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