Science-policy interaction is vital for addressing hydro-climatic disasters in data-limited regions, with modeling and analysis as key components. The utilization of the soil and water assessment tool (SWAT) model facilitated an evaluation of water balance variations across time and space within Narayani Basin through multi-site calibration. The adjustment of all parameters via the SUFI-2 algorithm revealed that precipitation and temperature lapse rate (PLAPS and TLAPS) exhibit higher sensitivity in scenarios where observed stations fail to capture orographic effects. The calibrated model accurately replicated evapotranspiration, net water yield, and groundwater flow for each sub-basin, including average flow and flow duration curve at calibration points. Findings indicated that 22% of precipitation is lost to evaporation, while 75% contributes to basin runoff, showcasing significant spatial and temporal variability in water balance components. Notably, net water yield comprises 44% lateral flow, 38% surface flow, and 16% groundwater flow, with distinct spatial patterns favoring lateral flow in the Himalayas and groundwater flow in the plains due to topographical variations. These outcomes offer actionable insights for policymakers and water resource managers, enabling assessments of climate and land use impacts and facilitating the formulation of policies for sustainable water resource utilization.

  • Variability of water balance components in rugged topography, including groundwater and surface water interactions.

  • The research output would be applicable for planning water resource development and policy formulation.

  • A hydrological model has been developed using a multi-site calibration and validation approach, which will be valuable for future studies, such as climate change impact assessments.

Water availability within hydrological systems is governed by intricate interplays of geophysical, hydro-climatic factors, and human interventions, crucial on a global scale (Gleick 2000; Tijerina et al. 2021). Understanding these dynamics is paramount for effective water resource management and is critical to achieving sustainable development goals and ensuring community resilience (Wada et al. 2017). Nepal is richly endowed with water resources like glaciers, rivers, lakes, and groundwater that consider water a pivotal driver for economic growth, especially in agriculture and hydropower generation (WECS 2011). Optimal utilization of these resources demands a nuanced understanding of water balance variations across diverse spatial and temporal scales.

In Nepal, data scarcity poses a significant challenge, prompting the adoption of hydrologic simulation methods as cost-effective tools for water resource management and hazard mitigation (Yanto et al. 2017; Talchabhadel et al. 2021; Ghobadi & Kang 2023). Existing research within Nepal's river systems, including Narayani River Basin (NRB) and its tributaries, has primarily focused on single-site optimization, often lacking comprehensive considerations of hydrological processes' discretization (Bhattarai et al. 2018; Chand et al. 2019; Dawadi et al. 2020). Hydrological modeling offers a robust framework for comprehending water patterns over watersheds, with different techniques categorized as lumped, semi-distributed, or distributed models, catering to specific aspects of flow generation (Brirhet & Benaabidate 2016). Among these, the soil and water assessment tool (SWAT) stands as a widely implemented model, facilitating assessments across diverse geographical scales.

However, in the context of highly diverse basins like NRB, employing a multi-site calibration approach becomes imperative to enhance model efficiency and reduce uncertainties in hydrological simulations (Molina-Navarro et al. 2017; Malik et al. 2022). Although only studies have applied multi-site calibration strategies in adjacent basins (Bharati et al. 2014; Pandey et al. 2020c), comprehensive investigations encompassing the entirety of NRB are limited. Addressing these complexities demands sophisticated uncertainty analysis, which involves a range of approaches within the SWAT-Calibration and Uncertainty Programs (SWAT-CUP) auto-calibration tool.

These methods encompass various algorithms, each with distinct assessment methodologies and parameter range estimates for a specific target function (SalimiRad et al. 2020). Notably, studies in diverse river basins worldwide have identified SUFI-2 as consistently producing superior results and optimal parameter ranges while requiring the shortest running time (Molina-Navarro et al. 2017; Desai et al. 2021; Mahmudi et al. 2021). SUFI-2, a semi-automatic optimization technique utilizing the Latin Hypercube sampling scheme, efficiently achieves optimal outcomes and facilitates calibration and validation at multiple hydrological stations, accommodating various objective functions (Serur & Adi 2022).

This study aims to bridge this gap by adopting a multi-site calibration approach using SUFI-2 within NRB. It seeks to establish a hydrological model that thoroughly examines the spatial and temporal distribution of water balance. The fully calibrated and validated hydrological model's applicability would extend to climate change impact assessment in conjunction with land use change, facilitating the subsequent quantification of uncertainties.

The general methods used in this study are shown in Figure 1. In general, it incorporates the pre-processing of time series and spatial data, hydrological model setup, its parameterization, as well as the baseline hydrological characterization. The methodology will be detailed in the subsequent sub-sections.
Figure 1

Methodological framework for hydrological model setup and application for the evaluation of water yield. DEM, digital elevation model; LULC, land use/cover; PCP, precipitation (mm); Tmax, Tmin, maximum and minimum temperature (°C); RH, relative humidity (fraction); WS, wind speed (m/s); and SR, solar radiation (MJ/m2/day).

Figure 1

Methodological framework for hydrological model setup and application for the evaluation of water yield. DEM, digital elevation model; LULC, land use/cover; PCP, precipitation (mm); Tmax, Tmin, maximum and minimum temperature (°C); RH, relative humidity (fraction); WS, wind speed (m/s); and SR, solar radiation (MJ/m2/day).

Close modal

Study area

The centrally located NRB has an area of 36,498 km2 out of which approximately 13% is situated in the Tibetan Plateau, China. The source of the river is the Trans-Himalayas, and it passes through mountains, hills, and Siwalik in the southern part as shown in Figure 2. The fan shape of the watershed and dendritic river networks in most of the area could result in flash floods in the monsoon. Trishuli, Budigandaki, Marshyangdi, Daraudi, Madi, Seti Gandaki, and Kaligandaki are the seven primary tributaries of the Narayani, with the exception of the Daraudi and Madi, which have catchment areas with glaciers (Shrestha & Bajracharya 2011). Many rivulets are also drained into the Narayani for example, the East Rapti River which joins the river at Chitwan.
Figure 2

Study area and distribution of meteorological station.

Figure 2

Study area and distribution of meteorological station.

Close modal

The NRB has a wide diversity in biophysical and climatic characteristics. The topography varies greatly, with elevations ranging from 100 to 8,174 m above sea level. The area is divided into nine different land use categories, with forests being the most dominant, covering approximately 43% of the land. There are also 21 different soil types in the basin, with Gleic Leptosole (Lpi) being the most prevalent, making up around 30% of the total soil in the area.

Agriculturally, the region covers approximately 62,000 ha of land, with practices primarily reliant on rainfed cultivation in the hills. However, the southern regions are prone to water-related disasters such as flash floods and inundation as highlighted by Dandekhya et al. (2017). These calamities often result in significant siltation and flooding, adversely impacting nearby fields and reducing land productivity during flood occurrences.

Swat model

SWAT is a hydrologic model developed by the United States Department of Agriculture (USDA), Agriculture Research Service (ARS) that is based on physical principles and is used to simulate the quantity and quality of surface and groundwater (Phan et al. 2011). It has been extensively employed for predicting runoff, sediment yield, and nutrient transport (Agarwal et al. 2014; Khoi & Suetsugi 2014) and has also been adapted to incorporate the effects of climate change (Dahal et al. 2016).

In SWAT, a watershed is discretized into sub-basins, and each sub-basin is further subdivided into distinct hydrologic response units (HRUs) that are characterized by their homogeneous hydrologic properties (soil type and land cover/use) (Shope et al. 2014). The input parameters for the model include precipitations, wind speed (WS), temperature, solar radiation, soil type, land cover/use, and slope. Time-varying inputs are provided on a daily time step. The water balance is simulated in SWAT using the following equation (Arnold et al. 1998):
formula
(1)
where SWf is the final moisture content (mm), SWin is the initial moisture content (mm), t is the time (days) for each ith day, Rd is the rainfall depth (mm), Qrunoff is the surface runoff (mm), Ei is the evapotranspiration (mm), Wspg is the water seepage (mm), and QGW is the groundwater recharge to the channel network (mm). SWAT implements the Soil Conservation Service curve number (SCS-CN) method or the Green and Ampt infiltration method to calculate surface runoff. The SCS-CN method is used in this study. To compute the potential evapotranspiration and channel routing, the Penman–Monteith method and variable storage method, respectively, applied in this study.

SWAT required spatial data and continuous time series climatic data as input. The three types of spatial data are digital elevation model (DEM), land use/cover (LULC), and soil. The resampled 90 m DEM of ALOS World 3D–30 m (JAXA EORC 2021) is used for this study. As per DEM, the elevation difference across the NRB varies from 100 to 8,174 m. Similarly, the LULC for HinduKush developed by ICIMOD (2013) is used for the study. Also, a 1:1 million resolution SOTER map for Nepal and China is used for this study (Gambia et al. 2008; Huting 2009).

Moreover, time series hydro-meteorological data are obtained from the Department of Hydrology and Meteorology (DHM) for the period of 1980–2019. There are 86 stations for precipitation that have been used, among them 8 stations from the third pole environment were made available. Similarly, 18 climatological, 14 agrometeorological, and 1 aeronautical station have reliable data to use. The missing data for the precipitation and normal ratio method (Paulhus & Kohler 1952) have been used to fill the gaps, while an annual average technique was applied for temperature and other variables.

Model setup

The three types of spatial data – DEM, LULC, and soil as shown in Figure 3 – were utilized to create the model. Once the pre-processing was completed, the climate data were converted to a format that was compatible with SWAT. The rainfall and minimum and maximum temperatures were used in similar units of all measurements by DHM (in millimeters and degrees Celsius). However, the daily observations of relative humidity (taken in the morning and evening) were averaged and then converted to a fraction. To convert the daily sunshine hours into solar radiation (MJ/m2/day), the Angstrom–Prescott model was employed. Additionally, the daily WS data, initially measured in km/h, was converted into m/s.
Figure 3

Spatial data for SWAT model setup: (a) digital elevation model, (b) land use map of 2010, (c) soil map, and (d) discretized sub-basin of the model.

Figure 3

Spatial data for SWAT model setup: (a) digital elevation model, (b) land use map of 2010, (c) soil map, and (d) discretized sub-basin of the model.

Close modal
Figure 4

Calibration and validation stations based on available observed daily discharge.

Figure 4

Calibration and validation stations based on available observed daily discharge.

Close modal

The SWAT model is set up with the ArcSWAT2012 platform in ArcGIS. During the model setup, 10,000 ha of threshold area was bound to generate the stream network. With consideration of 48 hydropower projects at different stages, 8 existing irrigation projects, 13 proposed storage projects, and to apprehend spatial diversity, the basin was discretized into 199 sub-basins with areas ranging from 0.67 to 924.80 km2. Then, considering a 10% threshold of land use, soil and slope along with five slope classes (0–20, 20–35, 35–50, 50–70, and >70%) were applied so that approximately equal area falls under each slope interval. Five elevation bands at an interval of 500 m reduce the orographic effect of temperature and precipitation, and these elevation bands are more significant than increasing HRUs or sub-basins (Bhatta et al. 2019). And finally, 2,838 HRUs were defined.

Meteorological data were collected from various stations and included daily precipitation from 86 stations, maximum and minimum temperature from 18 stations, relative humidity from 14 stations, WS from 1 station, and sunshine hours from 1 station. The SCS curve number technique was used to calculate the surface runoff, and the daily curve number was determined based on the soil moisture. The Penman–Monteith method was used to assess potential evapotranspiration (Amini et al. 2022). For channel routing, a variable storage method was used.

Model optimization

A multi-site calibration approach was carried out to effectively capture spatial variation (Figure 4). The calibration and validation started from upstream and moved downstream after locking the calibrated upstream stations. The calibration and validation time frames vary for each calibration point based on the availability of quality data (Table 1). Here only the station having less than 10% of missing data is considered. First, the model was run from 1980 to 2019 and 5 years of warm-up period was used to stabilize the model. Calibration of the model was carried out as per Neitsch et al. (2002), on the three stages: (i) evaluation of sensitivity, (ii) auto calibration, and (iii) manual calibration for fine-tuning.

Table 1

Station used for calibration and validation

S.No.Sub-basinStation nameCal period (year)Val period (year)
Kaligandaki Myagdi Malangaghat Q404.7 1990–2005 (16) 2006–2019 (14) 
Kali Gandaki_Angsing Q419.1 1996–2010 (15) 2011–2019 (9) 
Kaligandaki Kotagaun Q420 1985–2005 (21) 2006–2019 (14) 
Seti Gandaki Seti Gandaki_Damauli Q430.5 2000–2011 (12) 2012–2019 (12) 
Marshynagdi Marshyandgi Bimalnagar Q439.7 1988–2008 (21) 2009–2019 (11) 
Budhi Gandaki Budhi Gandaki Arughat Q445 1985–2000 (16) 2001–2010 (10) 
Trishuli Trishuli Betrawati Q447 1985–2000 (16) 2001–2010 (10) 
Tadi Belkot Q448 1985–2005 (21) 2006–2019 (14) 
Trishuli Kalikhola Q449.91 1996–2010 (15) 2011–2019 (9) 
10 Narayani Main Out let Narayani Devghat Q450 1985–2005 (21) 2006–2019 (14) 
S.No.Sub-basinStation nameCal period (year)Val period (year)
Kaligandaki Myagdi Malangaghat Q404.7 1990–2005 (16) 2006–2019 (14) 
Kali Gandaki_Angsing Q419.1 1996–2010 (15) 2011–2019 (9) 
Kaligandaki Kotagaun Q420 1985–2005 (21) 2006–2019 (14) 
Seti Gandaki Seti Gandaki_Damauli Q430.5 2000–2011 (12) 2012–2019 (12) 
Marshynagdi Marshyandgi Bimalnagar Q439.7 1988–2008 (21) 2009–2019 (11) 
Budhi Gandaki Budhi Gandaki Arughat Q445 1985–2000 (16) 2001–2010 (10) 
Trishuli Trishuli Betrawati Q447 1985–2000 (16) 2001–2010 (10) 
Tadi Belkot Q448 1985–2005 (21) 2006–2019 (14) 
Trishuli Kalikhola Q449.91 1996–2010 (15) 2011–2019 (9) 
10 Narayani Main Out let Narayani Devghat Q450 1985–2005 (21) 2006–2019 (14) 

The most responsive parameters of the basin were identified initially by reviewing various literature sources (Devkota & Gyawali 2015; Pandey et al. 2019; Pandey et al. 2020a, 2020b) that covered similar catchment areas and based on previous experiences. It was observed that certain parameters were considerably sensitive in specific sub-watersheds, while different parameters were more influential in other sub-watersheds. Twenty-nine (29) sensitive parameters were selected for calibration as presented in Table 2. As stated by Kouchi et al. (2017), the parameter optimization options replace (v_), and relative change (r_) was applied through the SUFI2 algorithm.

Table 2

Fitted value of each calibrated parameters of each sub-basin

Parameter nameMethodSuggested rangeKaligandaki
Seti GandakiMarshyangdiBudhi GandakiTrishuli
Narayani
Q404.7Q419.1Q420Q430.5Q439.7Q445Q447Q448Q449.91Q450
CN2 r_ 35–98 0.14 0.09 0.09 −0.19 0.32 0.08 0.09 −0.05 0.11 −0.20 
PLAPS v_ −1,000–1,000 35.59 47.48 47.48 −3.38 109.13 219.67 300.00 16.28 38.20 94.12 
TLAPS v_ −10–10 −5.61 −6.61 −6.61 −5.41 −3.15 −7.23 −6.68 −7.05 −7.43 −0.89 
SFTMP v_ −20–20 – – −5.29 – −5.29 v5.29 −5.29 – – – 
SMTMP v_ −20–20 – – −3.53 – −3.53 −3.53 −3.53 – – – 
SMFMX v_ 0–20 – – 4.04 – 4.04 4.04 4.04 – – – 
SMFMN v_ 0–20 – – 5.00 – 5.00 5.00 5.00 – – – 
TIMP v_ 0–1 – – 0.10 – 0.10 0.10 0.10 – – – 
SURLAG v_ 0.05–24 – – 2.65 – 2.65 2.65 2.65 – – – 
SOL_ALB r_ 0–0.25 0.10 −0.06 −0.06 0.06 −0.10 −0.02 0.22 0.08 0.03 −0.06 
SOL_Z r_ 0–3,500 −0.06 −0.03 −0.03 −0.08 −0.08 −0.05 −0.18 −0.04 −0.08 0.04 
SOL_AWC r_ 0–1 −0.15 −0.09 −0.09 −0.14 0.89 0.04 −0.07 0.06 0.05 −0.08 
SOL_BD r_ 0.9–2.5 0.03 −0.08 −0.08 −0.08 0.10 0.05 0.01 −0.09 0.04 −0.07 
SOL_K r_ 0–2,000 0.10 0.07 0.07 −0.07 −0.13 0.27 −0.05 −0.13 0.05 −0.19 
CANMX v_ 0–100 49.07 24.50 24.50 12.57 30.05 82.84 58.40 25.28 34.50 80.50 
ESCO v_ 0–1 0.55 0.03 0.03 0.79 0.99 0.48 0.51 0.05 0.38 0.38 
EPCO v_ 0–1 0.77 0.39 0.39 0.51 0.23 0.81 0.66 0.59 0.55 0.55 
LAT_TIME v_ 0–180 40.14 17.29 17.29 76.41 14.64 17.48 17.26 14.49 23.05 27.00 
OV_N v_ 0.01–1 0.16 0.39 0.39 0.72 0.36 0.08 0.05 0.12 0.12 0.08 
ALPHA_BNK v_ 0–1 0.98 0.69 0.69 0.19 0.71 0.20 0.05 0.88 0.16 0.87 
CH_N2 v_ −0.01–0.3 0.13 0.02 0.02 0.09 0.13 0.23 0.07 0.14 0.04 0.09 
CH_K2 v_ −0.01–500 195.98 149.30 149.30 37.49 188.82 41.17 61.17 90.01 139.50 199.75 
SHALLST v_ 0–50,000 1,000 3,299 3,299 11,143 43,876 7,167 1,000 21,825 6,795 4,895 
REVAPMN v_ 0–500 375 122 122 240.48 324.90 304.13 500.00 417.25 127.50 488.38 
GW_REVAP v_ 0.02–0.2 0.02 0.05 0.05 0.068 0.132 0.120 0.122 0.138 0.023 0.132 
GWQMN v_ 0–5,000 1,000 1,742 1,742 2,475 2,298 480 327 1,201 870 287 
GW_DELAY v_ 0–500 31.00 47.47 47.47 11.01 52.54 76.36 103.77 37.39 43.13 80.00 
ALPHA_BF v_ 0–1 0.05 0.79 0.79 0.43 0.18 0.16 0.46 0.45 0.36 0.36 
RCHRG_DP v_ 0–1 0.05 0.09 0.09 0.03 0.06 0.05 0.05 0.09 0.02 0.05 
Parameter nameMethodSuggested rangeKaligandaki
Seti GandakiMarshyangdiBudhi GandakiTrishuli
Narayani
Q404.7Q419.1Q420Q430.5Q439.7Q445Q447Q448Q449.91Q450
CN2 r_ 35–98 0.14 0.09 0.09 −0.19 0.32 0.08 0.09 −0.05 0.11 −0.20 
PLAPS v_ −1,000–1,000 35.59 47.48 47.48 −3.38 109.13 219.67 300.00 16.28 38.20 94.12 
TLAPS v_ −10–10 −5.61 −6.61 −6.61 −5.41 −3.15 −7.23 −6.68 −7.05 −7.43 −0.89 
SFTMP v_ −20–20 – – −5.29 – −5.29 v5.29 −5.29 – – – 
SMTMP v_ −20–20 – – −3.53 – −3.53 −3.53 −3.53 – – – 
SMFMX v_ 0–20 – – 4.04 – 4.04 4.04 4.04 – – – 
SMFMN v_ 0–20 – – 5.00 – 5.00 5.00 5.00 – – – 
TIMP v_ 0–1 – – 0.10 – 0.10 0.10 0.10 – – – 
SURLAG v_ 0.05–24 – – 2.65 – 2.65 2.65 2.65 – – – 
SOL_ALB r_ 0–0.25 0.10 −0.06 −0.06 0.06 −0.10 −0.02 0.22 0.08 0.03 −0.06 
SOL_Z r_ 0–3,500 −0.06 −0.03 −0.03 −0.08 −0.08 −0.05 −0.18 −0.04 −0.08 0.04 
SOL_AWC r_ 0–1 −0.15 −0.09 −0.09 −0.14 0.89 0.04 −0.07 0.06 0.05 −0.08 
SOL_BD r_ 0.9–2.5 0.03 −0.08 −0.08 −0.08 0.10 0.05 0.01 −0.09 0.04 −0.07 
SOL_K r_ 0–2,000 0.10 0.07 0.07 −0.07 −0.13 0.27 −0.05 −0.13 0.05 −0.19 
CANMX v_ 0–100 49.07 24.50 24.50 12.57 30.05 82.84 58.40 25.28 34.50 80.50 
ESCO v_ 0–1 0.55 0.03 0.03 0.79 0.99 0.48 0.51 0.05 0.38 0.38 
EPCO v_ 0–1 0.77 0.39 0.39 0.51 0.23 0.81 0.66 0.59 0.55 0.55 
LAT_TIME v_ 0–180 40.14 17.29 17.29 76.41 14.64 17.48 17.26 14.49 23.05 27.00 
OV_N v_ 0.01–1 0.16 0.39 0.39 0.72 0.36 0.08 0.05 0.12 0.12 0.08 
ALPHA_BNK v_ 0–1 0.98 0.69 0.69 0.19 0.71 0.20 0.05 0.88 0.16 0.87 
CH_N2 v_ −0.01–0.3 0.13 0.02 0.02 0.09 0.13 0.23 0.07 0.14 0.04 0.09 
CH_K2 v_ −0.01–500 195.98 149.30 149.30 37.49 188.82 41.17 61.17 90.01 139.50 199.75 
SHALLST v_ 0–50,000 1,000 3,299 3,299 11,143 43,876 7,167 1,000 21,825 6,795 4,895 
REVAPMN v_ 0–500 375 122 122 240.48 324.90 304.13 500.00 417.25 127.50 488.38 
GW_REVAP v_ 0.02–0.2 0.02 0.05 0.05 0.068 0.132 0.120 0.122 0.138 0.023 0.132 
GWQMN v_ 0–5,000 1,000 1,742 1,742 2,475 2,298 480 327 1,201 870 287 
GW_DELAY v_ 0–500 31.00 47.47 47.47 11.01 52.54 76.36 103.77 37.39 43.13 80.00 
ALPHA_BF v_ 0–1 0.05 0.79 0.79 0.43 0.18 0.16 0.46 0.45 0.36 0.36 
RCHRG_DP v_ 0–1 0.05 0.09 0.09 0.03 0.06 0.05 0.05 0.09 0.02 0.05 

The SUFI-2 algorithm on SWAT-CUP facilitated auto-calibration, initially running 1,000 iterations and then narrowing down to the best-fit values. Nash–Sutcliffe's efficiency guided the auto-calibration. Acknowledging Abbaspour's (2022) argument that achieving the best-fit solution alone may not sufficiently represent a well-calibrated hydrological model, this study emphasizes the significance of visually examining graphical representations of observed and simulated flows and additional statistical parameters such as coefficient of determination (R2) and percentage bias (PBIAS) were concurrently assessed during graphical evaluation. This includes hydrographs, flow duration curves, scatter plots, mass curves, and water balance statistics to enhance the model's performance. Daily and monthly simulations were conducted for a comprehensive evaluation, with strict attention to maintaining physically based parameters within acceptable ranges throughout the calibration process.

Model performance

Following a rigorous assessment of data quality, the model underwent calibration and validation at 10 hydrological stations, resulting in optimized parameter values listed in Table 2. Initial default settings led to underestimations of base flow and peak values, prompting calibration of 29 parameters, revealing variable sensitivities within sub-basins. Utilizing the SUFI-2 algorithm for global sensitivity analysis and considering p-values, various parameters exhibited differing levels of sensitivity.

As per Abbaspour (2015) in the SWAT-CUP user manual, parameters with a p-value below 0.05 signify higher sensitivity. For instance, at the calibration point of Q439.7, PLAPS, CN2, LAT_TIME, SOL_BD (..), TLAPS, GW_REVAP, ALPHA_BNK, ALPHA_BF, SOL_K (..), GW_DELAY, EPCO, ESCO, and OV_N demonstrated heightened sensitivity. Different calibration points showcased distinct sets of sensitive parameters. The majority of climatic stations were strategically positioned within the central basin (see Figure 1). Notably, PLAPS and TLAPS emerged as the most sensitive parameters for the Trishuli and Marshyangdi Basins, considering the orographic impact of elevation-based rainfall and temperature. Furthermore, in rivers exhibiting lower baseflows such as Myagdi Q404.7 and Tadi Q448, LAT_TIME and ALPHA_BF displayed higher sensitivity. Manning's n values for the main channel (CH_N2) and overland (OV_N) were fine-tuned based on land use classes. Similar studies by Marahatta et al. (2021) in the Budhi Gandaki Basin and Bajracharya et al. (2018) in the Kaligandaki Basin have also highlighted the sensitivity of these parameters.

After finalizing the best-optimized parameters from SWAT-CUP, the SWAT backup (.mdb) file is updated with the optimized values. Subsequently, graphical evaluations for all calibration points are conducted. Detailed performance assessments of the model in primary tributaries are presented in the following sections.

Figure 5 shows a visual comparison between the observed and simulated flow at the Narayani main outlet located in Devghat station (Q450). A comparable analysis was conducted for each calibration station and presented in Annex I.
Figure 5

Model performance at Narayani River main outlet at Devghat Station Q450. (a) Daily hydrograph of observed and simulated flow; (b) monthly hydrograph of observed and simulated flow; (c) daily flow duration curve; and (d) scatter plot of the daily observed versus simulated flow.

Figure 5

Model performance at Narayani River main outlet at Devghat Station Q450. (a) Daily hydrograph of observed and simulated flow; (b) monthly hydrograph of observed and simulated flow; (c) daily flow duration curve; and (d) scatter plot of the daily observed versus simulated flow.

Close modal

As the Narayani is the junction of seven main tributaries, most downstream observed stations are available at Devghat with station number Q450. The discharge of the NRB is a compounding form of snow melt, surface runoff, groundwater flow and rainfall. After considering the 40 years of daily data with 5 years of warm-up period to simulate the model, the daily and monthly statistical indicators are presented in Table 3. These indicators as suggested by Moriasi et al. (2007) show that the model performance is above satisfactory level for all calibration stations.

Table 3

Performance indication for the model calibration and validation

Sub-basinStation nameCalibration
Validation
Daily
Monthly
Daily
Monthly
R2NSEPBIASi2NSEPBIASR2NSEPBIASR2NSEPBIAS
Kaligandaki Myagdi Khola Q404.7 0.7 0.7 −5.5 0.83 0.83 −6 0.55 0.51 −23.5 0.63 0.57 −24.6 
Kali Gandaki Angsing Q419.1 0.87 0.86 −12.3 0.93 0.92 −12.3 0.85 0.73 −25.2 0.93 0.8 −25.3 
Kaligandaki Kotagaun Q420 0.81 0.81 −8.5 0.85 0.93 −12.6 0.83 0.78 −10.3 0.95 0.94 −10.1 
Seti Gandaki Seti Gandaki Damauli Q430.5 0.67 0.64 −14.9 0.81 0.76 −14.8 0.78 0.74 22 0.88 0.83 −22.1 
Marshynagdi Marshyandgi Bimalnagar Q439.7 0.79 0.77 −15.1 0.91 0.87 −15.7 0.79 0.73 −26 0.9 0.81 −25 
Budhi Gandaki Budhi Gandaki Arughat Q445 0.82 0.8 −7.6 0.88 0.86 −7.5 0.85 0.84 8.8 0.9 0.89 
Trishuli Trishuli Betrawati Q447 0.74 0.73 −8.1 0.82 0.79 −8 0.71 0.68 −11.2 0.78 0.74 −11.1 
Tadi Belkot Q448 0.76 0.75 −13.1 0.92 0.91 −13.3 0.76 0.69 7.5 0.92 0.85 7.4 
Trishuli Kalikhola Q449.91 0.85 0.71 −24.7 0.92 0.77 −24.6 0.79 0.77 16.1 0.85 0.83 16.2 
Narayani Narayani Devghat Q450 0.93 0.9 −12.2 0.96 0.93 −12.1 0.88 0.87 11.73 0.92 0.9 −15.4 
Sub-basinStation nameCalibration
Validation
Daily
Monthly
Daily
Monthly
R2NSEPBIASi2NSEPBIASR2NSEPBIASR2NSEPBIAS
Kaligandaki Myagdi Khola Q404.7 0.7 0.7 −5.5 0.83 0.83 −6 0.55 0.51 −23.5 0.63 0.57 −24.6 
Kali Gandaki Angsing Q419.1 0.87 0.86 −12.3 0.93 0.92 −12.3 0.85 0.73 −25.2 0.93 0.8 −25.3 
Kaligandaki Kotagaun Q420 0.81 0.81 −8.5 0.85 0.93 −12.6 0.83 0.78 −10.3 0.95 0.94 −10.1 
Seti Gandaki Seti Gandaki Damauli Q430.5 0.67 0.64 −14.9 0.81 0.76 −14.8 0.78 0.74 22 0.88 0.83 −22.1 
Marshynagdi Marshyandgi Bimalnagar Q439.7 0.79 0.77 −15.1 0.91 0.87 −15.7 0.79 0.73 −26 0.9 0.81 −25 
Budhi Gandaki Budhi Gandaki Arughat Q445 0.82 0.8 −7.6 0.88 0.86 −7.5 0.85 0.84 8.8 0.9 0.89 
Trishuli Trishuli Betrawati Q447 0.74 0.73 −8.1 0.82 0.79 −8 0.71 0.68 −11.2 0.78 0.74 −11.1 
Tadi Belkot Q448 0.76 0.75 −13.1 0.92 0.91 −13.3 0.76 0.69 7.5 0.92 0.85 7.4 
Trishuli Kalikhola Q449.91 0.85 0.71 −24.7 0.92 0.77 −24.6 0.79 0.77 16.1 0.85 0.83 16.2 
Narayani Narayani Devghat Q450 0.93 0.9 −12.2 0.96 0.93 −12.1 0.88 0.87 11.73 0.92 0.9 −15.4 

For instance, At Q450, a negative PBIAS during calibration indicates an overestimation in simulated volume, contrasting with the model's underestimation of runoff volume during validation. Across other calibration points, the model tends to overestimate volume during calibration, while at certain points, it underestimates runoff volume during validation.

All sub-basins, except Myagdi Mangalaghat (Q404.7) and Tadi (Q448), receive snowmelt discharge. The graphical evaluation of each calibration point, as illustrated in Figure 5 for Q450 and ANNEX I, reveals a challenge in capturing peak events during calibration and validation. Despite the PBIAS being above the satisfactory level, it is concluded that the calibrated model excels in reproducing average flow estimates.

Spatial variation of water balance

There are four significant hydrological components in the water balance, namely precipitation (PPT), actual evapotranspiration (AET), net water yield (NWY), and change in storage (Δ storage) The trend of NYW and PPT may not be identical because of factors such as rainfall intensity, soil properties, and land cover characteristics (Bharati et al. 2014). Furthermore, the change in storage is an umbrella term encompassing groundwater recharge, changes in soil moisture storage in the unsaturated zone, and inaccuracies in the model.

The hydrological model has been used to simulate the distribution of significant water balance factors within the NRB on a sub-basin level. This simulation covers the period from 1980 to 2019 and includes a warm-up period of 5 years. The mean annual precipitation is 1,902 mm, the evapotranspiration is 413 mm about 22% of precipitation, and the net water yield is 1,432.45 mm about 75% of annual precipitation for the entire basin; however, all components of water balance vary spatially as well as temporally. The net water yield (NWY) indicates the union of surface runoff, sub-surface runoff, groundwater flow with consideration of transmission, and pond abstractions (Bharati et al. 2019; Jibesh et al. 2021). Out of this, 44% comes from lateral flow, 38% from surface flow, and 16% from groundwater flow, contributing to the NWY. Due to the steep topography and intense rainfall for short monsoon period the lateral flow contribution is higher than other components.

From the insights of total water yield, at the Tibetan Plateau, the lateral flow contributes 29%, the groundwater contributes 20%, and the surface runoff contributes 45% to the total water yield; similarly, in the High Himalayas, the contribution from the lateral flow is high (49%) and the groundwater flow is low (8%) to the net water yield. Also, in the high mountain, 64% from lateral flow, 8% from groundwater flow and 27% from surface runoff as well as in the middle mountain, 51% from lateral flow and 14% from groundwater flow and 34% from surface runoff contribution to the water yield. However, in the Siwalik, only 30% of the contribution is from lateral flow, 45% of endowment is from groundwater flow and 21% is from surface runoff to the total water yield. Here, the lateral flow contribution at the high Himalayas, high mountains, and the middle mountain is higher than Tibetan plateau and Siwalik; this is because of steep topography and intense rainfall for short periods and the groundwater contribution is higher in the Siwalik zone due to the sub-horizontal topography (Andermann et al. 2012; Cochand et al. 2019). In a study by Yao et al. (2021) on the Himalayas River, the groundwater contribution is inversely proportional to topographic slope and rainfall intensity. Also, in our case, from the upstream Himalayan region to the downstream Siwalik region, the groundwater is increasingly contributed.

Moreover, the surface runoff is higher in the Tibetan Plateau because of the high snow pack and glaciers. On the other hand, surface water contribution is low at the Siwalik; this is due to flat topography and also the groundwater is contributing high, which shows that the higher rate of groundwater recharge is during the monsoon period in the Siwalik zone.

Figure 6 shows the spatial distribution of PPT, AET, and NWY. From the investigation, the central part of the basin receives more rainfall (>5,000 mm); however, less precipitate (<200 mm) in the northern part is the back side of the Himalayan range. Similarly, the AET is declining toward the north. As the ET is a respective function of rainfall, land cover, and temperature, warmer climate and most of the area are covered by forest and cropland in the south. More than 650 mm from the south to less than 200 mm in the north evaporate the moisture.
Figure 6

Sub-basin level of average annual precipitation (PPT), actual evapotranspiration (AET), and net water yield (NWY) for Narayani Basin.

Figure 6

Sub-basin level of average annual precipitation (PPT), actual evapotranspiration (AET), and net water yield (NWY) for Narayani Basin.

Close modal
Here, Figure 7 shows the components of the water balance distribution of each sub-basin. The negative sign for storage indicates the contribution to the groundwater recharge (Bharati et al. 2014) and here also the amount of storage is higher in the glacier zone which is due to freezing temperature, and the rainfall is stored in the form of ice at the northern part. Similarly, Δ storage is positive in the southern part, which means that groundwater is contributing to the water yield.
Figure 7

Sub-basin wise water balance component, precipitation is higher in the central part of the basin, storage is higher in the northern part of the basin, and evapotranspiration is higher in the southern part of the basin.

Figure 7

Sub-basin wise water balance component, precipitation is higher in the central part of the basin, storage is higher in the northern part of the basin, and evapotranspiration is higher in the southern part of the basin.

Close modal

Temporal variation of water balance

There is a high variation of monthly as well as seasonal variation of water balance. Figure 8 indicates the monthly average variation of the water balance within the basin. For May, June, July, and August, the storage value is indicating negative within the basin; this means that groundwater is recharging in that month. And similarly, Δ storage is positive for the dry month, which indicates that the groundwater is contributing to the stream flow. Since the wet season is warmer, AET is also higher. Moreover, net water yield is a respective function of precipitation and hence followed a similar trend.
Figure 8

Average monthly water balance for baseline period (1980–2019) in the NRB.

Figure 8

Average monthly water balance for baseline period (1980–2019) in the NRB.

Close modal

In view of the fact that the soil is fully saturated in August, the delta storage is lower than in July and even positive in September, even though the rainfall and water yield are about too similar in that month. This means that the recharging capacity is reduced in August and after September the ground water is donative to the water yield.

Similarly, Figure 9 shows the seasonal variation of three main components of water balance. The pre-monsoon month (March–May), also a warmer month, receives less than 500 mm in most of the area; however, the northern part of the basin where the high snowpack and glacier perceive rainfall above 600 mm. While the central to the southern part of the basin evaporates more moisture greater than 150 mm and also the water yield is very low, less than 250 mm for most of the basin.
Figure 9

Seasonal distribution of precipitation, actual evapotranspiration, and net water yield, (a) precipitation, (b) actual evapotranspiration, and (c) water yield.

Figure 9

Seasonal distribution of precipitation, actual evapotranspiration, and net water yield, (a) precipitation, (b) actual evapotranspiration, and (c) water yield.

Close modal

In addition, for the monsoon period (June to September), the central parts are wetter than the southern to northern part. While the central part receives more rain above 2,000 mm for four monsoon months, the back side of the Himalayas receives little rain less than 500 mm. Also, AET is higher in the southern part, greater than 250 mm and very low (<50 mm) in the Higher Himalayas. Moreover, the water yield follows the rainfall pattern and hence, a greater amount of water yield from the central part only less than 500 mm water yield from the back part of the Himalayas. There is a similar trend for post-monsoon and winter seasons, less precipitate as well as AET and water yield. Even if the temperature is higher in pre-monsoon, the actual ET is greatest in monsoon.

This study employed a multi-site approach to calibrate and validate multiple parameters, aiming to understand the spatial and temporal variations in water balance within the Narayani Basin. Calibration at 10 separate sites produced diverse statistical outcomes influenced by orographic effects on rainfall and temperature, alongside data gaps during specific periods. Employing a multi-site calibration strategy in hydrological modeling aids in reducing the uncertainty associated with the model and its parameters, thus enhancing accuracy. The sensitivity of parameters varied among calibration sub-basins, predominantly relying on available observed data, topography, land use, etc. For instance, PLAPS and TLAPS exhibited increased sensitivity in regions with limited rainfall and temperature measurements, while areas with significant snowmelt demonstrated sensitivity in snow contribution parameters. Notably, parameters like CN2, ALPHA_BF, and ALPHA_BNK played pivotal roles in optimizing the model, considering the basin's monsoon-fed nature. Furthermore, both statistical indicators and graphical evaluations demonstrated the model's satisfactory performance.

The study estimated that the average annual precipitation in the basin is 1,902 mm, with evapotranspiration at 413 mm (representing approximately 22% of precipitation), and net water yield at 1,432.45 mm (representing approximately 75% of annual precipitation). However, all components of the water balance vary spatially and temporally. Specifically, 44% of the net water yield comes from lateral flow, 38% from surface flow, and 16% from groundwater flow.

In regions with mountainous slopes and high-intensity rainfall, lateral flow can play a significant role in the movement of water across the landscape. Due to the steep topography, surface runoff can occur rapidly, with water moving downhill and forming streams and rivers. Additionally, short but intense monsoon rainfall periods can further increase the amount of water moving laterally, as the soil may not be able to absorb all of the water that falls during these events. This can result in flooding, landslides, and erosion, particularly in areas with steep slopes or where land use practices have disrupted the natural water cycle. In such regions, it is crucial to understand the dynamics of lateral flow and its interaction with surface and groundwater flows, in order to manage water resources effectively and minimize the negative impacts of extreme weather events. This may involve implementing measures, such as erosion control, terracing, and the construction of retention ponds or other forms of water storage infrastructure.

This well-calibrated hydrological model demonstrates significant potential for estimating flow at each sub-basin level, particularly in areas lacking observed data. Its utility extends to studies concerning water availability for irrigation, hydropower, and water supply projects, offering an alternative to the use of empirical formulas. Other output data from the model like soil moisture storage, and potential evapotranspiration at the local level would help to identify the crop water requirement.

Furthermore, its application is not limited to specific sectors; it also aids in developing water resources management strategies. Additionally, the model's adaptability enables assessments of climate and land use changes, providing crucial insights into potential alterations in water availability. Consequently, the model's outputs serve as invaluable tools for water resource policymakers and managers, facilitating the design of basin and location-specific policies. These policies foster the sustainable utilization of water resources, contributing significantly to national development.

The Department of Hydrology and Meteorology, Government of Nepal are highly acknowledged for providing the hydro-meteorological data. Moreover, cordial thanks are extended to Professor Vishnu Prasad Pandey, Tribhuvan University, Nepal and Assistant Professor Rocky Talchabhadel, Jackson State University: Jackson, Mississippi, US for their valuable comments that greatly improved the quality of this paper.

N.D. is responsible for overall structuring of the contents, data analysis, conceptualizing flow of contents, and drafting. S.L. is responsible for overall guidance, review and feedback/input for improving the overall quality of the entire manuscript. P.K.B. is responsible for overall guidance, review and feedback/input for improving the overall quality of the entire manuscript.

This research is supported by the University Grants Commission Nepal under the PhD fellowship for the first author, awarded with fellowship number PhD-79/80-Engg-08.

All relevant data are available from an online repository or repositories.

The authors declare there is no conflict.

Abbaspour
K. C.
2015
SWAT – CUP SWAT Calibration and Uncertainty Programs
.
Abbaspour
K. C.
2022
The fallacy in the use of the ‘best-fit’ solution in hydrologic modeling
.
Science of the Total Environment
802
,
149713
.
https://doi.org/10.1016/j.scitotenv.2021.149713
.
Agarwal
A.
,
Babel
M. S.
&
Maskey
S.
2014
Analysis of future precipitation in the Koshi river basin, Nepal
.
Journal of Hydrology
513
,
422
434
.
https://doi.org/10.1016/j.jhydrol.2014.03.047
.
Amini
A.
,
Karami
F.
,
Sedri
M. H.
&
Khaledi
V.
2022
Determination of water requirement and crop coefficient for strawberry using lysimeter experiment in a semi-arid climate
.
H2Open Journal
5
(
4
),
642
655
.
https://doi.org/10.2166/h2oj.2022.030
.
Andermann
C.
,
Longuevergne
L.
,
Bonnet
S.
,
Crave
A.
,
Davy
P.
&
Gloaguen
R.
2012
Impact of transient groundwater storage on the discharge of Himalayan rivers
.
Nature Geoscience
5
(
2
),
127
132
.
https://doi.org/10.1038/ngeo1356
.
Arnold
J. G.
,
Srinivasan
R.
,
Muttiah
R. S.
&
Williams
J. R.
1998
Large area hydrological modeling and assessment part I: Model development
. JAWRA
34
(
1
),
73
89
.
Bajracharya
A. R.
,
Bajracharya
S. R.
,
Shrestha
A. B.
&
Maharjan
S. B.
2018
Climate change impact assessment on the hydrological regime of the Kaligandaki Basin, Nepal
.
Science of the Total Environment
625
,
837
848
.
https://doi.org/10.1016/j.scitotenv.2017.12.332
.
Bharati
L.
,
Gurung
P.
,
Jayakody
P.
,
Smakhtin
V.
&
Bhattarai
U.
2014
The projected impact of climate change on water availability and development in the Koshi Basin, Nepal
.
Mountain Research and Development
34
(
2
),
118
130
.
https://doi.org/10.1659/MRD-JOURNAL-D-13-00096.1
.
Bharati
L.
,
Bhattarai
U.
,
Khadka
A.
,
Gurung
P.
,
Neumann
L. E.
,
Penton
D. J.
,
Dhaubanjar
S.
&
Nepal
S.
2019
From the mountains to the plains: Impact of climate change on water resources in the Koshi River Basin. In IWMI Working Papers 187. https://doi.org/10.5337/2019.205
.
Bhatta
B.
,
Shrestha
S.
,
Shrestha
P. K.
&
Talchabhadel
R.
2019
Evaluation and application of a SWAT model to assess the climate change impact on the hydrology of the Himalayan River Basin
.
CATENA
181
,
104082
.
https://doi.org/10.1016/j.catena.2019.104082
.
Bhattarai
S.
,
Zhou
Y.
,
Man Shakya
N.
&
Zhao
C.
2018
Hydrological modelling and climate change impact assessment using HBV light model: A case study of Narayani River Basin, Nepal
.
Nature Environment and Pollution Technology
17
(
2007
),
691
702
.
Brirhet
H.
&
Benaabidate
L.
2016
Comparison of two hydrological models (lumped and distributed) over a pilot area of the Issen watershed in the Souss Basin, Morocco
.
European Scientific Journal, ESJ
12
(
18
),
347
.
https://doi.org/10.19044/esj.2016.v12n18p347
.
Chand
M.
,
Bhattarai
B.
,
Baral
P.
&
Pradhananga
N.
2019
Trend analysis of temperature data for Narayani River Basin, Nepal
.
Sci
1
(
1
),
21
.
https://doi.org/10.3390/sci1010021
.
Cochand
M.
,
Christe
P.
,
Ornstein
P.
&
Hunkeler
D.
2019
Groundwater storage in high alpine catchments and its contribution to streamflow
.
Water Resources Research
55
(
4
),
2613
2630
.
https://doi.org/10.1029/2018WR022989
.
Dahal
V.
,
Shakya
N. M.
&
Bhattarai
R.
2016
Estimating the impact of climate change on water availability in Bagmati Basin, Nepal
.
Environmental Processes
3
(
1
),
1
17
.
https://doi.org/10.1007/s40710-016-0127-5
.
Dandekhya
S.
,
England
M.
,
Ghate
R.
,
Goodrich
C. G.
,
Nepal
S.
,
Prakash
A.
,
Shrestha
A.
,
Singh
S.
,
Singh
M.
&
Udas
P. B.
2017
The Gandaki Basin Maintaining Livelihoods in the Face of Landslides, Floods, and Drought A ARE HI Himalayan Adaptation, Water and Resilience Research
.
.
Dawadi
B.
,
Acharya
R. H.
,
Pudasainee
S.
,
Lamichhane
D.
&
Shrestha
I. K.
2020
Observed trends and spatial distribution in daily precipitation indices of extremes over the Narayani River Basin, Central Nepal
.
Applied Ecology and Environmental Sciences
8
(
3
),
106
118
.
https://doi.org/10.12691/aees-8-3-6
.
Desai
S.
,
Singh
D. K.
,
Islam
A.
&
Sarangi
A.
2021
Multi-site calibration of hydrological model and assessment of water balance in a semi-arid river basin of India
.
Quaternary International
571
,
136
149
.
https://doi.org/10.1016/j.quaint.2020.11.032
.
Devkota
L. P.
&
Gyawali
D. R.
2015
Impacts of climate change on hydrological regime and water resources management of the Koshi River Basin, Nepal
.
Journal of Hydrology: Regional Studies
4
,
502
515
.
https://doi.org/10.1016/j.ejrh.2015.06.023
.
Gambia
T.
,
Van Engelen
V.
&
Huting
J.
2008
Global Assessment of Land Degradation Soil and Landform Properties for LADA Partner Countries. October
.
Ghobadi
F.
&
Kang
D.
2023
Application of machine learning in water resources management: A systematic literature review
.
Water (Switzerland)
15
(
4
).
https://doi.org/10.3390/w15040620
.
Gleick
P. H.
2000
A look at twenty-first century water resources development
.
Water International
25
(
1
),
127
138
.
https://doi.org/10.1080/02508060008686804
.
Huting
J.
2009
Soil and Terrain database for Nepal. June
.
ICIMOD 2013 Land cover of Nepal 2010 [Data set]. ICIMOD, Lalitpur, Nepal. https://doi.org/10.26066/rds.9224
.
JAXA EORC
2021
ALOS Global Digital Surface Model (DSM). 1.2, 1–21. https://www.eorc.jaxa.jp/ALOS/en/aw3d30/
Jibesh
K. C.
,
Dhaubanjar
S.
,
Pandey
V. P.
&
Subedi
R.
2021
Water balance component analysis of a spring catchment of western Nepal
.
Banko Janakari
31
(
1
),
23
32
.
https://doi.org/10.3126/banko.v31i1.37341
.
Kouchi
D. H.
,
Esmaili
K.
,
Faridhosseini
A.
,
Sanaeinejad
S. H.
,
Khalili
D.
&
Abbaspour
K. C.
2017
Sensitivity of calibrated parameters and water resource estimates on different objective functions and optimization algorithms
.
Water (Switzerland)
9
(
6
),
1
16
.
https://doi.org/10.3390/w9060384
.
Mahmudi
P.
,
Motamedvaziri
B.
,
Hosseini
M.
,
Ahmadi
H.
&
Amini
A.
2021
Study of climate change effects on hydrological processes in Siminehroud and Zarrinehroud watersheds northwest of Iran
.
Earth Science Informatics
14
(
2
),
965
974
.
https://doi.org/10.1007/s12145-021-00598-2
.
Malik
M. A.
,
Dar
A. Q.
&
Jain
M. K.
2022
Modelling streamflow using the SWAT model and multi-site calibration utilizing SUFI-2 of SWAT-CUP model for high altitude catchments, NW Himalaya's
.
Modeling Earth Systems and Environment
8
(
1
),
1203
1213
.
https://doi.org/10.1007/s40808-021-01145-0
.
Marahatta
S.
,
Devkota
L. P.
&
Aryal
D.
2021
Application of swat in hydrological simulation of complex mountainous river basin (Part I: Model development)
.
Water (Switzerland)
13
(
11
).
https://doi.org/10.3390/w13111546
.
Molina-Navarro
E.
,
Andersen
H. E.
,
Nielsen
A.
,
Thodsen
H.
&
Trolle
D.
2017
The impact of the objective function in multi-site and multi-variable calibration of the SWAT model
.
Environmental Modelling and Software
93
,
255
267
.
https://doi.org/10.1016/j.envsoft.2017.03.018
.
Moriasi
D. N.
,
Arnold
J. G.
,
Van Liew
M. W.
,
Bingner
R. L.
,
Harmel
R. D.
&
Veith
T. L.
2007
Model evaluation guidelines for systematic quantification of accuracy in watershed simulations
.
American Society of Agricultural and Biological Engineers
50
(
3
),
885
900
.
Neitsch
S. L.
,
Arnold
J. G.
,
Kiniry
J. R.
,
Srinivasan
R.
&
Williams
J. R.
2002
Soil and Water Assessment Tool User's Manual
.
Pandey
V. P.
,
Dhaubanjar
S.
,
Bharati
L.
&
Thapa
B. R.
2019
Hydrological response of Chamelia watershed in Mahakali Basin to climate change
.
Science of the Total Environment
650
,
365
383
.
https://doi.org/10.1016/j.scitotenv.2018.09.053
.
Pandey
V. P.
,
Dhaubanjar
S.
,
Bharati
L.
&
Thapa
B. R.
2020a
Spatio-temporal distribution of water availability in Karnali-Mohana Basin, Western Nepal: Climate change impact assessment (Part-B)
.
Journal of Hydrology: Regional Studies
29
,
100691
.
https://doi.org/10.1016/j.ejrh.2020.100691
.
Pandey
V. P.
,
Dhaubanjar
S.
,
Bharati
L.
&
Thapa
B. R.
2020b
Spatio-temporal distribution of water availability in Karnali-Mohana Basin, Western Nepal: Hydrological model development using multi-site calibration approach (Part-A)
.
Journal of Hydrology: Regional Studies
29
,
100690
.
https://doi.org/10.1016/j.ejrh.2020.100690
.
Pandey
V. P.
,
Shrestha
D.
,
Adhikari
M.
&
Shakya
S.
2020c
Streamflow alterations, attributions, and implications in extended east Rapti watershed, central-southern Nepal
.
Sustainability (Switzerland)
12
(
9
).
https://doi.org/10.3390/su12093829
.
Paulhus
J. L. H.
&
Kohler
M. A.
1952
Interpolation of missing precipitation records
.
Monthly Weather Review
80
(
8
),
129
133
.
https://doi.org/10.1175/1520-0493(1952)080&lt;0129:IOMPR > 2.0.CO;2
.
Phan
D. B.
,
Wu
C. C.
&
Hsieh
S. C.
2011
Impact of climate change on stream discharge and sediment yield in Northern Viet Nam
.
Water Resources
38
(
6
),
827
836
.
https://doi.org/10.1134/S0097807811060133
.
SalimiRad
H.
,
Dehvari
A.
,
Galavi
H.
&
Ebrahimian
M.
2020
Parameter identification and uncertainty analysis of SWAT in Kardeh streamflow simulation
.
Iran-Water Resources Research
16
(
3
),
212
221
. .
Shope
C. L.
,
Maharjan
G. R.
,
Tenhunen
J.
,
Seo
B.
,
Kim
K.
,
Riley
J.
,
Arnhold
S.
,
Koellner
T.
,
Ok
Y. S.
,
Peiffer
S.
,
Kim
B.
,
Park
J. H.
&
Huwe
B.
2014
Using the SWAT model to improve process descriptions and define hydrologic partitioning in South Korea
.
Hydrology and Earth System Sciences
18
(
2
),
539
557
.
https://doi.org/10.5194/hess-18-539-2014
.
Shrestha
S. R.
&
Bajracharya
B.
2011
The status of glaciers in The Hindu Kush–Himalayan region
.
Mountain Research and Development
33
(
1
),
114
115
.
http://dx.doi.org/10.1659/mrd.mm113
.
Talchabhadel
R.
,
Aryal
A.
,
Kawaike
K.
,
Yamanoi
K.
,
Nakagawa
H.
,
Bhatta
B.
&
Karki
S.
2021
Evaluation of precipitation elasticity using precipitation data from ground and satellite-based estimates and watershed modeling in Western Nepal
.
Journal of Hydrology: Regional Studies
33
,
100768
.
https://doi.org/10.1016/j.ejrh.2020.100768
.
Tijerina
D.
,
Condon
L.
,
FitzGerald
K.
,
Dugger
A.
,
O'Neill
M. M.
,
Sampson
K.
,
Gochis
D.
&
Maxwell
R.
2021
Continental hydrologic intercomparison project, phase 1: A large-scale hydrologic model comparison over the continental United States
.
Water Resources Research
57
(
7
),
e2020WR028931
.
https://doi.org/10.1029/2020WR028931
.
Wada
Y.
,
Bierkens
M. F. P.
,
De Roo
A.
,
Dirmeyer
P. A.
,
Famiglietti
J. S.
&
Hanasaki
N.
2017
Human–water interface in hydrological modelling: Current status and future directions
. Hydrology and Earth System Sciences 21,
4169
4193
.
WECS. (Water and Energy Commission Secretariat)
.
2011
Water Resources of Nepal in the Context of Climate Change. 68. https://wwfeu.awsassets.panda.org/downloads/water_resources_of_nepal_final_press_design.pdf.
Yanto
L.
,
Rajagopalan
B.
&
Kasprzyk
J.
2017
Hydrological model application under data scarcity for multiple watersheds, Java Island, Indonesia
.
Journal of Hydrology: Regional Studies
9
,
127
139
.
https://doi.org/10.1016/j.ejrh.2016.09.007
.
Yao
Y.
,
Zheng
C.
,
Andrews
C. B.
,
Scanlon
B. R.
,
Kuang
X.
,
Zeng
Z.
,
Jeong
S. J.
,
Lancia
M.
,
Wu
Y.
&
Li
G.
2021
Role of groundwater in sustaining Northern Himalayan Rivers
.
Geophysical Research Letters
48
(
10
),
1
10
.
https://doi.org/10.1029/2020GL092354
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).

Supplementary data