Environmental Challenges 8 (2022) 100579

Contents lists available at ScienceDirect

Environmental Challenges
journal homepage: www.elsevier.com/locate/envc

Hydrological modelling through SWAT over a Himalayan catchment using

high-resolution geospatial inputs
Sabyasachi Swain a,∗, S.K. Mishra a, Ashish Pandey a, A.C. Pandey b, Atul Jain c, S.K. Chauhan c,
Anil Kumar Badoni c
a
Department of Water Resources Development and Management, Indian Institute of Technology Roorkee, Roorkee, 247667, India
b
Irrigation Research Institute Roorkee, Roorkee, 247667, India
c
THDC India Limited, Rishikesh, 249201, India

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

Keywords: The concerns of water availability have been increasing due to population explosion, rapid industrialization,
SWAT and other anthropogenic activities. This urges for an accurate estimation of streamflow at the river basin scale,
Remote sensing which is typically carried out by the application of rainfall-runoff models. Though numerous models exist in
Tehri catchment
hydrology literature, their efficiency over the mountainous catchments is generally observed to be poor, mostly
Hydrological modelling
due to the lack of high-quality data over such regions. This study presents an application of the widely used soil
and water assessment tool (SWAT)-based hydrological modelling using high-resolution geospatial inputs over
the Tehri reservoir catchment (India) located in the lower Himalayan region. The Resourcesat-2 Linear Imaging
Self-Scanning System (LISS)-IV imageries of land use/land cover (LULC) and the Cartosat-1 digital elevation
model (DEM) are procured from National Remote Sensing Centre (NRSC), Indian Space Research Organization
(ISRO), India. The LULC and DEM have a spatial resolution of 5.8 m and 2.5 m, respectively. SWAT model is
applied for a duration of 12 years (2006-2017) with 2006, 2007-2013 and 2014-2017 as the warm-up, calibration
and validation periods, respectively. The results reveal an excellent performance of the model in streamflow
simulation. The efficacy measures viz., Nash–Sutcliffe efficiency (NSE) and Coefficient of determination (R2 ) are
obtained to be 0.83 and 0.84, respectively. The information reported in this study will be helpful to the water
resources engineers for hydrologic modelling over the mountainous catchments.

1. Introduction forests to farmlands or settlements leads to an increased surface runoff

and decreased groundwater recharge (Aragaw et al., 2021).
All living things rely on water as one of the most crucial natu- An accurate estimation of streamflow at the catchment scale is
ral resources. The hydrological cycle is a system of circulation that crucial. The increasing concerns of water availability due to popu-
connects the ocean, atmosphere, and land. There are different com- lation explosion, rapid industrialization, and other anthropogenic ac-
ponents of the water cycle and therefore, the water resource sys- tivities have further raised its importance (Himanshu et al., 2019;
tems are affected/altered by climate and land use/land cover changes Swain et al., 2022c). Numerous models exist in hydrology literature;
(Aragaw et al., 2021; Guptha et al., 2021, 2022; Palmate et al., 2021, however, physically-based rainfall-runoff models have been proven to
2022a, 2022b; Pandey and Palmate, 2018, 2019; Swain et al., 2020, be more successful in streamflow estimation studies because of their
2021a, 2021b, 2021c; 2022a, 2022b). The hydrologic processes (e.g., explicit representation of the runoff-generation strategy in terms of
infiltration, groundwater recharge, soil moisture storage, evapotranspi- the underlying physics (Pandey et al., 2016; Arnold and Fohrer, 2005;
ration loss, interception rates, etc.) are all affected by anthropogenic Dhami et al., 2018). Soil and water assessment tool (SWAT) is among
activities, resulting in changes in basin water resources (Swain et al., the most widely applied physics-based hydrological models. It is also
2022c, 2022d, 2022e, 2022f). These changes also affect the physio- a semi-distributed model as it runs on the hydrologic response unit
graphic aspects of the catchment (e.g., surface roughness, vegetation), (HRU) level. HRU is the smallest spatial unit of the model and the
which affects the volume and timing of surface runoff and groundwater standard approach of defining HRU lumps all similar soils, LULC, and
recharge, resulting in streamflow alterations. For example, transforming slopes within a sub-catchment (Aadhar et al., 2019; Himanshu et al.,
2017; Neitsch et al., 2011; Pandey et al., 2019, 2021; Swain 2017;

∗
Corresponding author.
E-mail address: [email protected] (S. Swain).

https://doi.org/10.1016/j.envc.2022.100579
Received 28 February 2022; Received in revised form 24 June 2022; Accepted 1 July 2022
2667-0100/© 2022 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)
S. Swain, S.K. Mishra, A. Pandey et al. Environmental Challenges 8 (2022) 100579

Fig. 1. Location details of the study area.

Swain et al., 2018). With the enhanced features and several in-built niques have gained substantial recognition for their ability to provide
hydrological processes, the SWAT model has been widely employed high-resolution physiographic information, which helps in improved as-
for different applications such as calibration, sensitivity, and/or uncer- sessment and monitoring of different physical features (Pandey et al.,
tainty analysis, land use/land cover (LULC) and climate change impact 2009; Singh and Pandey, 2021a, 2021b; Sood et al., 2021). Moreover, it
assessment, non-point source pollutant cycling/loss and transport study, is convenient to process and analyse the remotely-sensed datasets in the
snowmelt and/or glacier melt processes, hydropower projects, impacts contemporary computing systems. Remote sensing technique and the
on groundwater/surface water, crop growth/yield, evaluation of best satellite-retrieved datasets have been used in numerous fields of hydrol-
management practices, etc. (Abbaspour et al., 2015; Bressiani et al., ogy (Haque et al., 2020; Jain et al., 2001; Khan et al., 2020; Kumar and
2015; Gupta et al., 2020; Kim and Lee, 2010; Licciardello et al., 2011; Anbalagan 2015; Sarkar et al., 2020; Sindhu et al., 2015; Stehr et al.,
Omani et al., 2017; Pandey et al. 2016; Palmate and Pandey 2021; 2009). While global warming and climate change has increased the un-
Schuol et al., 2008; Sinnathamby et al., 2017; Swain 2017). However, certainty in estimating different hydro-climatic components, the high-
in this study, the focus is on applicability of the model for simulation of resolution remotely sensed information has played a crucial role in im-
streamflow. proving the representation and observation of hydrological processes,
The Himalayan region is a remarkable natural water tower as it which is well substantiated in literature (Guptha et al., 2021, 2022;
serves water to more than 20% of the world’s population. The Hi- Himanshu et al., 2018; Kumar et al., 2015).
malayan hills have a significant role in managing the climatology of Considering all the aforementioned issues, the objective of this study
India as they bring monsoon to the country and split the jet streams. is to assess the efficacy of the SWAT model for streamflow estimation
Hence, their role is vital for water resources planning and manage- over a Himalayan catchment using the high-resolution remotely-sensed
ment. However, climatic aberrations may have a considerable impact physiographic inputs. The details of study area, model description, re-
on the Himalayan region and are considered a threat to the sustainabil- sults and conclusions are presented in subsequent sections.
ity of their natural resources (Immerzeel et al. 2020; Swain et al. 2022b).
Therefore, it is important to set up a robust hydrological model over the
2. Study area
Indian Himalayan catchments to ensure effective addressal of the wa-
ter availability concerns. A well-calibrated SWAT model may be effec-
The Tehri catchment is located in the state of Uttarakhand, India.
tive to simulate streamflow over the Himalayan catchments; however,
The total catchment area is approximately 7300 km2 . The annual rain-
the availability of accurate input data is always a constraint, particu-
fall over the catchment normally varies in the range of 1500-2000 mm,
larly over the mountainous regions (Lu and Chiang 2019; Malik et al.,
whereas the normal minimum and maximum temperature over a year
2019; Rahman et al., 2013; Singh et al., 2018). While the constraints of
is 0°C and 36°C, respectively. Topography is of mixed type; however,
rainfall data can be circumvented by installation of more rain gauges,
being in the mountainous region, majority of the catchment lies under
collecting physiographic information (e.g., elevation, slope, land use)
steep slopes (Kumar and Anbalagan 2015; Swain et al. 2022b). The lo-
is a tedious task. In recent years, remote sensing and geospatial tech-
cation map of the study area is shown in Fig. 1. It can be observed that

2
S. Swain, S.K. Mishra, A. Pandey et al. Environmental Challenges 8 (2022) 100579

Fig. 2. Flowchart representing the overall framework of

SWAT-based modelling.

the catchment possesses a diverse variation in terms of elevation (715 The Cartosat data is available at 2.5 m resolution and hence, it per-
m to 6996 m above mean sea level). The upper portions of the catch- forms better than other global datasets of coarser resolution (Patel et al.,
ment mostly lie in very high elevation zones. The catchment is formed 2022). The Resourcesat 2 Linear Imaging Self-Scanning System (LISS)
by two rivers i.e., Bhagirathi and Bhilangna (Fig. 1), which drains into IV MX satellite imageries of LULC with a spatial resolution of 5.8 m
Tehri reservoir. In general, the anthropogenic influences are minimal are used in this study. Both Cartosat-1 DEM and LISS IV data are ob-
over the Himalayan catchments. Therefore, a large Himalayan catch- tained from National Remote Sensing Centre (NRSC), Indian Space Re-
ment like Tehri can be considered an ideal site for hydrological mod- search Organization (ISRO), India. The soil data is obtained from the
elling studies. National Bureau of Soil Survey and Land Use Planning (NBSS&LUP),
India.
3. Hydrological modelling through SWAT SWAT model requires meteorological inputs viz., precipitation, tem-
perature (maximum and minimum), wind speed, relative humidity,
3.1. Data used and solar radiation (Tripathi et al., 2004). Among these, precipitation
and temperature data are taken from India Meteorological Department
Rainfall-runoff modelling through SWAT requires physiographic and (IMD) gridded observation dataset available at a daily temporal scale.
hydro-meteorological inputs. The digital elevation model (DEM) is re- The spatial resolution of rainfall and temperature are 0.25° x 0.25°
quired for watershed/sub-watershed delineation. For developing hy- and 1° x 1°, respectively. The observed data for wind speed, relative
drologic response units (HRUs), three physiographic maps, i.e., land humidity and solar radiation are difficult to obtain in desirable accu-
use/land cover (LULC), slope (derived from DEM) and soil data, are re- racy. These are obtained from Climate Forecasting System Reanalysis
quired. For the present study, Cartosat-1 DEM was procured. The qual- (CFSR) dataset, which is a global reanalysis product provided by Na-
ity of this DEM in terms of accuracy has been studied by comparing tional Centre for Atmospheric Research, USA. This global weather data
with other globally available DEMs (Muralikrishnan et al., 2006, 2013). is also available at SWAT website (http://globalweather.tamu.edu/).

3
S. Swain, S.K. Mishra, A. Pandey et al. Environmental Challenges 8 (2022) 100579

Fig. 3. Slope map (extracted from DEM) of

Tehri catchment.

CFSR data is most often used for SWAT model applications, especially the modelling process will be more complex and time consuming. The
for wind speed, relative humidity and solar radiation, which are mostly Tehri catchment area is divided into 33 sub-catchments.
unavailable at desirable spatiotemporal resolution. Since precipitation In the HRU definition step, all the geospatial inputs (LULC, soil and
is the most prominent meteorological parameter for hydrological mod- slope maps) are reclassified in order to achieve compatibility with Arc-
elling, it is recommended to use the actual/observed precipitation data SWAT environment. After defining the thresholds for each input and
(Swain 2017). overlaying them, a total 445 HRUs are obtained in 33 sub-basins over
For model calibration and validation, the observed streamflow data the catchment. The outlet of the basin falls in the sub-basin 32 with a
is required for the entire simulation period. The 10-daily streamflow total contributing area of 7268 km2 . SWAT operates at HRU level and
data for twelve years (2006-2017) at the outlet of the catchment, which thus, is regarded as a semi-distributed model.
is also the inflow to the Tehri reservoir, was procured from Tehri Hydro Following the HRU analysis step, the meteorological data (from IMD
Development Corporation India Limited (THDCIL), Rishikesh. and CFSR) are provided into the model as inputs. As CFSR data format
is most compatible with SWAT model, the IMD rainfall and temperature
3.2. Steps of SWAT model simulation data are arranged with respect to them, i.e., rainfall and temperature
from the CFSR input tables are replaced by those from IMD. These mete-
The SWAT-based hydrological modelling involves the following orological inputs are written into the model and the database is updated.
steps: (1) Setup of SWAT project, (2) Watershed/sub-watershed delin- After providing all the meteorological inputs, the model is run and the
eation, (3) HRU analysis, (4) writing and editing input tables, (5) model output (i.e., streamflow at the sub-catchment outlets) is simulated.
run, (6) calibration and validation of model. The overall framework of In general, the default simulation of a model does not produce sat-
SWAT-based hydrological modelling is shown in Fig. 2. isfactory results; hence, calibration and validation of the model should
The SWAT model is operated in ArcGIS through the ArcSWAT in- be carried out with respect to the observed hydrologic data. Calibration
terface. The first step of SWAT project setup is to specify the directory aims to identify the parameters most influential on the hydrologic pro-
that includes all the necessary databases and folders to store the data. In cesses of the catchment and their optimal values to obtain the best pos-
the second step of catchment delineation, a threshold has to be provided sible model structure pertinent to the regional hydrological conditions.
for delineating sub-catchments. If the number of sub-catchments is high, This can be accomplished by SWAT-Calibration and Uncertainty Proce-

4
S. Swain, S.K. Mishra, A. Pandey et al. Environmental Challenges 8 (2022) 100579

Fig. 4. Classified Land use land cover map over Tehri

catchment.

dures (SWAT-CUP) software, which is used for the auto-calibration of Where, n is the total number of observations.
the SWAT-simulated flows (Abbaspour 2014). The calibration can also Similarly, R2 can be computed as,
be done manually by checking individual parameter’s influence on the 2
model and finding their optimum value by repetitive trials. However, ⎡ ∑𝑛 ( 𝑜𝑏𝑠 𝑜𝑏𝑠
)(
𝑠𝑖𝑚 − 𝑌 𝑠𝑖𝑚
) ⎤
⎢ 𝑖=1 𝑌 𝑖 − 𝑌 𝑌 𝑖 ⎥
this process is tedious and requires significant efforts. The SWAT-CUP 𝑅2 = ⎢ √ ⎥
)2 √∑ (
(2)
provides a wide range of options to the user, which circumvents the ⎢ ∑ ( )2 ⎥
⎢ 𝑛 𝑜𝑏𝑠 𝑜𝑏𝑠 𝑛 𝑠𝑖𝑚 𝑠𝑖𝑚 ⎥
issues associated with the manual calibration and provides more accu- ⎣ 𝑖=1 𝑌𝑖 − 𝑌 𝑖=1 𝑌𝑖 − 𝑌
⎦
rate solution in less time and efforts. SWAT-CUP uses different types of
algorithms, among which, sequential uncertainty fitting (SUFI2) tool is
4. Results and discussions
most widely used and the same is applied in this study. The model is
validated to substantiate its applicability over the region for future.
The objective of this study is to carry out hydrological modelling
3.3. Performance measures over the Tehri catchment using the high-resolution imageries procured
from NRSC (ISRO). Prior to being forced as geospatial inputs for HRU
The model’s performance can be evaluated using several efficacy analysis, the remotely sensed products must be used for classification of
measures. Among these, Nash-Sutcliffe Efficiency (NSE) and coefficient LULC and slope over the study area.
of determination (R2 ) are most commonly used (Moriasi et al., 2015). The slope map generated from DEM of Tehri catchment is presented
These efficacy measures represent the reconciliation of the model sim- in Fig. 3. Based on slope characteristics, the catchment area is classified
ulated values with respect to the observed values. NSE may range from into 5 categories, i.e., < 20%, 20-40%, 40-60%, 60-80%, and > 80%. It
-∞ to 1, whereas R2 varies from 0 to 1. The ideal value of both NSE and is evident from Fig. 3 that majority of the catchment area falls under
R2 is 1, which represents a perfect match of simulated outputs with the high slope categories.
observed data. The classified LULC map for the Tehri catchment is shown in Fig. 4.
Given that 𝑌𝑖𝑠𝑖𝑚 , 𝑌𝑖𝑜𝑏𝑠 , 𝑌 𝑠𝑖𝑚 , and 𝑌 𝑜𝑏𝑠 represent the simulated, ob- Based on LULC classification, nine different LULC classes over the catch-
served, mean of simulated values and mean of observed values in re- ment are identified. These are: forest (dense), open forest, agricultural
spective time steps i, NSE can be computed as, lands, shrubs and bushes, settlements, waterbodies, degraded lands,
rock, and snow/glacier. The southern portion of the catchment is dom-
⎡ ∑𝑛 ( 𝑜𝑏𝑠 )2 ⎤ inated by forests, open forests, and shrubs and bushes. The reservoir
⎢ 𝑖=1 𝑌𝑖 − 𝑌𝑖𝑠𝑖𝑚 ⎥
𝑁𝑆𝐸 = 1 − ⎢ ( )2 ⎥ (1) (waterbody) is clearly shown in the bottommost (southern) portion of
⎢ ∑𝑛 𝑌𝑖𝑜𝑏𝑠 − 𝑌 𝑜𝑏𝑠 ⎥⎦ the catchment. Similarly, the northern portion of the catchment is dom-
⎣ 𝑖=1

5
S. Swain, S.K. Mishra, A. Pandey et al. Environmental Challenges 8 (2022) 100579

Fig. 5. NDVI map of the Tehri catchment.

inated by rock and snow. Overall, the settlements and waterbody cover The SWAT model provides several outputs at the outlet of each
the least area among all the LULC classes. sub-catchment; however, the streamflow at the outlet of the entire
For the Tehri Catchment, the LULC imagery pertaining to 2017 was catchment is the focus of this study as the observed data is available
used to derive normalised difference vegetation (NDVI) map, which is for streamflow into the Tehri reservoir. As it is mentioned previously
presented in Fig. 5. It can be clearly noticed that the highest value of that the outlet of the entire catchment lies in sub-basin 32 with a
NDVI during 2017 was 0.815 and the lowest NDVI was -0.623. The NDVI total contributing area of 7268 km2 , the discharge therein provided
of upper (northern) portions are mostly negative, which represents ab- by SWAT should be calibrated with respect to the corresponding ob-
sence of vegetation. These areas are mostly covered by rock or snow. served data. As the observed discharge data for 2006-17 was avail-
The positive NDVI representing green cover is mostly found in lower able at the Tehri catchment outlet, the duration of 2006-17 is divided
(southern) portions of the catchment. These results are consistent with into 3 parts i.e. the year 2006 is taken as the warm-up period, 2007-
the LULC classified map presented in Fig. 4. Since reservoir is located 2013 as the calibration period and 2014-2017 as the validation pe-
in the lower portions of the catchment, presence of dense vegetation is riod. The warm period is the duration when the model simulations
desired in its nearby regions, which can be helpful for soil and water are carried out for initial soil-water balance (Pandey et al. 2021). In
conservation. other words, the warm-up of a model is an adjustment process such
The classified soil map of the Tehri catchment is presented in Fig. 6 . that the internal stores (e.g., soil moisture) reach an optimal state from
It can be clearly noticed that there are six different types of soil over the the estimated initial condition, thereby putting the model in an opti-
catchment, i.e., clayey loam, gravelly loam, sandy loam, sandy clayey mal state (Kim et al., 2018). Then the model is calibrated for the pe-
loam, rock outcrops and glacier/snow. From Fig. 5 and Fig. 6, it can be riod of 2007-13 with the observed flows. It is to mention that SWAT
inferred that vegetation cover is present over sandy loam, clayey loam model can be calibrated at daily or monthly time step; however, the ob-
and sandy clayey loam soil types. served flows are available at a 10-daily scale and therefore, the model

6
S. Swain, S.K. Mishra, A. Pandey et al. Environmental Challenges 8 (2022) 100579

Fig. 6. Classified Soil map over Tehri catchment.

Fig. 7. Comparison of simulated flow and observed flow after model calibration.

7
S. Swain, S.K. Mishra, A. Pandey et al. Environmental Challenges 8 (2022) 100579

Fig. 8. Dotty plots of all the parameters consid-

ered for SWAT calibration (Note: For each dotty
plot, the X-axis represents the parameter’s value for
each simulation and the Y-axis represents the cor-
responding NSE values).

8
S. Swain, S.K. Mishra, A. Pandey et al. Environmental Challenges 8 (2022) 100579

Fig. 9. Sensitivity chart (based on p-value) of the pa-

rameters considered for calibration of SWAT model.

was calibrated at monthly time steps. To this end, the observed 10- at the catchment outlet. From the relevant SWAT calibration litera-
daily data was rearranged to monthly scale and was provided as in- tures (Abbaspour, 2014; Abbaspour et al., 2015; Arnold et al., 2012,
put to the SWAT-CUP model with maximizing NSE as the objective 2013), these parameters were found to have significant influence on
function. streamflow. Obtaining their optimum value can be helpful to under-
The SWAT-CUP is simulated for 500 times with the ranges of the stand the hydrology of the catchment. The parameters considered for
possible sensitive parameters. The list of parameters and their ranges uncertainty analysis for discharge modeling for the Tehri catchment,
are selected based on the SWAT calibration and validation literature. their calibration range (minimum and maximum) and best-fitted val-
Higher number of simulations helps in improving the model perfor- ues are presented in Table 1. The detailed description of these as well
mance. The results of the best simulation (based on efficacy measures as other parameters of SWAT model can be referred from literature
on monthly scale) and its comparison with respect to the observed (Abbaspour, 2014; Arnold et al., 2013).
flows along with the observed rainfall at 10-daily scale for 7 years It is to mention that the upper portion of the catchment is dominated
i.e., 2007-2013 are presented in Fig. 7. It can be seen that the ob- by snow cover and hence, the snow-related parameters are considered
served flow and the best-simulated flow possess a good reconciliation for calibration.
with each other. The pattern of the flow is exquisitely replicated by The details of these parameters, and their importance can be ex-
the model. The efficacy measures viz., R2 and NSE were found to be plored from the previously cited studies as well as from the SWAT web-
0.74 and 0.68, respectively, at monthly scale, which justifies the ap- site (https://swat.tamu.edu/).
plicability of SWAT model to simulate discharge over the catchment. Further, it can be observed from Table 1 that the parameters are hav-
However, in some years (e.g. 2010 and 2013 at 10-daily scale), the ing the prefixes (R_, V_) and suffixes (.mgt, .hru, .bsn, .sol, .gw). The suf-
model clearly overestimated the flows. Upon detailed investigation, it is fixes represent which category the parameters belong to, i.e., .mgt, .hru,
found that the mismatch can be attributed mainly to the input rainfall .bsn, .sol, and .gw represent management, hydrologic response units,
data and other factors. The model results were relatively more propor- basin, soil and groundwater, respectively. On the other hand, the pre-
tional to the rainfall amounts, which is a sign of a good hydrological fixes in SWAT-CUP viz., R_ and V_ represents “relative” and “replace”
model. respectively. For the V_ prefixed parameters, their optimum value are to
The uncertainty analysis is carried out considering a list of 18 param- be replaced in the SWAT model whereas, for the R_ prefixed parameters,
eters, which were selected for assessing their effects on the discharge the initially set parameter is to be multiplied by (1+ best-fitted value).

9
S. Swain, S.K. Mishra, A. Pandey et al. Environmental Challenges 8 (2022) 100579

Fig. 10. Comparison of simulated flow and observed flow during model validation (2014-2017) at 10-daily scale.

Table 1 A well-calibrated hydrological model should be validated before be-

The parameters considered, their range and best fitted value obtained ing applied for future simulations. As the observed data at the Tehri
through SWAT calibration. catchment outlet was available until 2017, the validation was carried
Sl. No. Parameters Fitted Value Minimum Maximum out for the duration of 4 years, i.e., 2014-17. A comparison of the sim-
ulated flows with respect to the observed flows is presented in Fig. 10.
1 R__CN2.mgt 0.04535 -0.1 0.35
2 V__ALPHA_BF.gw 0.407 0 1
The calibrated model possessed exquisite results for the validation
3 V__GW_DELAY.gw 213.89 30 400 period, i.e., 2014-17. From the SWAT Simulated vs. Observed 10-Daily
4 V__GWQMN.gw 0.621 0 3 discharge plot in Fig. 10, it is evident that the model simulated flows
5 V__GW_REVAP.gw 0.0963 0 0.3 matched excellently with respect to the observed flows, particularly in
6 R__SOL_K(..).sol 0.21855 -0.05 2
terms of the temporal flow pattern. The peak-flows are also well cap-
7 R__SOL_AWC(..).sol 0.11114 0.02 1
8 V__ESCO.hru 0.75585 0.05 1 tured by the SWAT simulations. Hence, the SWAT model is validated
9 V__SURLAG.bsn 8.25875 0.05 10 over the catchment for simulating discharge, which is also reflected from
10 V__GDRAIN.mgt 3.07895 0.05 8 the efficacy measures. At monthly scale, the efficacy measures viz., R2
11 V__RCHRG_DP.gw 0.1589 0.05 0.95
and NSE values are found to be 0.84 and 0.83 respectively. Therefore,
12 V__SMTMP.bsn 8.38 0 20
13 V__SMFMX.bsn 0.62 0 20
the model can be utilised for simulating discharge over the basin as well
14 V__SMFMN.bsn 0.1 0 20 as to assess the impacts of climate change on hydrology of the catch-
15 R__OV_N.hru 0.32325 -0.2 0.95 ment. This study substantiates the utility of high-resolution remotely
16 R__SOL_Z(..).sol 1.7424 -0.2 3 sensed information for hydrological modelling over the Tehri catch-
17 R__SLSUBBSN.hru 123.82 10 150
ment. This research should be extended to other mountainous catch-
18 V__CANMX.hru 2.7 0 100
ments using state-of-the-art geospatial inputs, which can be helpful in
improving the hydrological model performances, thereby enabling a bet-
ter addressal of the water availability concerns over such regions.
All the parameters are updated in the SWAT model and the model is 5. Conclusions
calibrated.
For calibration of SWAT model, the Nash-Sutcliffe Efficiency (NSE) An application of the widely used SWAT hydrological model using
is selected as the objective function. There are eleven efficacy measures high-resolution geospatial inputs (LISS IV MX imageries of LULC and the
out of which, one can be selected as the objective function to obtain the Cartosat-1 DEM) over the Tehri reservoir catchment (India) is presented.
best simulation. NSE is recommended as one of the most reliable mea- The LULC and DEM have a spatial resolution of 5.8 m and 2.5 m, respec-
sure for checking model accuracy (McCuen et al., 2006; Moriarsi et al., tively. SWAT model is applied for a duration of 12 years (2006-2017)
2015). The effects of the selected parameters on model performance is with 2006, 2007-2013 and 2014-2017 as the warm-up, calibration and
presented in terms of dotty plots in Fig. 8. The X- axis represents the validation periods, respectively. The results reveal an excellent perfor-
value of the parameter and the Y-axis shows the resulting NSE for all mance of the model in streamflow simulation, as evident from the effi-
the 500 simulations. cacy measures (R2 = 0.84, NSE = 0.83) during the validation period. The
It can be seen from the dotty plots that CN2 (Curve Number) and calibrated and validated model will be useful for water resources plan-
CANMX (Canopy Storage Capacity) are the most sensitive parameters ning and management over the Tehri catchment. The information re-
as there is a clear change in the performance metric of the model with ported in this study substantiates the utility of high-resolution remotely
their changing values. The sensitivity analysis of various parameters for sensed information for hydrological modelling over the mountainous
their effects on discharge is a crucial step in hydrological modeling. For catchments in general, and the Tehri catchment in particular.
the calibration period (2007-13), the model sensitivity is assessed. The
ranking of the sensitive parameters is done using p-value. The p-value is CRediT author statement
computed to assess the probabilistic significance of the sensitivity anal-
yses. The p-value is called the significance level and (1-p) is regarded Sabyasachi Swain: Conceptualization, Data curation, Methodology,
as the confidence level. For example, if the p-value is 0.05, there is a Figures, Writing (Original draft); Surendra Kumar Mishra and Ashish
95% confidence level that the results are significant. The most sensi- Pandey: Conceptualization, Supervision, Writing (Review and editing);
tive parameters along with their significance (determined by p-value) is A.C. Pandey: Data curation, Writing (Review and editing); Atul Jain,
presented in Fig. 9. S. K. Chauhan and A. K. Badoni: Writing (Review and editing).

10
S. Swain, S.K. Mishra, A. Pandey et al. Environmental Challenges 8 (2022) 100579

Funding Himanshu, S.K., Pandey, A., Patil, A., 2018. Hydrologic evaluation of the TMPA-3B42V7
precipitation data set over an agricultural watershed using the SWAT model. J. Hy-
drol. Eng. 23 (4), 05018003.
This study is a by-product of the R&D project (THD-1176-WRC, IIT Himanshu, S.K., Pandey, A., Shrestha, P., 2017. Application of SWAT in an Indian river
Roorkee) sponsored by THDCIL, Rishikesh, Uttarakhand (India). basin for modeling runoff, sediment and water balance. Environ. Earth Sci. 76 (1), 3.
Himanshu, S.K., Pandey, A., Yadav, B., Gupta, A., 2019. Evaluation of best management
practices for sediment and nutrient loss control using SWAT model. Soil Tillage Res.
Declaration of Competing Interest 192, 42–58.
Immerzeel, W.W., Lutz, A.F., Andrade, M., Bahl, A., Biemans, H., Bolch, T., Hyde, S.,
Brumby, S., Davies, B.J., Elmore, A.C., Emmer, A., Feng, M., Fernández, A.,
The authors declare that they have no known competing financial Haritashya, U., Kargel, J.S., Koppes, M., Kraaijenbrink, P.D.A., Kulkarni, A.V.,
interests or personal relationships that could have appeared to influence Mayewski, P., Nepal, S., Pacheco, P., Painter, T.H., Pellicciotti, F., Rajaram, H., Rup-
per, S., Sinisalo, A., Shrestha, A.B., Viviroli, D., Wada, Y., Xiao, C., Yao, T., Bail-
the work reported in this paper.
lie, J.E.M., 2020. Importance and vulnerability of the world’s water towers. Nature
577, 364–369.
Data availability statement Jain, S.K., Kumar, S., Varghese, J., 2001. Estimation of soil erosion for a Himalayan wa-
tershed using GIS technique. Water Resour. Manage. 15 (1), 41–54.
The geospatial inputs used in this study are not open-access and Khan, A., Govil, H., Taloor, A.K., Kumar, G., 2020. Identification of artificial groundwater
recharge sites in parts of Yamuna River basin India based on Remote Sensing and
can’t be provided by the authors. They can be purchased from Geographical Information System. Groundw. Sustain. Dev. 11, 100415.
their source organizations, whose details are available inside the Kim, K.B., Kwon, H.H., Han, D., 2018. Exploration of warm-up period in conceptual hy-
manuscript. The use of 10-daily flow data from THDCIL is duly drological modelling. J. Hydrol. 556, 194–210.
Kim, N.W., Lee, J., 2010. Enhancement of the channel routing module in SWAT. Hydrol.
acknowledged. To the best of the authors’ knowledge and belief,
Process. 24 (1), 96–107.
the data used is of no commercial value. However, the flow data Kumar, A., 2015. Geomorphologic instantaneous unit hydrograph based hydrologic re-
can’t be shared in the repository. The interested readers may con- sponse models for ungauged hilly watersheds in India. Water Resour. Manage. 29 (3),
863–883.
tact THDCIL requesting for observed streamflow data. Only the
Kumar, R., Anbalagan, R., 2015. Landslide susceptibility zonation in part of Tehri reservoir
meteorological data used in this can be available from the corre- region using frequency ratio, fuzzy logic and GIS. J. Earth Syst. Sci. 124 (2), 431–448.
sponding author, upon reasonable request. Licciardello, F., Rossi, C.G., Srinivasan, R., Zimbone, S.M., Barbagallo, S., 2011. Hydro-
logic evaluation of a Mediterranean watershed using the SWAT model with multiple
PET estimation methods. Trans. ASABE 54 (5), 1615–1625.
Acknowledgments Lu, C.M., Chiang, L.C., 2019. Assessment of Sediment Transport Functions with the Modi-
fied SWAT-Twn Model for a Taiwanese Small Mountainous Watershed. Water 11 (9),
1749.
The authors would like to thank THDCIL and its project evaluation Malik, A., Kumar, A., Kandpal, H., 2019. Morphometric analysis and prioritization of sub-
committee members, whose suggestions were helpful for this work. The -watersheds in a hilly watershed using weighted sum approach. Arabian J. Geosci. 12
(4), 118.
authors extend the acknowledgment to National Remote Sensing Centre McCuen, R.H., Knight, Z., Cutter, A.G., 2006. Evaluation of the Nash–Sutcliffe efficiency
(Indian Space Research Organization) and Indian Institute of Technol- index. J. Hydrol. Eng. 11 (6), 597–602.
ogy Roorkee for provision of data and facilities. We also thank both Moriasi, D.N., Gitau, M.W., Pai, N., Daggupati, P., 2015. Hydrologic and water qual-
ity models: Performance measures and evaluation criteria. Trans. ASABE 58 (6),
the anonymous reviewers, whose suggestions helped us to improve the
1763–1785.
manuscript. Muralikrishnan, S., Pillai, A., Narender, B., Reddy, S., Venkataraman, V.R., Dadhwal, V.K.,
2013. Validation of Indian national DEM from Cartosat-1 data. J. Indian Soc. Remote
Sens. 41 (1), 1–13.
References Muralikrishnan, S., Senthilkumar, A., Manjunath, A.S., Rao, K.M.M., 2006. Geometric
quality assessment of Cartosat-1 data products, ISPRS Commission IV, WGIV/9. Int.
Aadhar, S., Swain, S., Rath, D.R., 2019. Application and performance assessment of SWAT Archs. Photogram. Rem. Sens. 36 (4).
hydrological model over Kharun river basin, Chhattisgarh, India. In: World Environ- Neitsch, S. L., Arnold, J. G., Kiniry, J. R., & Williams, J. R. (2011). Soil and water assessment
mental and Water Resources Congress 2019: Watershed Management, Irrigation and tool theoretical documentation version 2009. Texas Water Resources Institute technical
Drainage, and Water Resources Planning and Management, Reston, VA. American report No. 406, Texas A & M University System, Texas.
Society of Civil Engineers, pp. 272–280. Omani, N., Srinivasan, R., Smith, P.K., Karthikeyan, R., 2017. Glacier mass balance simu-
Abbaspour, K.C., 2014. SWAT-CUP 2012: SWAT Calibration and Uncertainty Programs— lation using SWAT distributed snow algorithm. Hydrol. Sci. J. 62 (4), 546–560.
A User Manual. Swiss Federal Institute of Aquatic Science and Technology, p. 106. Palmate, S.S., Pandey, A., 2021. Effectiveness of best management practices on dependable
Abbaspour, K.C., Rouholahnejad, E., Vaghefi, S., Srinivasan, R., Yang, H., Kløve, B., 2015. flows in a river basin using hydrological SWAT model. In: Water Management and
A continental-scale hydrology and water quality model for Europe: calibration and Water Governance. Springer, Cham, pp. 335–348.
uncertainty of a high-resolution large-scale SWAT model. J. Hydrol. 524, 733–752. Palmate, S.S., Amrit, K., Jadhao, V.G., Dayal, D., Himanshu, S.K., 2022a. Prioritization of
Aragaw, H.M., Goel, M.K., Mishra, S.K., 2021. Hydrological responses to human-induced erosion prone areas based on a sediment yield index for conservation treatments: a
land use/land cover changes in the Gidabo River basin, Ethiopia. Hydrol. Sci. J. 66 case study of the upper Tapi River basin. In: Advances in Remediation Techniques for
(4), 640–655. Polluted Soils and Groundwater. Elsevier, pp. 291–307.
Arnold, J.G., Fohrer, N., 2005. SWAT2000: current capabilities and research opportunities Palmate, S.S., Pandey, A., Pandey, R.P., Mishra, S.K., 2021. Assessing the land degrada-
in applied watershed modelling. Hydrol. Processes 19 (3), 563–572. tion and greening response to changes in hydro-climatic variables using a conceptual
Arnold, J.G., Kiniry, J.R., Srinivasan, R., Williams, J.R., Haney, E.B., Neitsch, S.L., 2013. framework: A case-study in central India. Land Degrad. Dev. 32 (14), 4132–4148.
SWAT 2012 input/output documentation. Texas Water Resources Institute. Palmate, S.S., Wagner, P.D., Fohrer, N., Pandey, A., 2022b. Assessment of uncertainties in
Arnold, J.G., Moriasi, D.N., Gassman, P.W., Abbaspour, K.C., White, M.J., Srinivasan, R., modelling land use change with an integrated cellular automata–Markov chain model.
Santhi, C., Harmel, R.D., Van Griensven, A., Van Liew, M.W., Kannan, N., Jha, M.K., Environ. Model. Assess. 27 (2), 275–293.
2012. SWAT: Model use, calibration, and validation. Trans. ASABE 55 (4), 1491–1508. Pandey, A., Palmate, S.S., 2018. Assessments of spatial land cover dynamic hotspots em-
Bressiani, D.A., Gassman, P.W., Fernandes, J.G., Garbossa, L.H.P., Srinivasan, R., Bon- ploying MODIS time-series datasets in the Ken River Basin of Central India. Arabian
umá, N.B., Mendiondo, E.M., 2015. Review of soil and water assessment tool (SWAT) J. Geosci. 11 (17), 1–8.
applications in Brazil: challenges and prospects. Int. J. Agricult. Biolog. Eng. 8 (3), Pandey, A., Palmate, S.S., 2019. Assessing future water–sediment interaction and critical
9–35. area prioritization at sub-watershed level for sustainable management. Paddy Water
Dhami, B., Himanshu, S.K., Pandey, A., Gautam, A.K., 2018. Evaluation of the SWAT Environ. 17 (3), 373–382.
model for water balance study of a mountainous snowfed river basin of Nepal. Envi- Pandey, A., Bishal, K.C., Kalura, P., Chowdary, V.M., Jha, C.S., Cerdà, A., 2021. A soil wa-
ron. Earth Sci. 77 (1), 1–20. ter assessment tool (SWAT) modeling approach to prioritize soil conservation man-
Gupta, A., Himanshu, S.K., Gupta, S., Singh, R., 2020. Evaluation of the SWAT model for agement in river basin critical areas coupled with future climate scenario analysis.
analysing the water balance components for the upper Sabarmati Basin. In: Advances Air, Soil and Water Research 14 11786221211021395.
in Water Resources Engineering and Management. Springer, Singapore, pp. 141–151. Pandey, A., Himanshu, S.K., Mishra, S.K., Singh, V.P., 2016. Physically based soil erosion
Guptha, G.C., Swain, S., Al-Ansari, N., Taloor, A.K., Dayal, D., 2022. Assessing the role of and sediment yield models revisited. Catena 147, 595–620.
SuDS in resilience enhancement of urban drainage system: A case study of Gurugram Pandey, A., Mathur, A., Mishra, S.K., Mal, B.C., 2009. Soil erosion modeling of a Hi-
City, 41. Urban Climate, India. malayan watershed using RS and GIS. Environ. Earth Sci. 59 (2), 399–410.
Guptha, G.C., Swain, S., Al-Ansari, N., Taloor, A.K., Dayal, D., 2021. Evaluation of an Pandey, B.K., Khare, D., Kawasaki, A., Meshesha, T.W., 2021. Integrated approach to sim-
urban drainage system and its resilience using remote sensing and GIS. Remote Sens. ulate hydrological responses to land use dynamics and climate change scenarios em-
Applic. 23, 100601. ploying scoring method in upper Narmada basin, India. J. Hydrol. 598, 126429.
Haque, S., Kannaujiya, S., Taloor, A.K., Keshri, D., Bhunia, R.K., Ray, P.K.C., Chauhan, P., Pandey, B.K., Khare, D., Kawasaki, A., Mishra, P.K., 2019. Climate change impact assess-
2020. Identification of groundwater resource zone in the active tectonic region of ment on blue and green water by coupling of representative CMIP5 climate models
Himalaya through earth observatory techniques. Groundw. Sustain. Dev. 10, 100337. with physical based hydrological model. Water Resour. Manage. 33 (1), 141–158.

11
S. Swain, S.K. Mishra, A. Pandey et al. Environmental Challenges 8 (2022) 100579

Patel, P., Thakur, P.K., Aggarwal, S.P., Garg, V., Dhote, P.R., Nikam, B.R., Swain, S., Swain, S., Mishra, S.K., Pandey, A., 2020. Assessment of meteorological droughts over
Al-Ansari, N., 2022. Revisiting 2013 Uttarakhand flash floods through hydrological Hoshangabad district, India. IOP conference series: earth and environmental science,
evaluation of precipitation data sources and morphometric prioritization. Geomat. 491. IOP Publishing.
Nat. Hazard. Risk 13 (1), 646–666. Swain, S., Mishra, S.K., Pandey, A., 2021a. A detailed assessment of meteorological
Rahman, K., Maringanti, C., Beniston, M., Widmer, F., Abbaspour, K., Lehmann, A., 2013. drought characteristics using simplified rainfall index over Narmada River Basin, In-
Streamflow modeling in a highly managed mountainous glacier watershed using dia. Environ. Earth Sci. 80 (6), 221.
SWAT: the Upper Rhone River watershed case in Switzerland. Water Resour. Man- Swain, S., Mishra, S.K., Pandey, A., 2021b. Assessing contributions of intensity-based rain-
age. 27 (2), 323–339. fall classes to annual rainfall and wet days over Tehri catchment, India. In: Advances
Sarkar, T., Kannaujiya, S., Taloor, A.K., Ray, P.K.C., Chauhan, P., 2020. Integrated study of in Water Resources and Transportation Engineering. Springer, Singapore, pp. 113–
GRACE data derived interannual groundwater storage variability over water stressed 121. doi:10.1007/978-981-16-1303-6_9.
Indian regions. Groundw. Sustain. Dev. 10, 100376. Swain, S., Mishra, S.K., Pandey, A., 2022a. Appraisal of land use/land cover change
Schuol, J., Abbaspour, K.C., Srinivasan, R., Yang, H., 2008. Estimation of freshwater avail- over Tehri catchment using remote sensing and GIS. In: Geospatial Technolo-
ability in the West African sub-continent using the SWAT hydrologic model. J. Hydrol. gies for Land and Water Resources Management. Springer, Cham, pp. 37–51.
352 (1-2), 30–49. doi:10.1007/978-3-030-90479-1_3.
Sindhu, D., Ravikumar, A.S., Shivakumar, B.L., 2015. Quantitative analysis of catch- Swain, S., Mishra, S.K., Pandey, A., Dayal, D., 2021c. Identification of meteorological
ment using remote sensing and geographic information system. Aquat. Procedia 4, extreme years over central division of Odisha using an index-based approach. In: Hy-
1421–1428. drological Extremes. Springer, Cham, pp. 161–174.
Singh, A., Malik, A., Kumar, A., Kisi, O., 2018. Rainfall-runoff modeling in hilly watershed Swain, S., Mishra, S.K., Pandey, A., Dayal, D., 2022b. Spatiotemporal assessment of precip-
using heuristic approaches with gamma test. Arabian J. Geosci. 11 (11), 261. itation variability, seasonality, and extreme characteristics over a Himalayan catch-
Singh, G., Pandey, A., 2021a. Evaluation of classification algorithms for land use land ment. Theor. Appl. Climatol. 147, 817–833. doi:10.1007/s00704-021-03861-0.
cover mapping in the snow-fed Alaknanda River Basin of the Northwest Himalayan Swain, S., Mishra, S.K., Pandey, A., Kalura, P., 2022c. Inclusion of groundwater and so-
Region. Appl. Geomat. 13 (4), 863–875. cio-economic factors for assessing comprehensive drought vulnerability over Narmada
Singh, G., Pandey, A., 2021b. Hybrid ensemble modeling for flash flood potential assess- River Basin, India: a geospatial approach. Appl. Water Sci. 12 (2), 1–16.
ment and susceptibility analysis of a Himalayan river catchment. Geocarto Int. 1–28. Swain, S., Sahoo, S., Taloor, A.K., 2022d. Groundwater quality assessment using geospatial
doi:10.1080/10106049.2021.2017007. and statistical approaches over Faridabad and Gurgaon districts of National Capital
Sinnathamby, S., Douglas-Mankin, K.R., Craige, C., 2017. Field-scale calibration of Region, India. Appl. Water Sci. 12 (4), 1–14.
crop-yield parameters in the Soil and Water Assessment Tool (SWAT). Agric. Water Swain, S., Sahoo, S., Taloor, A.K., Mishra, S.K., Pandey, A., 2022e. Exploring recent
Manage. 180, 61–69. groundwater level changes using Innovative Trend Analysis (ITA) technique over
Sood, V., Gusain, H.S., Gupta, S., Taloor, A.K., Singh, S., 2021. Detection of snow/ice cover three districts of Jharkhand, India. Groundw. Sustain. Dev., 100783.
changes using subpixel-based change detection approach over Chhota-Shigri glacier, Swain, S., Taloor, A.K., Dhal, L., Sahoo, S., Al-Ansari, N., 2022f. Impact of climate change
Western Himalaya, India. Quat. Int. 575, 204–212. on groundwater hydrology: a comprehensive review and current status of the Indian
Stehr, A., Debels, P., ARUMI, J.L., Romero, F., Alcayaga, H., 2009. Combining the Soil and hydrogeology. Appl. Water Sci. 12 (6), 120. doi:10.1007/s13201-022-01652-0.
Water Assessment Tool (SWAT) and MODIS imagery to estimate monthly flows in a Swain, S., Verma, M.K., Verma, M.K., 2018. Streamflow estimation using SWAT model
data-scarce Chilean Andean basin. Hydrol. Sci. J. 54 (6), 1053–1067. over Seonath river basin, Chhattisgarh, India. In: Hydrologic modeling. Springer, Sin-
Swain, S., 2017. Hydrological modeling through soil and water assessment toolin a cli- gapore, pp. 659–665.
mate change perspective a brief review. In: 2017 2nd International conference for Tripathi, M.P., Panda, R.K., Raghuwanshi, N.S., Singh, R., 2004. Hydrological modelling
convergence in technology (I2CT). IEEE, pp. 358–361. of a small watershed using generated rainfall in the soil and water assessment tool
model. Hydrol. Process. 18 (10), 1811–1821.

12