Journal of Hydrology: Regional Studies 4 (2015) 502–515

Contents lists available at ScienceDirect

Journal of Hydrology: Regional

Studies
journal homepage: www.elsevier.com/locate/ejrh

Impacts of climate change on hydrological regime and water

resources management of the Koshi River Basin, Nepal
Laxmi Prasad Devkota ∗ , Dhiraj Raj Gyawali
Nepal Development Research Institute, Shree Durbar Tole, Lalitpur, Nepal

a r t i c l e i n f o a b s t r a c t

Article history: Study region: The middle hilly region of the Koshi River Basin in Nepal.
Received 28 September 2014 Study focus: Assessment is made of the hydrological regime of the basin under climate
Received in revised form 10 May 2015 change. Results from two Regional Climate Models (PRECIS-HADCM3Q0 and PRECIS-
Accepted 8 June 2015
ECHAM05), based on IPCC-SRES A1B scenario, were bias corrected against historical gauged
Available online 2 September 2015
data. Hydrological impact simulations were conducted using SWAT model. Design flood
estimation was done after extreme value analysis based on annual flow maxima.
Keywords:
New hydrological insights for the region: The study found that climate change does not pose
Koshi Basin
major threat on average water availability. However, temporal flow variations are expected
SWAT modeling
Climate change to increase in the future. The magnitude of projected flow for given return periods, however,
Design standard strongly depends on the climate model run considered. The ECHAM05 results show higher
Design values flow changes than those estimated from the HADCM3 outputs. A relation was derived to
Uncertainties estimate projected flood flow as a function of return period and flow estimated from his-
torical series. Amidst the uncertainties, these predictions provide reasonable insight for
re-consideration of design standards or design values of hydraulic structures under climate
change.
© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

The Koshi River is a trans-boundary river originating in China and passes through Nepal and finally meets the Ganga-River
in India. The total catchment area of the river at its confluence with Ganga is 74,030 km2 (FMIS, 2012). About 43% of this total
area lies in China, 42% in Nepal and the remaining 15% in India (Sharma, 1997; FMIS, 2012). The Koshi River drains most of
the eastern part of Nepal.
Livelihood of hundreds of millions of people of Nepal and India is dependent on the water availability in the Koshi River
and, at the same time, frequent floods and drought have rendered millions vulnerable to their impacts (Chen et al., 2013).
With the changing development trajectories of these countries, rapid development that includes urbanization, construction
of roads and buildings, industrial development have taken place in recent years in the lower part of the basin. These activities
are going to continue in the future as well. As in other parts of the world, long term water resources management is, thus,
becoming a key driver for sustainable development of this region to meet the growing water demand for domestic, irrigation,
fisheries and industrial uses, and mitigating the distressing impact of the water induced disasters, mainly floods.

∗ Corresponding author at: Nepal Development Research Institute, Shree Durbar Tole, Lalitpur, GPO Box 8975, EPC 2201 Kathmandu, Nepal.
E-mail addresses: lpdevkota@ndri.org.np, lpdevkota@yahoo.com (L.P. Devkota).

http://dx.doi.org/10.1016/j.ejrh.2015.06.023
2214-5818/© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515 503

The world climate is changing more rapidly in recent years (Vijaya et al., 2011). Based on GCMs and RCMs, ISET-N (2009)
reported that the mean annual temperature of Nepal is expected to increase by 1.4 ◦ C by 2030, 2.8 ◦ C by 2060, and 4.7 ◦ C by
2090. The average annual mean of maximum temperature was predicted to increase by 2.1 ◦ C under A2 scenario and by 1.5 ◦ C
under B2 scenario in 2080s in the Bagmati River Basin (Babel et al., 2014) while Bharati et al. (2012) projected that it will
increase by 0.86 ◦ C under A2 and by 0.79 ◦ C under B1 scenarios in 2030s for the Koshi Basin compared to baseline 1976–2005.
Increased temperature leads to greater evaporation and thus surface drying which thereby increases the intensity and
duration of drought. However, the water holding capacity of air increases by about 7% per 1 ◦ C warming which leads to
increased water vapor in the atmosphere (Trenberth, 2011; Agarwal et al., 2014). It consequently, produces more frequent
and intense precipitation events (Gurung and Bhandari, 2009). Agarwal et al. (2014), in their study of future precipitation in
Koshi River Basin of Nepal, stated that projected precipitation under A1B scenario for the future periods (2020s, 2055s and
2090s) is expected to increase. Frequent and intense extreme climate and weather events will lead to increasing climatic
variability that significantly increases the intra-annual variability of stream flow (Agrawala et al., 2003; Chaulagain, 2006).
Climate change is, therefore, making water resources management further challenging adding additional uncertainty on
the hydrological conditions. Assessment of the long term river hydrology with climate change impact on it has become
imperative in recent years for long term sustainable water resources management (APN, 2008; Gosain et al., 2006; Chiew
et al., 2011; Obeysekera et al., 2010; Shivakumar, 2010 and WECS, 2011) because it helps to build resilience to the possible
impact of climate change through enhanced institutional flexibility and consideration of climate related risks in the planning
process (Sharma and Shakya, 2005).
Developing countries are more vulnerable to extreme weather events under present day climatic variability which causes
substantial economic damage (Monirul and Mirza, 2003). For the past thirty years, the number of flood disasters has increased
compared to other forms of disasters in South Asia (Dutta and Herath, 2005) taking a heavy toll on properties and lives and
caused more economic losses including infrastructure such as drinking water as well as water treatment plant, agricultural
area degradation than any other hazards in Nepal (MoF, 2011). The recurrent devastating flood events in the Koshi River
are quite enough to substantiate those observations and findings which have claimed huge damage in property, agriculture
and loss of lives in the Terai plain of Nepal and Bihar of India at different time moments, e.g. 1869, 1870, 1954, 1968, 1971,
1980, 1984, 1991 (Mishra, 2008). The recent devastating one was the flood event of August 2008. This flood event caused the
death of 4–6 persons, affected 6000 ha of agricultural land, damaging crops worth USD 3.7 million, displaced 40,378 people
from 7102 families and damaged 4 km of the east-west highway in Nepal (ICIMOD, 2008). The devastation was much bigger
in India. It claimed 42 lives (as of 25 August 2008), destroying 35,000 ha of crop, affected more than one million people with
more than 70,000 being displaced (ICIMOD, 2008).
Due to the frequent flooding in Koshi River and its damages, there have been some efforts to control flood in both India
and Nepal with the Koshi River Agreement in 1954. However, Koshi Flood in 2008 has alerted both countries regarding the
functioning of the existing Koshi barrage and the river embankments. As a result, Koshi High Dam Project, which was envi-
sioned in 1937 to effectively “cure” the flooding problem, has once again come forward as a revitalized issue for flood control,
irrigation and hydropower generation. Koshi High Dam, the concrete dam of 269m height, is proposed to be constructed
in Barahkshetra (near Chatara) with live storage of 4420 million cubic meters (mcm) and gross storage of 8500 mcm. The
expected benefit from the dam is irrigation of 66,450 ha of land in Nepal and millions of hectares in India, flood control and
3489 MW of hydropower (DoED, 2013). As such, Nepal (the upper riparian), India (middle riparian), and even Bangladesh
(lower riparian of the Ganga River Basin) have shown interest in the construction of the high dam. Recently, the sixth meeting
of the India Nepal Joint Ministerial Commission on Water Resources (JCWR), held in November 2011 at New Delhi, agreed
to expedite the completion of a Detailed Project Report for the Koshi High Dam Multipurpose Project. Similarly, the meeting
of the JCWR held in January 2013 at Kathmandu decided to continue the feasibility study of the Dam (MoEn, 2014).
The beneficial aspects of the Koshi High Dam Project are very promising, but the project is not that simple as it seems.
Apart from the existent socio-political issues, the proposed Koshi High Dam Project is going to face the challenges and uncer-
tainties on its hydrological regime brought upon by the changing climate. Similarly, adequacy of the existing embankment to
contain future floods, taking climate change into account, is yet to be assessed. The assumption of stationarity in traditional
hydrologic considerations may no longer be valid under climate change (Milly et al., 2008) or indeed under natural variability
(Frank and Kuczera, 2002) as cited in Chiew et al. (2011) while discussing on the implications of climate change for water
resources management. Design of hydraulic structures such as the Koshi High Dam or embankments based on the analysis
of historical hydrological data is, thus, simply not a good design considering the non-stationarity of the hydrological time
series. Hennrgriff et al. (2006) reported in their study in the Baden-Wurtemberg area, that the ratio of design runoff under
climate change and historical conditions ranged from 1.75 to 1.00. It was found to be decreasing with increasing return
periods. For the 1000-year return period, the ratio was obtained to be 1.00, referring to the null effect of climate change.
However, Das and Simonovic (2012) conducted a study to assess the impact of climate change to assess in the upper Thames
River Basin of Canada using the results of fifteen different climate models for the future conditions in 2020, 2050 and 2080.
They stated that the 100-year and 250-year flood magnitudes were found to be respectively 12% and 32% higher than for
the baseline period (1979–2005) with negligible impact for the 10-year return period. It indicates that the magnitudes of
flood flow increase under climate change impact, with higher changes for higher return periods.
It is thus prudent to predict the changes in the hydrological cycle including the extreme events under climate change
scenario using appropriate hydrological models. Such models need to be robust enough to capture the hydrological phe-
nomena with sufficient spatial detail. The results from such analysis will aid in making wise and rational decisions on the
504 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515

Fig. 1. Koshi River Basin in Nepal.

hard and soft approaches to face and overcome the challenges of the water resources management activities in the basin in
the context of climate change.
This study assesses the changes in the hydrological regime of the Koshi River Basin with and without climate change using
the Soil and Water Assessment Tool (SWAT model). This model has been widely used to assess the climate change impacts
on the hydrological regime of various catchments across the world (Bharati et al., 2012; Gurung et al., 2013; Narsimlu et al.,
2013; Lubini and Adamowski, 2013; Singkran et al., 2014; Bossa et al., 2014; Fontaine et al., 2002). Considering the projected
data availability, for this purpose the results from two regional climate models based on the IPCC SRES greenhouse gas
emission scenario A1B were used to assess the future climatic conditions (IPCC, 2007).

2. Study area: The Koshi River Basin

The Koshi River has three main tributaries viz. Tamor in the eastern part, Arun in the middle and Sunkoshi in the western
part of the basin. The Sunkoshi River consists of 5 major tributaries: the Indrawati, the Bhote-Koshi, the Tama Koshi, the
Likhu and the Dudh Koshi (Fig. 1). The Sun-koshi, the Arun and the Tamor meet at Tribenighat, flow through Barahkshetra
gorge for a length of about 15 kms and enters into the Terai Region (plains) of Nepal after Chatara.
The Koshi River Basin covers three major ecological zones of Nepal with a transverse length (north-south) of about
150 km. These zones are: (i) Snow covered Himalaya in the north, (ii) hilly region in the middle and (iii) plain region of Terai
in the south. The variation of altitude in this short north–south reach is quite sharp ranging from 95 m to 8848 m. The High
Himalayan region of the Koshi basin within Nepal is about 8220 km2 (>3000 m) where glacial lakes are common. ICIMOD
(2011) mapped 599 glacial lakes in the Koshi Basin covering an area of 26 km2 .
The upstream Himalaya part of the Koshi Basin covers an area of about 17,620 km2 , mainly covered with forests and
agricultural land. This region is the high rainfall receiving zone of the basin. The downstream part in the Terai region of
Nepal covers an area of 2000 km2 before it enters into Indian Territory. The Terai region is highly populated. Agricultural
land occupies the main part of this plain. This study focused on the middle, hilly part of the basin. Because this region is a
high rainfall receiving zone, rainfall-runoff (hydrologic simulation) study needs to be carried out to assess the impact of the
climate change on the hydrology. The SWAT model was employed to assess the impact on the basin hydrology, and the Snow
Runoff Model (SRM) for quantifying the boundary flows from the Himalayan region. Given that snow and glacial hydrology
are dominant in the Himalayan region, SRM modeling was performed to comprehend the snowmelt phenomenon dividing
the Himalayan catchment into five sub-catchments; see Khadka et al. (2014) for details.
 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515 505

Fig. 2. Digital Elevation Model (DEM), landuse and soil maps of the Koshi River Basin.

3. Materials and methods

3.1. Hydrologic simulation

A semi-distributed, time continuous watershed model, Soil and Water Assessment Tool (SWAT) (Arnold et al., 1998; SWAT,
2005) was used for rainfall-runoff modeling in this study. Conceptually, SWAT divides a watershed into sub-watersheds.
Each sub-watershed is connected through a stream channel and further discretized into Hydrologic Response Units (HRUs). A
HRU is a unique combination of soil and vegetation type in a sub-watershed, and SWAT simulates the hydrological variables
and runoff results at the HRU level and aggregates these results to the catchment scale by applying a weighted average to
the HRU results. The runoff is routed to obtain the total runoff for the watershed at the outlet.
The SWAT model for Koshi Basin was set up with the Arc SWAT2009 interface (Neitsch et al., 2011). The data required
for the model implementation and the steps involved are briefly discussed below. Coefficient of Determination (R2 ) and
Nash–Sutcliffe Simulation Efficiency (ENS ) were used as the goodness of fit measures during calibration and validation of the
model.

3.2. Data used in simulation

3.2.1. Topographic, land use and soil data

Topographic, land use and soil map data are the spatial data required for hydrological simulations using the SWAT model.
The topography of the basin is defined by the Aster GDEM v2.01 Digital Elevation Model (DEM) for the Koshi River Basin. It
has a spatial resolution of 30 m. Fig. 2 shows this DEM.
Landuse data for the study area were obtained from the Department of Survey, Nepal (SD, 1996). Reclassification of the
landuse map was done and the respective model parameters were selected from the SWAT database corresponding to the
specific landuse/cover types. Fig. 2 shows the landuse map for the Koshi River Basin. The FAO soil map2 was applied and the
soil data properties obtained from the FAO soil properties database (FAO, 2002). Refer Fig. 2.

3.2.2. Historical time series data

Historical meteorological and hydrological data were collected from the Department of Hydrology and Meteorology
(DHM), Government of Nepal (DHM, 2012). The meteorological data used for the SWAT modeling were daily precipitation,
daily maximum and minimum temperature, daily relative humidity, wind speed and sunshine hours. Gauged flow data were
used in the model for two purposes; first to define inlet discharge points for the basin to simulate the existing conditions
and second for performing sensitivity analysis, calibration and validation of the model at the outlet. Since the SWAT model
was used for the hilly part of the basin, inlet discharge points were defined in the model at five points (Fig. 3) to incorporate
the flow generated from the mountainous/snow-covered part of the basin.

3.2.3. Projected future climate data

Simulation results for the region from two Regional Climate Models (RCMs), PRECIS-HADCM3Q0 and PRECIS-ECHAM05,
were used in this study to generate the future flows. They both are based on the IPCC greenhouse gasses emission scenario
(SRES) A1B. Detailed discussion of these models is provided in ADPC et al. (2012). The RCM outputs were considered for the
baseline period 1976–2000 and for the future period 2040–2060. The two RCM models are hereafter denoted as ECHAM05
and HADCM3. Spatial resolution of the data is 25 km. Bias correction of the climate data (rainfall and mean, maximum
and minimum temperature) was done against gauged historical data. Recently, several studies focused on methods for
bias correction of climate model outputs prior to use in hydrological impact studies. In comparison with the basic method
of bias correcting the mean, more advanced methods are recommended bias correcting both the mean and the variance,
and – for rainfall – bias correcting a range of quantile values versus probability (Sunyer et al., 2014). Methods have been

1
AsterGDEMv2.0 is a property of METI and NASA.
2
http://www.tucson.ars.ag.gov/agwa/index.php?option=com jdownloads&Itemid=&view=viewdownload&catid=17&cid=113.
506 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515

Fig. 3. Delineation of the catchment area.

developed to conduct this quantile based bias correction in a non-parametric (Gudmundsson et al., 2012; Ntegeka et al.,
2014) or parametric way (Rana et al., 2014; Terink et al., 2010; Leander and Buishand, 2007). In order to capture the day-
to-day climatic variability, the parametric Power Transform method for precipitation and the mean and standard deviation
adjustment for temperature were opted in this study following their impressive performance in the Meuse river basin
(Leander and Buishand, 2007) and the Rhine river basin (Terink et al., 2010). Further, Mean Bias Error (MBE) and Root Mean
Square Error (RMSE) of uncorrected and bias corrected versus observed values were considered for spatial evaluation of
bias correction, as opted by Terink et al. (2010). For temporal bias, coefficient of determination (R2 ) has been considered to
evaluate the performance of the bias correction. Monthly precipitation values for 1976–2000 in each region were considered
for evaluating the statistics. Significant improvement on R2 values for both hilly and mountain regions have been obtained
between observed and bias corrected precipitation data [R2 = 0.26 to 0.61 for HADCM3 and 0.34 to 0.62 for ECHAM for SWAT
domain; similarly R2 = 0.38 to 0.73 for HADCM3 and 0.36 to 0.72 for ECHAM for SRM domain]. MBE and RMSE were also
improved significantly for both models and regions.

3.3. Flow simulation

3.3.1. Model setup

To account for the flow contribution from the mountainous part of the basin, the region was divided into five sub-basins.
Outlets of these sub-basins were defined in SWAT as inlet points, viz. Majhitar for Tamor sub-basin, Uwagaon for Arun
sub-basin, Rabuwa Bazar for Dudh Koshi sub-basin, Busti for Tama Koshi sub-basin and Pachuwarghat for both Indrawati
and Sunkoshi-Bhotekoshi sub-basin. It is noted here that the Likhu river sub-basin has an insignificant snow covered area
and thus was included in the SWAT domain. The sub basins with the river reaches, inlet and outlet points are shown in Fig. 3.
Five slope classes (0–5%, 5–10%, 10–20%, 20–50% and >50%) were applied and the slope grids were reclassified based on
the DEM. These were combined with the landuse and soil data to define the HRUs to each sub-basin.
Daily data from 16 rainfall stations, 12 temperature stations and 3 synoptic stations (daily wind speed, relative humidity
and solar radiation data) were used for the period of 1976–2008 (Fig. 3). Missing temperature data for various stations
were filled using an elevation correction approach with the data from a reference station and monthly lapse rate calculated
comparing with the reference stations.
 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515 507

Table 1
Calibrated parameters and the values.

Parameters Description Bound Calibration results

Lower Upper Fitting value Method

CN2.mgt Initial SCS CNII value −25 25 1.05 (Forest, Multiply

Cultivated,
Rangeland, Barren
Land)
Sol Awc.sol Available water capacity (mmH2 0/mm soil) −25 25 0.75 Multiply
Canmx.hru Maximum Canopy Storage, (mm H2 0) 0 10 2.5 (Forest), Replace
1(Cultivation)
Gw Delay.gw Groundwater Delay (days) 0 500 8 Replace
Ch K2.rte Hydraulic conductivity in main channel, mm/hr 0 150 55 Replace
CH N.rte Manning’s Roughness in main canal 0 1 0.035 Replace
SURLAG.bsn Surface runoff lag time (days) 1 24 1 Replace
SFTMP.bsn Snowfall temperature, ◦ C −5 5 1.5 Replace
SMTMP.bsn Snowmelt base temperature, ◦ C −5 5 3 Replace
SMFMX.bsn Melt factor for snow on June 21 (mm H2 0/◦ C/day) 0 10 5.6 Replace
SMFMN.bsn Melt factor for snow on December 21 (mm H2 0/◦ C/day) 0 10 1.5 Replace
TIMP.bsn Snowpack temperature lag factor 0 1 0.51 Replace

3.3.2. Sensitivity analysis

Sensitivity analysis was carried out in order to evaluate the response of model output with reference to change in various
flow related parameters. A Latin Hypercube Sampling and One-at-a-time sensitivity analysis (LH-OAT) method was applied
using the observed flow. Top ten sensitive parameters affecting flow were then further used in calibrating the model.

4. Results and discussion

4.1. Flow characteristics

Flow in the Koshi River is the combination of snow melt runoff and rainfall. The observed daily flow data (from 1977 to
2008) at the outlet station Chatara (catchment area of 54,100 km2 ; DHM, 2008) showed that it carries a flow of 1500 m3 s−1
on average. However, the inter-annual variation is quite high, ranging from 618 m3 /s to 2055 m3 /s. The monthly minimum
and maximum flows for this period were 231 and 6180 m3 /s respectively. The daily minimum and maximum flow values
recorded during this period are respectively 201 and 11,900 m3 /s, with the coefficient of variation greater than one. It shows
that daily fluctuation of flow in the dry season is quite low while it is very high during monsoon season as depicted in Fig. 8. It
is because the monsoon flow is mainly governed by rainfall which is a common characteristic in all rivers of Nepal (Sharma,
1997; Thapa and Pradhan, 1995). An instantaneous maximum flow of 24,000 m3 /s was recorded in 1980. The instantaneous
flow for this period is, thus, more than double of the daily maximum. After comparison of the simulated flows for the
historical period with the observed daily flows, the average of the simulated daily flow was found only 2% lower than the
average observed flow. However, the daily minimum and maximum simulated flows were respectively 14% higher and 19%
lower than those of the observed flow. This implies that there is a slight overestimation of low flows and underestimation
of the high flows by the model.

4.2. Model results

4.2.1. Model calibration

The observed flow time series at the outlet station, Chatara, was divided into three distinct periods, the ‘warming-up’,
‘calibration’ and ‘validation’ periods, i.e. from 1986–1990, 1991–2000 and 2001–2006, respectively. Table 1 below shows
the parameters used for the calibration with the final calibrated values.
Observed versus simulated hydrographs are shown in Fig. 4. In Fig. 5 cumulative volume of flows for the simulation period
is shown. From these graphs it can be clearly seen that the model simulates the flow very well and the hydrographs are in
good agreement with the rainfall pattern. The high values of the Nash–Sutcliffe efficiency (ENS = 0.97) and the coefficient of
determination ( R2 = 0.98) and the small volume difference between the observed and simulated values of 0.21%, show a very
strong predictive capability of the model.

4.2.2. Validation
Figs. 6 and 7 show the comparison between observed and simulated flows for the validation period. Both graphs show
that the simulated values are again in good agreement with the observed values. This is justified by the high ENS of 0.87 and
R2 of 0.88, and the small volume difference of 6.5%. However, for the year 2003, the model underestimates the flow. This
underestimation may be due to erroneous data in the observed flow values.
508 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515

8000 0
7000
300
6000

Precipitaon, mm
Flows, m3/s
5000 600
4000
3000 900

2000
1200
1000
0 1500
Jan-91 May-92 Sep-93 Feb-95 Jun-96 Nov-97 Mar-99 Aug-00

Precipitaon Observed Flows Simulated Flows

Fig. 4. Observed vs. simulated hydrographs for calibration period.

1,80,000
Cumulave flow volume, m3/s

1,60,000
1,40,000
1,20,000
1,00,000
80,000
60,000
40,000
20,000
0
May-90 Jan-93 Oct-95 Jul-98 Apr-01
Observed Flows Simulated Flows

Fig. 5. Comparison of observed vs. simulated water volumes for calibration period.

8000 0
7000
300
6000

Precipitaon, mm
5000 600
Flows, m3/s

4000
3000 900

2000
1200
1000
0 1500
Jan-01 May-02 Sep-03 Feb-05 Jun-06
Precipitaon Observed Flows Simulated Flows

Fig. 6. Observed vs. simulated hydrographs for validation period.

120000

100000
Cumulave Flows, m3/s

80000

60000

40000

20000

0
Jan-01 May-02 Sep-03 Feb-05 Jun-06
Observed Flows Simulated Flows

Fig. 7. Comparison of observed vs. simulated volumes for validation period.

 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515 509

Fig. 8. Daily and monthly averaged flows at Chatara (DHM, 2008).

4.2.3. Future flow simulation

After simulation of the bias corrected ECHAM05 and HADCM3 daily results in the SWAT model, the climate change
impacts on design floods were assessed.
These design values are of importance while constructing hydraulic structures like dams, embankments or bridges over
rivers. They were derived from the annual maximum values based on the following steps:

i. Frequency analysis of annual instantaneous maximum flow recorded at the gauging station (Chatara) for different return
periods (QT,i ), where a Log-Pearson III distribution was calibrated to the observed flow.
ii. Similarly, frequency analysis of annual maximum daily flows of the historic data for different return periods (QT,d ).
iii. A ratio was then obtained by dividing the maximum instantaneous and daily flows for each return period T, i.e.
QT,i
RT = (1)
QT,d

iv. Frequency analysis of annual maximum daily flows of the projected period for different return periods (QT,d  ).

v. Calculation of instantaneous maximum flows (QT,i ), also called design floods, by multiplying the maximum daily flows
derived from the projected series for different return periods with ratio obtained from step (iv) for the respective return
periods, i.e.
 
QT,i = QT,d × RT (2)

4.3. Comparison of flow statistics of historical and projected flow series

To consider a 20-year period for both the historical and future series, the historical (observed) flows were limited to
the period 1987–2006. The period 2040–2060 was considered for the future (RCM based) flows. The 20-years period was
further divided into 2 parts of 10 years each to assess the changes within the periods. Note that model based future flows
were compared with the historical observed data in this study. The flow statistics of historical and projected flow series
are presented in Table 2. The data show that the estimated average future flow is slightly less than the baseline flow of
1987–1996, except for the ECHAM05 results and the 2050s. From the water availability point of view, climate change is
not going to impact greatly the basin water resources. However, both daily and monthly variations in flow are quite high.
It demands for strong regulation of storage facilities such as the Koshi High Dam, in order to fulfill the downstream water
requirements for irrigation, domestic and industrial uses, to generate stipulated hydropower and for flood control. Quite
low value of the safe yield (i.e. Q95 : probability of exceeding the given flow by 95% of the time) and very high values of flood
flows (Q1 and Q0.1 ) justifies the need of such water storage.

4.3.1. Change in annual flow

Comparison of the annual average flows between the historical data and future flows, show a slightly decreasing trend
of about 15 m3 /s per year (Fig. 9). Sharma et al. (2000) also found decreasing flow trends for the Koshi River when analyzing
the flow data of 1974–1994. Similar observations were made for other major Nepalese rivers such as Narayani, Karnali,
Kaligandaki (WECS, 2011). However, the increasing trend on the annual average flow of 17 m3 /s (25 m3 /s by ECHAM05 and
10 m3 /s by HADCM3) is clearly seen in both projected cases (ECHAM05 and HADCM3).
510 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515

Table 2
Comparison of flow statistics of historical and projected flow series, Unit: m3 /s.

Variables Observed Projected ECHAM05 Projected HADCM3

1987–1996 (baseline) 1997–2006 2041–2050 2051–2060 2041–2050 2051–2060

Qavg 1654 1471 1429 1717 1448 1500

CV 0.97 1.07 1.15 1.29 1.17 1.2
Qavg + Sdv ( + ) 3257
Min daily 255 201 219 247 230 245
Max daily 9610 9480 23,780 36,020 15,000 19,830
Min monthly 299 231 257 256 237 261
Max monthly 6089 6181 5620 8614 4673 6844
Annual min 1354 1251 1135 1247 1280 1296
Annual max 1795 2055 1713 2088 1757 1881
Q90 378 315 289 290 272 284
Q95 350 294 275 268 256 265
Q1 6230 6470 7625 9752 8443 8148
Q0.1 7730 8300 14,530 26,450 12,700 15,760

Fig. 9. Future changes in annual flows.

Table 3
Future change in monthly flows.

Months Observed ECHAM05 HADCM3

1987–1996 1997–2006 2041–2050 2051–2060 2041–2050 2051–2060

(Baseline flow, m3 /s) (%) (%) (%) (%) (%)

Jan 442 −18 −17 −10 −21 −9

Feb 379 −16 −18 −14 −13 −16
Mar 390 −19 −23 −11 −23 −27
Apr 455 −12 −23 −26 −34 −11
May 876 −22 −22 −13 −31 −23
Jun 2024 −13 19 19 5 10
Jul 3928 −2 −7 26 −7 −8
Aug 4740 −8 −26 −14 −23 −13
Sep 3420 −13 −14 4 −15 −16
Oct 1644 −16 −18 14 8 3
Nov 874 −21 −19 6 −8 −9
Dec 572 −17 −13 −1 −13 −13

4.3.2. Change in monthly flow

Development of water storage systems can be the strategies for climate change adaptation (ICIMOD, 2009). However,
possible water withdrawals from the river under no storage condition or the storage reservoir capacity and/or operation of
the reservoir depend on the monthly availability of the flow and its variation. To evaluate this aspect of flow, the percentage
change in the decadal average of monthly flows for the projected periods versus the baseline average monthly values were
calculated and are depicted in Table 3. The changes in monthly flows vary from decreases with 30% to increases by more
than 27%. Most of the flow is decreasing during the lean period and increasing in June and July, which are in the period
of high flows. This means that unfavorable water withdrawal conditions are more likely in the future. It demands higher
volumes of water to be stored to meet the dry season water demand if water storage projects such as the Koshi High Dam
are constructed. Changes in the operation rules moreover become a necessity to deal with these changes in monthly flows
ultimately.
 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515 511

6000
5000

Flow, m3/s
4000
3000
2000
1000
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Observed 1987-96 Observed 1997-06 ECHAM05 2041-50

ECHAM05 2051-60 HADCM3 2041-50 HADCM3 2051-60

Fig. 10. Monthly averaged flows for historical and future conditions.

Table 4
Frequency of highest monthly flows.

Period June July August September

Observed 1987–1996 0 0 10 0
1997–2006 0 2 8 0
ECHAM05 2041–2050 4 2 1 3
2051–2060 3 4 0 3
HADCM3 2041–2050 3 0 5 2
2051–2060 2 2 3 3

4.3.3. Early shifting of peak flow

Changes in the timing of water flows or water availability in the river are of special importance for the water management.
Temporal shifting of monthly peak flows from July to August was observed in the Bagmati River Basin by Sharma and Shakya
(2005), while Shrestha (2004) found early shifting of peak flows along the Kaligandaki River in western Nepal. To assess if
there is any shift in timing of river flows, 10 year average monthly flow hydrographs for the historical and future periods are
plotted in Fig. 10. The figure suggests a clear shift of peak monthly flows from August under existing conditions (baseline)
to July for future projected conditions (ECHAM05). However, the shift in peak monthly flows is not distinct for the HADCM3
model. As an attempt to get more clarity, frequencies of highest monthly flows are presented in Table 4. The table vividly
shows that the maximum monthly flows in the Koshi River have been taking place in August for the baseline period. For the
future scenarios, the number of cases of maximum flows occurring in June and July show that early shifting of the hydrograph
peaks is likely.

4.3.4. Flows with different exceedance probabilities

Important values of flow duration curves, i.e., flows with 0.1%, 1%, 90% and 95% probability of exceedence for both observed
and projected flows are listed in Table 2 for different decades. The table shows a significant increase in flows for 1% and
0.1% exceedance probabilities for the 2050s. They are as high as 57% (ECHAM05) and 31% (HADCM3) for the 1% exceedance
flow and as high as three and half times (ECHAM05) and two times (HADCM3) for the 0.1% exceedence flow. However the
90% and 95% exceedance flows are decreasing by 25% as compared with the historical case for the same period. This result
indicates a higher probability for extreme flows to increase and low flows, hence water availability, to decrease during the
lean seasons in the future.

4.3.5. Flood frequencies and volumes

Frequency and volumes of peak flows and floods are essential information for the design of a dam with the objective of
flood control. Before we can analyze such flood frequencies and volumes, we first need to define floods. A flood is in this
study considered as any daily flow event for which the peak flow is higher than the long term mean daily flow plus one
times the standard deviation ( + ) based on the baseline period (1987–1996). This flow equals 3257 m3 /s (Table 2). This
definition is employed in previous studies by Sharma and Shakya (2005) and WECS (2011). Calculated number of flood days,
maximum number of consecutive flood days and corresponding flood volumes are for both the historical and future periods
given in Table 5. It shows that the number of consecutive flood days is decreasing in the future. This obviously is due to
the more intense rainfall. However, the number of such rainy days producing floods will be less. It moreover is noted that
the number of consecutive flood days and the flood volumes is higher in the 2050s than in the 2040s. When the average of
the flood volumes obtained from the ECHAM05 and HADCM3 results is considered to compute the future flood volume to
be kept in the flood storage zone of the reservoir, an average maximum flood volume of about 2400 million cubic meter is
obtained. The flood storage zone to be kept in the reservoir to accommodate such future flood should be three times more
than the flood zone capacity estimated from the baseline data.
512 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515

Table 5
Flood information of historical and projected scenarios.

Flood Volumes Observed Projected ECHAM05 Projected HADCM3

1987–1996 1997–2006 2041–2050 2051–2060 2041–2050 2051–2060

No. of days 685 567 415 604 435 488

Exceeding
( + ),
3257 m3 /s
Maximum No. of 71 104 34 69 26 48
consecutive days
exceeding
( + ),
3257 m3 /s
Maximum Volume 830 819 2055 3112 1296 1713
(in 106 m3 )

Table 6
Return period of flows for historical and projected conditions.

Return period (year) Flow (m3 /s)

Historical ECHAM05 HADCM3 Projected average

2 7246 11,878 11,207 11,543

5 9464 19,184 16,601 17,893
10 11,579 26,349 21,388 23,869
20 14,197 35,617 27,147 31,382
50 18,666 52,381 36,773 44,577
100 23,041 69,758 46,042 57,900
500 38,034 134,648 77,088 105,868
1000 47,445 178,539 96,165 137,352
10,000 101,080 457,143 201,103 329,123

4.4. Impacts of climate change on design floods

The impacts of climate change on water resources development works does not only depend on changes in the flood
volumes but also on timing and magnitude of the peak flows. The magnitude of the floods is particularly important when
we deal with hydraulic structures like the spillway of the dam to dispose the flood water and the embankments to contain
the flood water within the river bed. Timing and duration of peak flows dictate the operation rules of the reservoir both
for water use and safe disposal of flood water. Therefore, design floods were estimated for various return periods with and
without climate change scenarios.
The estimated instantaneous floods for various return periods are for historical as well as future conditions given in
Table 6. Average values of the ECHAM05 and HADCM3 based floods were calculated. The following conclusions can be
drawn from the results in Table 6.

1. For a given return period, the magnitudes of the future design flows are for both models higher than that of the historical
ones. This shows higher future design flows. It demands higher capacity of the flow disposing structures such as the dam
spillway and of the river embankments.
2. The higher flood values point out the insufficiency to assume that the design flood values are stationary over time. In other
words, design standard should be revisited to consider the climate change in hydraulic infrastructure design, i.e. higher
flood flows of lower return periods should be considered than what we are using now. For example, to design a hydraulic
structure for a return period of 100 years (peak flow of 57,900 m3 /s, Table 6), the results after the future projections show
that we need to increase the design return period to more than 1000 years when historical design floods keep being
applied (47,445 m3 /s, Table 6). It can be observed from the table that the 10,000-year flood will become a 500-year flood
under the future conditions considered in this study. These results suggest extreme increases in the instantaneous river
flows.
3. The changes in design floods are different when they are based on one versus the other RCM. The design floods estimated
from the ECHAM05 results are higher than those estimated form the HADCM3 model. It shows the existence of uncer-
tainties in future estimations. This uncertainty should be taken into account when designing hydraulic structures. For
example, for the 100-year return period flood, the ECHAM05 estimate is almost 50% higher than the HADCM3 estimate.
4. The ratio of future flow over the historical one is higher for higher return periods. In other words, if we want to adjust the
design floods to accommodate to climate change, the adjustment depends on the return period considered.
 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515 513

Fig. 11. Ratio of future over historical instantaneous peak flows versus return period T.

To simplify the analysis, an attempt was made to develop a relationship by calibrating the ratio of the future over historical
peak flows versus the corresponding return periods (Fig. 11). The following relation was obtained:

Qp
= 1.592 + 0.188 ln(T ) (3)
Qh

where Qp is the future peak flow, Qh the historical peak flow and T the return period (years).
The future design flood for a given return period can, thus, be estimated based on that relationship as follows:

Qp = [1.592 + 0.188 ln(T )]Qh (4)

5. Conclusion

This study assessed the expected climate change impact on river hydrology of the Koshi River and its implication on the
proposed Koshi High Dam Project. Two Regional Climate Model future projection runs viz. PRECIS-HADCM3Q0 and PRECIS-
ECHAM05 were acquired and used for this purpose. Both model runs are based on the IPCC SRES A1B scenario. The run results
for daily rainfall, and maximum and minimum temperatures were bias corrected against observed historical data using the
power transformation method for rainfall and the mean and standard deviation adjustment method for temperature. The
objective functions used while performing bias correction were the Mean Bias Error (MBE) and Root Mean Square Error
(RMSE) for the spatial variations and the coefficient of determination (R2 ) for the temporal bias. Satisfactory improvement
on MBE, RMSE and R2 values have been obtained between observed and bias corrected data.
Bias corrected rainfall and temperature data were used in the calibrated and validated SWAT model to assess the future
hydrological regime of the basin. The study found that climate change is less likely to pose the threat on average water
availability in the Koshi River Basin. However, temporal variation in river flows is expected to increase in the future. Most of
the flow is decreasing during the lean season and increasing during the high flow season. The future RCM based projections
show a decrease in the long term monthly flow by more than 30% in the drier months and an increase by more than 25%
in the high flow months when compared to the baseline values. The results suggest a shift of the peak monthly flow from
August under baseline conditions to July under ECHAM05 based projected future conditions. However, the shift in peak
monthly flow is not obvious for HADCM3. The peak flows for 0.1% exceedance probability are as high as three and half times
the corresponding baseline value for ECHAM05 and two times for HADCM3.
Design flood estimation method was proposed to estimate the peak and flood flows for different return periods. The
method is based on extreme value analysis of the daily maximum flows. The magnitudes of the design flood values were
found to be higher than those based on the historical series. The flood magnitudes obtained from the ECHAM05 results are,
however, higher than those estimated from the HADCM3 results. A relation was derived to estimate future design flood
flows as a function of return period and corresponding historical design flood flow.
It was concluded that any design based on a 100-year design flood flow (57,900 m3 /s) may need to be changed to a design
flood flow for a return period of more than 1000 years, when based on historical data (47,445 m3 /s), in order to account for
the impact of climate change. The results in this study furthermore show that the 10,000-year return flood may occur on
average every 500 years in the future. Amidst the uncertainties, these projections provide reasonable insight in support of
alterations or re-consideration of design standards or design values of hydraulic structures if impact of climate change is to
be taken into consideration during hydraulic design of water resources works.
514 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515

Acknowledgements

The authors would like to thank Climate Change Knowledge Network (CDKN) and Global Change SysTem Analysis
Research and Training (START) for the financial support to carry out this research. They are also grateful to Dr. Divas B.
Basnyat to improve the quality of the paper.

References

ADPC, AidIQ, DHM, ADB, 2012. Technical Approach and Methodology for Projected Data Preparation, Nepal Climate Data Portal. Asian Disaster Preparedness
Center (ADPC), Department of Hydrology and Meteorology (DHM), AiDIQ and Asian Development Bank (ADB).
Agarwal, A., Babel, M.S., Maskey, S., 2014. Analysis of future precipitation in the Koshi River Basin, Nepal. J. Hydrol. 513, 422–434.
Agrawala, S., Raksakulthai, V., Aalst, M., Larsen, P., Smith, J., Reynolds, J., 2003. Development and Climate Change in Nepal: Focus on Water Resources and
Hydropower. Organization for Economic Cooperation and Development, Paris, pp. 64.
APN, 2008. Climate Change in Southeast Asia and Assessment on Impact, Vulnerability and Adaptation on Rice Production and Water Resources, Project
Reference No.: CRP 2008-03CMY-Jintrawet, Asia-Pacific Network for Global Change Research.
Arnold, J.G., Srinivasan, R., Muttiah, R.S., Williams, J.R., 1998. Large area hydrologic modeling and assessment. Part 1. Model development. J. Am. Water
Resourc. Assoc. 34, 73–89.
Babel, M.S., Bhusal, S.P., Wahid, M.S., Agarwal, A., 2014. Climate change and water resources in the Bagmati River Basin, Nepal. Theor. Appl. Climatol. 115,
639–654.
Bharati, L., Gurung, P., Jayakody, P., 2012. Hydrologic Characterization of the Koshi Basin and the Impact of Climate Change. Hydro Nepal, Special Issue.
Bossa, A.Y., Diekkuger, B., Agbossou, E.K., 2014. Scenario-based impacts of land use and climate change on land and water degradation from the meso to
regional scale. J. Water 6 (10), 3152–3181, http://dx.doi.org/10.3390/w6103152.
Chaulagain, N.P., (Thesis for the degree of doctor of engineer) 2006. Impacts of Climate Change on Water Resources of Nepal: The Physical and Sociological
Dimension. University of Flensburg, Germany.
Chen, N.S., Hu, G.S., Deng, W., Khanal, N., Zhu, Y.H., Han, D., 2013. On the water hazards in the trans-boundary Kosi River Basin. Nat. Hazards Earth Syst. Sci.
13, 795–808, http://dx.doi.org/10.5194/nhess-13-795-2013 http:\\www.nat-hazards-earth-syst-sci.net/13/795/2013/
Chiew, F.H.S., Young, W.J., Cai, W., 2011. Current drought and future hydro-climate projection in Southeast Australia and implication for water management.
Stoch. Environ. Res. Risk Assess. 25, 601–612.
Das, S., Simonovic, S.P., 2012. Assessment of uncertainty in flood flows under climate change impacts in the upper Thames River Basin, Canada. Br. J. Environ.
Clim. Change 2 (4), 314–338.
DHM, 2008. Stream flow Summary (1962–2006). Department of Hydrology and Meteorology (DHM), Kathmandu, Nepal.
DHM, 2012. Meteorological Network. Department of Hydrology and Meteorology (DHM), Kathmandu, Nepal, http://www.dhm.gov.np/
meteorological-station (accessed 01.08.12).
DoED, 2013. Koshi High Dam. Department of Electricity Development, Kathmandu, Nepal.
Dutta, D., Herath, A., 2005. Trend of Floods in Asia and Flood Risk Management with Integrated River Basin Approach, Human Security and Climate Change,
An International Workshop, Asker, near Oslo.
FAO, 2002. FAO/UNESCO Digital Soil Map of the World and Derived Soil Properties. Land and Water Digital Media Series #1 rev 1. FAO, Rome.
FMIS, 2012. Flood Management Information System. Water Resources Department, Bihar, http://fmis.bih.nic.in/Riverbasin.html#kosi (accessed December,
2013).
Fontaine, T.A., Cruickshank, T.S., Arnold, J.G., Hotchkiss, R.H., 2002. Development of a snowfall–snowmelt routine for mountainous terrain for the soil and
water assessment tool (SWAT). J. Hydrol. 262 (1–4), 209–223.
Frank, S.W., Kuczera, G., 2002. Flood frequency analysis: evidence and implication of secular climate variability, New South Wales. Water Resour. Res. 38,
1062, http://dx.doi.org/10.1029/2001WR000232.
Gosain, A.K., Rao, S., Basurya, D., 2006. Climate change impact assessment on hydrology of Indian River Basins, climate change and India. Curr. Sci. 90,
346–353.
Gudmundsson, L., Bremnes, J.B., Haugen, J.E., Engen-Skaugen, T., 2012. Downscaling RCM precipitation to the station scale using statistical transformations
– a comparison of methods. Hydrol. Earth Syst. Sci. 16 (9), 3383–3390.
Gurung, G.B., Bhandari, D., 2009. Integrated approach to climate change adaptation. J. For. Livelihood 8 (1), 90–98.
Gurung, P., Bharati, L., Karki, S., 2013. Application of the SWAT model to assess climate change impacts on water balances and crop yields in the West Seti
River Basin. In: Conference Proceedings. SWAT Conference 2013, Toulouse, France.
Hennrgriff, W., Kolokotrinis, V., Weber, H., Bartels, H., 2006. Climate Change and Floods – Findings and Adaptation Strategies for Flood Protection, KA-
Abwasser, Abfall, vol. 53, http://www.afp.com/en/node/2424331 (accessed 13.04.14).
ICIMOD, 2008. Koshi Flood Disaster. International Centre for Integrated Mountain Development (ICIMOD), Kathmandu, Nepal.
ICIMOD, 2009. Water Storage: A Strategy for Climate Change Adaptation in the Himalayas. ICIMOD, Kathmandu, Nepal,
http://apps.icimod.org/elibrary/uploads/tmp/icimod-water storage.pdf (accessed 30.05.14).
ICIMOD, 2011. Glacial Lakes and Glacial Lake Outburst Floods in Nepal. International Centre for Integrated Mountain Development (ICIMOD), Kathmandu,
Nepal.
IPCC, 2007. Climate Change 2007. Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change.
ISET(N), 2009. Vulnerability through the Eyes of Vulnerable: Climate Change Induced Uncertainties and Nepal’s Development Predicaments. Institute of
for Social and Environmental Transition – Nepal (ISET-N), Kathmandu, Nepal.
Khadka, A., Devkota, L.P., Kayastha, R.B., 2014. Snowmelt contribution in Koshi River flow under climate change. J. Hydrol. Reg. Stud.
Leander, R., Buishand, T., 2007. Resampling of regional climate model output for the simulation of extreme river flows. J. Hydrol. 332, 487–496,
http://dx.doi.org/10.1016/j.jhydrol.2006.08.0062007.
Lubini, A., Adamowski, J., 2013. Assessing the potential impacts of four climate change scenarios on the discharge of the Simiyu River, Tanzania using the
SWAT Model. Int. J. Water Sci. 2, http://dx.doi.org/10.5772/56453.
Milly, P.C.D., Betancourt, J., Falkenmark, M., 2008. Stationarity is dead: wither water management? Science 319, 573–574.
Mishra, D.K., 2008. Trapped! Between the Devil and Deep Waters, http://el.doccentre.info/eldoc1/0901/090123zzz1B.pdf (accessed 15.12.13).
2014. MoEn, http://www.moen.gov.np/press releases.php (accessed 05.05.14).
MoF, 2011. Economic Survey-Fiscal Year 2010/2011. Ministry of Finance, Kathmandu, Nepal.
Monirul, M., Mirza, Q., 2003. Climate change and extreme weather events: can developing countries adapt? Clim. Policy 3, 233–248.
Narsimlu, B., Gosain, A.K., Chahar, B.R., 2013. Assessment of future climate change impacts on water resources of upper Sind River Basin, India using SWAT
model. Water Resour. Manage. 27 (10), 3647–3662.
Neitsch, S.L., Arnold, J.G., Kiniry, J.R., Williams, J.R., 2011. Soil and Water Assessment Tool Theoretical Documentation Version 2009. Texas Water Resources
Institute Technical Report No. 406. Texas A&M University System, College Station, TX, pp. 77843–82118.
Ntegeka, V., Baguis, P., Roulin, E., Willems, P., 2014. Developing tailored climate change scenarios for hydrological impact assessments. J. Hydrol. 508C,
307–321.
 L.P. Devkota, D.R. Gyawali / Journal of Hydrology: Regional Studies 4 (2015) 502–515 515

Obeysekera, J., Irizarry, M., Park, J., Barnes, J., Dessalegne, T., 2010. Climate change and its implication for water resources management in South Florida.
Stoch. Environ. Res. Risk Assess. 25, 495–516.
Rana, A., Foster, K., Bosshard, T., Olsson, J., Bengtsson, L., 2014. Impact of climate change on rainfall over Mumbai using distribution-based scaling of global
climate model projections. J. Hydrol. Reg. Stud. 1, 107–128.
SD, 1996. Topographical Survey Sheets. Survey Department, Kathmandu, Nepal.
Sharma, C.K., 1997. A Treatise on Water Resources of Nepal. MASS Printing Press, Kathmandu, Nepal.
Sharma, K.P., Moore, B., Vorosmarty, C.J., 2000. Anthropogenic, Climatic and Hydrologic Trends in the Koshi Basin, Himalaya. Climate Change 47. Kluwer
Academy Publishers, Dordrecht, the Netherlands, pp. 117–139.
Sharma, R.H., Shakya, N.M., 2005. Hydrological change and its impact on water resources of Bagmati Watershed, Nepal. J. Hydrol. 327, 315–322.
Shivakumar, B., 2010. Global climate challenge and its impacts on water resources planning and management: assessment and challenges. South Environ.
Res. Assess. 25, 583–600.
Shrestha, K.L., 2004. Kaligandaki: Global Change Impact Assessment for Himalaya Region for Environment Management and Sustainable Development.
Progress Report for APN.
Sunyer, M.A., Hundecha, Y., Lawrence, D., Madsen, H., Willems, P., Martinkova, M., Vermoor, K., Bürger, G., Hanel, M., Kriaučiūnienė, J., Loukas, A., Osuch, M.,
Yücel, I., 2014. Inter-comparison of statistical downscaling methods for projection of extreme precipitation in Europe. Hydrol. Earth Syst. Sci. Discuss.
11, 6167–6214.
SWAT, 2005. User’s Manual Soil and Water Management Tool. Temple Texas, USA.
Singkran, N., Tosang, J., Waijaroen, D., Intharawichian, N., Vannarart, O., Anuntawong, P., Kunta, K., Wisetsopa, P., Tipvong, T., Janjirawuttikul, N., Masthawee,
F., Anornpatanawat, S., Kirtsaeng, S., 2014. Influences of land use and climate changes on hydrologic system in the Northeastern River Basin of Thailand.
J. Water Clim. Change, http://dx.doi.org/10.2166/wcc.2014.127.
Terink, W., Hurkmans, R.T.W.L., Torfs, P.J.J.F., Uijlenhoet, R., 2010. Evaluation of a bias correction method applied to downscaled precipitation and temper-
ature reanalysis data for the Rhine basin. Hydrol. Earth Syst. Sci. 14, 687–703.
Thapa, B.B., Pradhan, B.B., 1995. Water Resources Development: Nepalese Perspective. Institute for Integrated Development Studies, Kathmandu, Nepal.
Trenberth, K.E., 2011. Changes in precipitation with climate change. Clim. Res. 47, 123–138.
Vijaya, V.R.S., Iniyan, S., Goic, R., 2011. A review of climate change, mitigation and adaptation. Renew. Sustain. Energy Rev. 16 (1), 878–897.
WECS, 2011. Water Resources of Nepal in the Context of Climate Change. Water and Energy Commission Secretariat, Singha Durbar, Kathmandu, Nepal.