challenges and opportunities. The new 
system derives its strength from the combination of modelling expertise from 
different modelling backgrounds: actor-driven demand modelling and the integrated 
modelling of the urban and regional system is adopted from the Land Use and 
Transport Interaction models and this expertise is combined with detailed land- 
use data and modelling from a cell-based land-use model. The expertise of specific 
modules, such as demography, housing, labour, transport and agriculture, is further 
enriched within the context of the overall framework by knowledge and information 
derived from sector specific models in these fields . 

It is clear that the road ahead in developing a new LUMOS model will not always 
be easy and undoubtedly there are many pitfalls to be avoided, but all constituents 
required for its successful development by PBL and its partners are present. We 
therefore hope that this endeavour, which represents a truly collaborative effort, will 
be broadly welcomed, supported and rewarded with the success it deserves. 



References 

Atzema, O., Korteweg, P., Lambooy, J., & Van Oort, F. (2009). Factoren achter het ruimtegebruik 

van werken.Een literatuurverkenning . Utrecht: Utrecht University. 
Borsboom, J., & Zondag, B. (2009). Visiedocument modelontwikkeling voor ruimtegebruiks- 

modellering bij het PBL. Bilthoven: Internal note, PBL. 
Borsboom-van Beurden, J. A. M., Boersma, W. T., Bouwman, A. A., Crommentuijn, L. E. M., 

Dekkers, J. E. C, & Koomen, E. (2005). Ruimtelijke Beelden; Visualisatie van een veranderd 

Nederland in 2030. RIVM report 550016003. Bilthoven: Milieu- en Natuurplanbureau. 
Brocket', J. (2004). Computable general equilibrium analysis in transportation economics. 

Chapter 12. In DA. Hensher, K.J. Button, KE. Haynes, & P.R. Stopher (Eds.), Handbook 

of transport geography and spatial systems (pp. 269-289). Elsevier, Amsterdam. 
CPB, MNP and RPB (2006). Welvaart en Leefomgeving. Een scenariostudie voor Nederland 

in 2040. The Hague: Centraal Planbureau, Milieu- en Natuurplanbureau en Ruimtelijk 

Planbureau. 

De Bruijn, K. (2008). Bepalen van schade ten gevolge van overstromingen. Voor verschillende 
scenario 's en bij verschillende beleidsopties. Deltares, Utrecht: Report in commission of IVM. 



208 



J. Borsboom-van Beurden and B. Zondag 



De Graaff, T., Van Oort, F. G., & Florax, R. J. G. F. (2009). Sectoral differences in regional 

population - employment dynamics in the Netherlands. Paper prepared for the North American 

Regional Science conference, San Francisco. 
De Niet, R., & Van Middelkoop, M. (2009). Drijvende krachten achter recreatie en toerisme. 

Quickscan voor LUMOS 1.0. Internal note. The Hague/Bilthoven: Planbureau voor de 

Leefomgeving. 

Dekkers, J. E. C„ & Koomen, E. (2006). De rol van sectorale inputmodellen in 

ruimtegehruiksimulatie ; Onderzoek naar de modellenketen voor de LUMOS toolbox. SPINlab 

research memorandum SL-05. Amsterdam: Vrije Universiteit. 
Delta Committee. (2008). Working together with water. A living land build for its future. Findings 

of the Delta Commissie 2008. Summary and conclusions. 
Ettema, D., Floor, H., & De Jong, T. (2009). Behavioural mechanisms in land use change models. 

Literature scan. Utrecht: Faculty of Geosciences/Utrecht University. 
Groot, A., Moors, E., Vos, C., Vellinga, P., & Opdam, P. (2009). Drijvende krachten achter 

veranderingen in ruimtegebruik voor de sectoren water en natuur, rapport in opdracht van 

PBL. Wageningen: University of Wageningen. 
Guhathakurta, S. (2003). Advances in urban and environmental modeling: Surveying the terrain 

and demarcating frontiers. In S. Guhathakurta (Ed.), Integrated land use and environmental 

models. A survey of current applications and research (pp. 3-12). Berlin: Springer. 
Hilferink, M., & Grashoff, P. (2009). ICT aspecten Lumos 2.0 ontwikkeling. Object Vision and 

Demis. 

Jacobs, C. (2010). De drijvende kracht van mobiliteit. Achtergrondstudie bij de vervaardiging van 

een nieuw LUMOS model. Amsterdam: VU University. 
Koomen, E., Kuhlman, T., Groen, J., & Bouwman, A. A. (2005). Simulating the future of 

agricultural land use in The Netherlands. Tijdschrift voor Economische en Sociale Ceografie 

(Journal of Economic and Social Geography), 96(2), 218-224. 
Kuijpers-Linde, M., & Koomen, E. (2009). Beleidsvragen en indicatoren voor een nieuw 

ruimtegebruiksmodel, In opdracht van het Planbureau voor de Leefomgeving. Amsterdam: 

Geodan Next. 

Ligtenberg, A. (2006) . Exploring the use of multi-agent systems for interactive multi-actor spatial 

planning. Wageningen: Thesis Wageningen University. 
Miller, E . J . , Kriger, D . S . , & Hunt ,J.D.(1998). Integrated urban models for simulation of transit 

and land-use policies. Final Report TCRP Web Document 9, University of Toronto. 
MNP (2007). Nederland Later; Tweede Duurzaamheidsverkenning deel fysieke leefomgeving 

Nederland. MNP-publicatienr.500 12700 1/2007. Bilthoven: Milieu- en Natuurplanbureau. 
MNP (2004a). Quality and the future. Sustainability outlook. Bilthoven: Milieu- en 

Natuurplanbureau. 

MNP (2004b). Milieu- en Natuureffecten Nota Ruimte, RIVM-rapport 711931009, RIVM, 
Bilthoven. 

Oosterhaven, J., Knaap, T., Ruigrok, C., & Tavasszy, L. (2001). On the development of RAEM: 
the Dutch spatial general equilibrium model and its first application to a new railway 
link. Proceedings 41st Congress of the European Regional Science Association, Zagreb, 
Croatia. 

PBL (2010). The Netherlands in the future. Second sustainability outlook: The physical living 
environment in the Netherlands. Bilthoven: Netherlands Environmental Assessment Agency 
(PBL). 

Priemus, H., & Hoekstra, J. (2009). Drijvende krachten achter woningvraag en ruimtegebruik 

wonen. Achtergrondstudie in opdracht van het Planbureau voor de Leefomgeving. 

Onderzoeksinstituut OTB/Technische Universiteit Delft. 
Rijken, B. (2009). De impact van energie als drijvende kracht van ruimtegebruiksveranderingen. 

PBL, Bilthoven: Literature Scan. Internal PBL note. 
RIVM (1997). Nationale Milieuverkenning 1997-2020. Bilthoven: Rijksinstituut voor 

Volksgezondheid en Milieu. 



1 1 Developing a New, Market-Based Land-Use Model 



209 



RIVM and Stichting DLO. (2001). Who is afraid of red, green and blue? Toets van de Vijfde 
Nota Ruimtelijke Ordening op ecologische effecten, RIVM-rapportnr. 711931005, Wilco BV, 
Amersfoort. 

RIVM and Stichting DLO. (2002). National nature outlook 2: 2000-2030. Summary. Bilthoven: 
RIVM. 

RPB/CBS (2005). Achtergonden en verondertsellingen bij het model PEARL - Naar een nieuwe 

regionale bevolkings- en allochtonenprognose , NAi Uitgeverij Rotterdam. 
Salvini, P., & Miller, E. J. (2005). ILUTE: An operational prototype of a comprehensive 

microsimulation model of urban systems . Networks and Spatial Economics, 5, 217-234. 
Simmonds, D. (1999). The design of the Delta land-use modelling package. Environment and 

Planning B: Planning and Design, 26, 665-684. 
Simmonds, D. C, & Echenique, M. (1999). DETR report, Review of land use/transport interaction 

models, London. 

Timmermans, H. (2007). Welvaart en Leefomgeving: reflectie op scenario ontwikkeling. Internal 

report. Technical University Eindhoven. 
Van Bruchem, C, & Silvis, H. (2009). Drijvende krachten toekomstig landgebruik landbouw. The 

Hague: LEI. 

Van Wissen, L. (2000). A micro- simulation model of firms: Applications of concepts of the 

demography of the firm. Papers of Regional Science, 79(2), 11 1-134. 
Van de Velde, R. J., Schotten, C. G. J., Van der Waals, J. F. M., Boersma, W. T, Ouwersloot, H., & 

Ransijn, M. (1997). Ruimteclaims en ruimtelijke ontwikkelingen in de zoekgebieden voor de 

toekomstige nationale luchtinfrastructuur (TNLI). Quickscan met de Ruimtescanner. RIVM- 

rapport 71 1901024. Bilthoven: RIVM. 
Verburg, P., & Overmars, K. (2009). Combining top-down and bottom-up dynamics in land use 

modeling: Exploring the future of abandoned farmlands in Europe with the Dyna-CLUE model. 

Landscape Ecology, 24, 1 167-1 181 . 
VROM (2008). Randstad towards 2040, summary of the structural vision. The Hague: Ministry of 

Housing, Spatial Planning and the Environment. 
Wagner, P., & Wegener, M. (2007). Urban land use, transport and environment models - 

experiences with an Integrated Microscopic Approach. 
Wegener, M. (2004). Overview of land use transport models. Chapter 9. In D.A. Hensher, 

K.J. Button, K.E. Haynes, & P.R. Stopher (Eds.) , Handbook of 'transport geography and spatial 

systems (pp. 127-146). Elsevier, Amsterdam. 
Weidner, T, Donelly, R., Freedman, J., Abraham, I. E„ & Hunt, I. D. (2006). TLUMIP - transport 

and land use model in Portland - current state, Stadt Region Land 81, Aachen: institute fur 

Stadtbauwesen und stadtverkehr, RWTH Aachen, pp. 91-102. 
Zondag, B. (2007). loint modeling of land-use, transport and economy, PhD thesis TRAIL thesis 

series nr. T2007/4, Delft University of Technology, Delft. 
Zondag, B.,& Borsboom,!. (2009). Driving forces of land-use change. Paper prepared for the 49th 

ERSA conference August 2009, Lodz, Poland. 
Zondag, B., & Pieters, M. (2001). Literature Review of Land-Use model, prepared by RAND 

Europe for the Transportation Research Centre of the Netherlands Ministry of Transport, Public 

Works and Water Management, Leiden. 



Index 



A 

Accessibility, 11 , 25, 46, 62, 65-67 , 70-7 1 , 
74-75 ,79,81-83,85, 90-93 , 99 , 1 25 , 
129, 134, 137, 155, 164, 171, 181, 183, 
186,199,201 

Activity-based model, 27-29, 34, 43^4, 46, 
192,201 

Actor, 11,24,27,31,37-38,43^16,94, 154, 
171, 182-184, 188, 191, 194-195, 
197-202,204-207 

Agent based simulation, 27-29, 38-40, 43^4, 
46,51,200,204 

Agriculture, 3, 7, 10, 17,28,43,71,73-75, 
84,86,94,98, 105, 109, 112, 114, 
134, 137, 141, 153-154, 159-160, 
162-163, 165, 173, 180-183, 185, 
187, 193, 196-199,201-204, 
206-207 

Air pollution, 25, 27, 97, 129, 194 

Allocation module, 9-10 

Autonomous development, 6, 32, 132 

B 

Baseline scenario, 6-8, 63-67, 70-74, 76, 79, 

89-92, 132, 137 
Bid price, 12,42,45,99, 108-109, 111, 

168-172 

Biodiversity, 7, 62, 66-67, 69, 74, 79, 183-184, 

187,196-197,201,205-206 
Bioenergy,97-114, 177 
Biomass, 97-114, 199 
Business park, 7, 10, 17,62 

C 

Calibration, 16, 24-25 , 28 , 3 1 , 5 1 , 1 70 
Climate adaptation, 89, 131, 133, 141-145, 
181, 191 



Climate change, 3-4, 61-62, 66-70, 74, 76, 
89, 97-114, 118, 126, 137, 141-146, 
177-178, 183, 191, 193, 196, 198-199, 
201-202,206 
Commercial development, 1 1 , 83, 1 19-120 
Common Agricultural Policy (CAP), 71 , 75, 
198,201 

Communication, 24, 29, 32-34, 76, 136 
Conflicts, 61, 64, 74, 122, 143, 145, 182, 185, 
193 

Consistency, 19, 80, 84, 86, 93, 129, 185, 

194-195,205 
Continuous model, 11, 13-14, 16,24 
Core principles, 35-53 
Cost-benefit analysis, 100-105, 108, 

180, 187 

D 

DamageScanner, 18,206 
Data Model Server (DMS), 18 
Decision-maker, 6, 26, 29, 31-32 
Demographic development, 4-6, 27-28, 63, 

65,81-84, 87,90-91,99-100, 106, 

118-119, 124, 127, 138, 193, 196, 198, 

200-201,203 
Demographic model, 25, 43, 187-188, 

200-201,203 
Densification, 34, 82-83, 135-141, 147-148, 

183,186, 196,198,201 
Density,41,44,66,87,93, 137-139, 147-148 
Discrete model, 14-16,24 
Driving forces, 45, 99, 107, 154, 173, 

197-199 

Dynamic modelling, 25, 36, 43, 45, 187-188, 
194 

E 

Economic theory, 25, 99, 180, 194 
Energy consumption, 27, 97 



E. Koomen, 1. Borsboom-van Beurden (eds.), Land-Use Modelling in Planning 
Practice, The GeoJournal Library 101, DOI 10.1007/978-94-007-1822-7, 
© Springer Science+Business Media B.V. 201 1 



211 



212 



Index 



Energy transition, 181, 199 

Environmental Impact Assessment, 24, 119, 

123, 127, 133, 147, 181 
Environmental issue, 27, 180 
Environmental policy, 4-8, 23, 1 18, 178, 182 
Equilibrium model, 95, 192, 194 
Ethanol, 97-98, 100-102, 104, 109, 1 13 

F 

Filtration, 98, 100, 104-105 

Flooding risk, 18, 62, 66, 74, 92, 120, 122, 

125-126, 132, 137, 178, 188, 195, 199, 

206 

Flood protection, 70, 73 

Future spatial patterns, 4, 63, 132-133, 177 

G 

GIS-data, 4, 203-205, 207 
Green heart, 75, 147 

Greenhouses, 7-8, 10,71,74-75,97, 120, 122, 
124, 133, 141, 161-162, 164, 167-168, 
171,202,204 

H 

Hedonic Pricing Method (HPM), 153, 
155, 172 

Household, 4, 26-27, 42, 80-83, 86-88, 90, 
106, 138 

Housing, 5, 7, 10, 17,27-28,61-62,65-66,68, 
70,72,76,79-89,94, 131, 134-135, 
137-141, 177, 180, 183, 188, 192-195, 
197-199,201-207 
market, 17, 80-83, 86-87, 134, 177, 188, 
194-195,201,203,206 

Hydrology, 7, 10, 16-18,86, 108, 111-112, 
126, 186-188, 199 

I 

Impact assessment, 6, 17, 24, 31-32, 72, 

119-120, 123, 125-127, 129, 133, 147, 
181 

Implementation issue, 4, 16-19, 107 
Indicator, 18, 64, 66, 70-72, 74, 79, 85, 88, 91, 

93-94, 123, 125-127, 129, 178-179, 

185, 194-196,206 
Industrial development, 82, 87-88, 99, 

133-135, 146, 181 
Inertia, 10,26-27 

Instrument , 1 7 , 3 1 -32 , 70 , 79-80 , 94 , 1 78- 1 79 , 

182-187, 198 
Integrated model, 17, 27-28, 36, 88, 170, 187, 

193, 196,200,202,207 



Integration, 9, 32, 36, 61, 67, 75, 91 , 94, 131 , 
134, 142-143, 145, 180-182, 191-193, 
204, 206 

Intensification, 44, 70-71 , 74-75, 140-141, 

186, 188, 198 
International business establishment, 62, 66, 

70-71,74 

L 

Labour market, 17, 79, 81-83, 86-88, 90, 94, 
201,206 

Land demand, 4, 9-10, 16-18, 24-25, 30, 
61-63,71,76,79, 105, 107, 132, 
134-135, 137, 139, 141, 143, 192-193, 
196-198,202,204 

Land market, 12,81-83, 108, 153-174,203, 
205 

Land price, 13, 148, 153-155, 158, 162, 

164-165, 167, 172-173, 198 
Landscape, 5, 11, 18,36,41,61,63 
Landscape quality, 63, 66-67, 69-71 , 73-75, 

79, 112, 114, 120, 122, 124-127, 

131-132, 137, 180-185, 187, 193, 196, 

198,205-206 
Land use model, 5, 24, 38, 40, 44, 46^17, 53, 

79-95, 178, 187-188, 191-207 
Land use transition, 153-173 
Land-use and transport interaction (LUTI) 

model, 79-95, 192,200 
Linear probability model, 155, 157, 159, 172 
Literature review, 35-53 
Living environment, 6, 61-63, 66-67, 70-72, 

74,76, 106, 119, 180, 185, 195 
Logit model, 12,24,41,84 
LUMOS,3,23-35,40,45,52, 177-179, 182, 

185, 187-188, 195, 199-204,207 

M 

Map Comparison Kit, 3,18 

Micro simulation, 28, 43^14, 46, 200 

Model chain, 16, 18-19, 26, 28, 181, 191, 

194-195, 199 
Model development, 23, 30-31 , 80, 186, 200 
Model validation, 35, 47^18 
Mono-functional land-use, 30,41, 187, 198 
Municipality, 80-83, 87-88, 94, 205 

N 

National Government, 165, 182-183 
Natura 2000 area, 66, 69, 71, 74-75, 123, 
125-126 

Natura 2000 policy, 7-8, 126, 198 



Index 



213 



Nature area, 5, 7-8, 13, 81, 99, 120, 126, 

133-134, 165, 193, 198 
The Netherlands in the Future, 61 , 76, 79, 192 
Noise pollution, 66, 86, 121 , 200, 203 

O 

Optimisation, 4, 6-8, 1 1 , 15-16, 24, 36, 121 , 
127, 193 

P 

Path dependency, 5 1 , 80, 95, 194-195 
Planning concepts, 128, 131, 134-136, 138, 

146-148, 180-181 
Policy alternatives, 117-118, 120-121, 

123-127, 133-134, 138, 177 
Policy formulation, 30, 1 17-118, 123, 128 
Policy preparation, 193 
Population decline, 10, 181, 196 
Private actor, 1 82 

Province, 8,65,85,90, 117-129, 131-133, 

139, 141-145, 155, 159-160, 162-163, 
168-169, 182-184 

Public actor, 182 

Q 

Quantitative assessment, 128-129 
R 

Raising awareness, 193 

Randstad, 61 , 65-67, 70, 74-75, 89-90, 1 19, 

131, 134-141, 147-148, 160-162, 164, 

167-168, 171, 181-183, 185, 193 
Reed, 98-105, 108-114 
Regional diversification, 197 
Regional spatial planning, 117-118, 120 
Regional Spatial Strategy, 1 17-129 
Residential development, 90, 121-122, 124 
Resolution, 11, 14,27,32,45,49,52-53,85, 

91,94, 99, 108, 118, 131, 142-143, 

146, 170,205 
River discharge, 69 
Road-use pricing, 70, 75 

S 

Scaling, 12-13, 15-16,49-50, 171-172 
Scenario-based simulation, 4— 6, 31, 82, 117, 
143, 147 

Scenarios, 3-6, 17, 28, 30-32, 63-65, 67, 

79,83,87,89,99, 104-107, 111-113, 
118-120, 122, 127, 129, 132, 134, 
139-141, 145-148, 185, 195,204 

Sectoral policy, 182-183 



Sector-specific information, 9-10, 12, 16, 79, 
128-129, 131, 134, 141-145, 148, 194 
model, 9-10, 16, 79, 148, 194 
Shadow price, 12, 15-16, 172 
Societal issues, 179 

Spatial analysis, 45, 117, 127, 129, 180-181, 
200 

Spatial Computed General Equilibrium model , 
192 

Spatial concepts, 135, 180-181 

Spatial exploration, 1 19-122 

Spatial planning, 4-8, 12, 40, 51, 53, 61, 
65,74-75,80-81,84,92, 117-118, 
120-121, 126-127, 131-132, 134-135, 
137-138, 153-155, 172-173, 177-188, 
192-193,203 
constraint, 153-154, 173 

Strategic environmental assessment, 123-127 

Suitability, 9-10, 12-16, 18,42,85, 107-111, 
119-120, 134, 138, 171-172, 194 

Sustainability, 5-6, 8, 41, 61-64, 66-72, 

74-76,79,84,87,89,93, 112,118-119, 
123, 125-129, 132-135, 138-140, 181, 
183, 185, 188, 191-194 

T 

Tangible user interface, 143-145 
Tigris XL, 17, 79-89, 91 , 93-94, 200-207 
Time steps , 9 , 25-26 , 40 , 45 , 49 , 52-53 , 80 , 
86,94,201 

Transition probability, 10, 153, 155-159, 162, 

165-168, 170-171 
Transport model, 25, 27, 81 , 83, 91, 93, 134, 

188,201,206 
Trend-based simulation, 4, 6-7, 133 

U 

Uncertainty, 4, 31,44, 118, 138, 146-147, 156, 

168-169, 181 
Urban fringe, 153, 172-173, 183 
Urbanisation^, 11, 18,27,45,65,67-71, 
73-75, 81, 83-84, 86-87, 120-122, 
124, 127-128, 131, 134-141, 143, 
146-147, 154, 158, 161-162, 165, 
167-168, 177-180, 184-188, 196-197, 
201 

Urban restructuring, 87, 186, 196 

Utility, 11-12,24-26,28,83-84,92,94, 171 

V 

Visualisation, 13, 19, 32-33, 120-122, 136, 
142 



214 



Index 



Water management, 10, 17,74,76,80,99, 114, 
118, 121-123, 134, 137, 142, 160, 174, 
183, 186-188, 191-192, 196-200, 202, 
204,206 

Water storage, 69, 100, 104-105, 107, 109, 
111-114, 198 



What-if approach, 5, 146-147, 192-193 
Willow, 98-105, 108-114 

Z 

Zoning, 11,25,86,99, 120, 128, 139, 147-148, 
155,183