Hydroclimatic Impact Assessment Using the SWAT Model in India—State of the Art Review

Abstract

The Soil and Water Assessment Tool (SWAT) has been widely employed to assist with decision making and management planning for assessing and mitigating the impact of climate change. This model has gained popularity in India as the country is facing increasing water issues under projected climate changes. However, a systematic review of the literature that discusses the applicability of the model, the impact assessment process, and the interpretation of the modeling results in India remains lacking. We synthesized and reviewed 110 recent SWAT modeling studies (published from 2012 to 2022) that evaluated the impact of future climate change on water resources in India to identify research gaps that need to be filled to advance SWAT modeling practices for impact assessments. The review revealed that the SWAT model provided acceptable accuracy statistics in most (90%) of the studies reviewed. Half of these studies identified the base curve number (CN2) as the parameter to which the water balance is the most sensitive; thus, this parameter was included in the calibration process. The accuracy of SWAT modeling is closely associated with the accuracy of the weather data fed to the model. However, extreme events, including heavy storm events and severe droughts, were rarely considered in climate change impact assessments using the SWAT model. Most studies downscaled global-scale climate modeling outputs to local weather stations when applying the SWAT model using various methods, such as the delta change method, multiple linear regression method, gamma–gamma transformation, fitted histogram equalization, and quantile mapping. Further, most studies investigated the performance of the SWAT model before applying the model to quantify the future hydrological consequences of projected climate change in a subsequent scenario analysis. This review suggests that further evaluations of the characteristics and development processes of existing climate data products are needed to effectively consider extreme events in impact assessments. In addition, this review finds that climate change impact modeling has been improved with advances in climate projection preparation, including ensemble averaging, bias correction, and downscaling methods. This regional review of current SWAT modeling practices for climate change impact assessments can be used to create reliable future hydrological projections in India.

1. Introduction

Climate change poses an unprecedented and multifaceted challenge to freshwater ecosystems and water supplies in many parts of the world. The average global surface air temperature increased by 0.74 °C during the 20th century, surpassing historical trends [1]. This increase will result in unfavorable or favorable changes in agriculture and water resources, profoundly impacting the livelihoods of large populations. Notably, the impacts of climate change vary depending on the landscape and human activities [2]. As such, a critical consideration of the ramifications of climate change for water resources in different river basins is imperative for policymakers and water resource planners [3,4]. A wealth of evidence underscores that climate change acutely disrupts the water balance of river basins, influencing both water quantities within basins and the availability of water in surrounding areas [5,6,7]. Consequently, researchers endeavor to elucidate the impacts of climate change on water systems at the basin scale, thereby understanding the comprehensive dimensions of this complex phenomenon [8,9,10,11].

A hydrological model can be used to demonstrate the potential impacts of expected changes in climate and water resource management. Among an array of such models, the Soil and Water Assessment Tool is a widely adopted framework which caters to various spatio-temporal scales and diverse agroclimatic scenarios [12]. Recent studies successfully applied the SWAT model to quantify the impacts of climate change on water resources [13,14]. Numerous studies have demonstrated the model’s capacity to accurately represent essential basin-level variables, including streamflow, groundwater dynamics, sediment yield, nutrient information, and vegetation/crop growth [10,13,15]. Leveraging global and regional climate models (GCMs and RCMs, respectively), the SWAT model has been widely used to simulate the basin water balance for different climate change scenarios [10,11,16].

Numerous reviews have comprehensively examined the potential and challenges associated with the use of the SWAT model for hydrological modeling across diverse landscapes and climate zones [17,18,19,20,21,22]. Arnold et al. (1998) emphasized the importance of the hydro-climatic impact assessment in the realm of water resources management and provided an informative overview of its development and application in large-scale hydrological modeling [23]. Abbaspour et al. (2015) reported the hydrological impacts of climate change to assess the importance of water resources management and disaster risk reduction, using the SWAT model across the European continent [19]. Tan et al. (2019) reviewed the SWAT model’s applications to evaluate the impacts of hydroclimatic extremes in Southeast Asia and discussed associated challenges and future research directions [24]. Given the increasing global application of the SWAT model, the compilation of the results from SWAT studies can provide valuable insights for developers and new users, offering new perspectives on how climate variability affects hydrological processes and water resources. These compilations also help identify the critical applications, capabilities, challenges, and limitations of using the SWAT in different regions. This, in turn, enhances our understanding of the potential consequences of climate and water resource changes.

In general, GCMs have been extensively used in SWAT modeling to project future hydrological processes to understand how climate change will affect the quality and quantity of water in the future [11,25,26,27]. Githui et al. (2009) employed the SWAT model and a GCM to estimate streamflow in the Nzoia catchment within the Lake Victoria Basin, providing valuable information about variations in the streamflow of the basin under different climate scenarios [28]. Shrestha et al. (2013) assessed future changes in sediment yield in the Nam Ou Basin in northern Laos, employing four GCMs to analyze historical and future periods for the A2 and B2 scenarios [29]. Similarly, Neupane et al. (2014) evaluated the effect of climate change on the streamflow discharge of the Kali Gandaki watershed in the central Himalayan region, predicting daily discharge using three different emission scenarios derived from 16 GCMs [30]. Such empirical studies must be reviewed to ensure that the insights can be applied to other regions with similar climatic and environmental conditions, utilizing the same modeling approach.

India provides a highly suitable testing ground for reviewing SWAT model studies for several reasons. India experiences significant impacts due to environmental change on the quantity and quality of water owing to its high population and intensified agricultural activities [31,32]. Moreover, its diverse range of geographical and climatic conditions, spanning from the north of the country to the south, offers an ideal setting for assessing the impacts of climate change on water resources. Notably, the River Basin of India stands out as one of the most vulnerable areas to climate change, highlighting the importance of conducting climate change studies using the SWAT model to quantitatively assess its potential impacts on water resources [33]. However, studies reviewing SWAT applications in the Indian River region are lacking. Therefore, detailed reviews of SWAT applications in the sub-basins and watersheds of Indian rivers—with a focused examination of future changes in the hydrological environment due to climate change—can improve our understanding of the model’s applicability and provide a more robust interpretation of the impacts of climate change.

In this study, we comprehensively reviewed state-of-the-art SWAT-based climate change studies in India, aiming to address the potential impacts of climate change on water resources issues (such as water quality, drought, and floods) and the hydrologic responses in river basins under different climatic conditions. We aimed to highlight existing SWAT applications reported in peer-reviewed journals, emphasizing their significance in simulating key variables like streamflow, water yield, sediment conditions, and nutrient cycles in Indian river basins. We summarized the main findings of SWAT applications in Indian river basins and quantitatively assessed the impacts of climate change on the hydrology and water resources of several Indian river basins. In addition, we evaluated the current availability of data, research challenges, and potential future research topics related to SWAT modeling in the context of climate change in India.

2. A General Framework of SWAT-Based Climate Change Studies

The SWAT model utilizes combinations of atmospheric data and internal model records to estimate various variables [34,35]. The inputs required for the SWAT model include climatic parameters (such as daily rainfall, maximum/minimum air temperature, solar radiation, wind speed, and relative humidity) as well as information about soil, land use, slope, and nutrients.

This paper presents a comprehensive review of hydroclimatic impact studies conducted on Indian river basins. Additionally, review articles on Indian river basins published from 2012 to 2022 were considered. The included studies were primarily retrieved from the SWAT literature database (https://www.card.iastate.edu/swat_articles/ accessed on 23 August 2023) and the Clarivate Analytics Web of Science database. Various search databases, including Web of Science, Scopus, Google Scholar, and the SWAT Database, were used to identify papers for this study. The literature was then selected based on screening and selection criteria and categorized according to the study’s themes, which included streamflow, water yield, and runoff (see Figure 1). The primary focus of a review paper is to systematically collect, analyze, and summarize the findings and methodologies of various original research studies, providing a comprehensive overview of the state of the field. Accuracy is assessed indirectly by evaluating the quality and reliability of the studies included in the review. Systematic review and meta-analysis are structured and rigorous methods used to identify, select, and evaluate studies relevant to a specific research question. Systematic reviews aim to minimize bias and error in the review process and provide an objective summary of the available evidence. A meta-analysis extends one step further by quantitatively combining the results of multiple studies to draw more robust conclusions.

2.1. SWAT Setup, Calibration, and Validation

The SWAT model is an open-source and widely available software (SWAT 2012) that was used extensively in previous studies [20,36,37,38,39,40,41,42,43,44]. Figure 2 depicts the setup, calibration, validation, and output of the SWAT model using different input parameters. The hydrological cycle was simulated in the SWAT model using the following water-balance equation:

$${\mathrm{S}\mathrm{W}}_{\mathrm{t}}={\mathrm{S}\mathrm{W}}_{\mathrm{o}}+{\sum }_{\mathrm{i}=1}^{\mathrm{t}}\left({\mathrm{R}}_{\mathrm{d}\mathrm{a}\mathrm{y}}-{\mathrm{Q}}_{\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{f}}-{\mathrm{E}}_{\mathrm{a}}{-\mathrm{W}}_{\mathrm{s}\mathrm{e}\mathrm{e}\mathrm{p}}-{\mathrm{Q}}_{\mathrm{g}\mathrm{w}}\right)$$

(1)

where

SW_t = the soil water content after a time step t on a day i (mm);

SW_o = the initial soil water content on day i (mm);

R_day = the amount of precipitation on day i (mm);

Q_surf = the amount of surface runoff (mm);

Ea = the amount of evapotranspiration on day i (mm);

Wseep = the amount of water entering the vadose zone from the soil profile on day i (mm);

Q_gw = the amount of return flow as drainage to surface water (mm).

Figure 2. Flow of data and information through SWAT modeling.

Figure 2. Flow of data and information through SWAT modeling.

During the calibration procedure, the simulation results are compared with actual standard values from the field. During the initial calibration, these fitted parameters must be fixed. They are then eliminated in a second run to recognize the calibration results for the model’s performance after calibration and validation [45]. The model’s performance is affected by the accuracy of the the SWAT input data during the calibration and validation processes [45,46]. Amatya and Jha (2011) described the SWAT model’s performance and reported that it improves more during the validation stage than during the calibration stage [47]. However, some studies described low performance during the validation process [48,49]. Krause et al. (2005) recommended several coupled statistical methods to optimize the model’s performance [50]. Moriasi et al. (2007) proposed model evaluation guidelines for hydrological models based on the determination coefficient (R²), the Nash–Sutcliffe model efficiency (NSE), the percent bias (PBIAS), the root-mean-square error (RMSE), and the RMSE observation standard deviation ratio (RSR) [49].

In India, the R², NSE, and PBIAS were used to evaluate the performance of the SWAT model, and the evaluation was primarily based on the guidelines of Moriasi et al. (2007, 2015) [49,51]. Based on the results, the performance of the SWAT model can be categorized as follows: satisfactory, 0.50 ≤ NSE < 0.65, 0.60 < RSR < 0.70, and 0.50 ≤ R² ≤ 0.65; good, 0.65 < NSE < 0.75, 0.50 < RSR < 0.60, and 0.65 < R² < 0.75; excellent, 0.75 < NSE < 1.00, 0.00 < RSR < 0.50, and 0.75 < R² < 1.00; and unsatisfactory, NSE < 0.50, RSR > 0.70, and R² < 0.50. The statistical indexes most commonly used for model calibration included the NSE, R², PBIAS, and RSR. The details of these statistical methods are presented in Equations (2)–(5).

The NSE is used to determine the relative magnitude of the residual variance associated with the measured variance [52]. The NSE ranges from ∞ to 1; higher NSE values indicate greater model accuracy. The NSE was calculated using Equation (2).

$$NSE=1-\frac{\sum _{i=1}^n{(Q_m-Q_s)}_i^2}{\sum _{i=1}^n{(Q_{m,i}-{\overline{Q}}_m)}^2}$$

(2)

where Q_m and Q_s are the observed and simulated mean discharge data, respectively, and n is the total number of observations.

The simulated and observed streamflow of the model can be assessed using the R² value, which varies from 0 to 1; a larger R² value implies sbetter model performance. The R² value was calculated using Equation (3).

$$R^2=\frac{{\left[\sum _{i=1}^n(Q_{m,i}-{\overline{Q}}_m)(Q_{s,i}-{\overline{Q}}_s)\right]}^2}{\sum _{i=1}^n{(Q_{m,i}-{\overline{Q}}_m)}^2\sum _{i=1}^n{(Q_{s,i}-{\overline{Q}}_s)}^2}$$

(3)

The average tendency of the simulated data to be greater or smaller than the observed data is measured using the PBIAS [53]. The optimal PBIAS value is 0; however, lower values indicate better performance. The PBIAS was calculated using Equation (4).

$$PBIAS=100*\frac{\sum _{i=1}^n{(Q_m-Q_s)}_i}{\sum _{i=1}^n{Q}_{m,i}}$$

(4)

where Q represents the discharge data and m and s are the measured and simulated values, respectively.

The RSR is the ratio of the RMSE to the standard deviation of the measured data. The RSR varies from 0 to a large positive value [51]. A lower RSR value corresponds to a lower RMSE, indicating improved model simulation performance [49]. The RSR was calculated using Equation (5).

$$RSR=\frac{\sqrt{\sum _{i=1}^n{(Q_m-Q_s)}_i^2}}{\sqrt{\sum _{i=1}^n{(Q_{m,i}-{\overline{Q}}_{m,i})}^2}}$$

(5)

2.2. Climate Projections

The description of climate change scenarios is a reasonable means of predicting future climate change, representing the climatology and norms of the radiative forcing of models [54]. GCMs and RCMs are complex, three-dimensional mathematical models that depict interactions between the atmosphere, land surface, oceans, and sea ice [55]. They are valuable for assessing the essential aspects of river basins in historical and future climate scenarios, even though they involve several fundamental assumptions [56]. Suppiah et al. (2007) selected the 15 best models in Australia after using a statistical approach to test how well each model simulated observed average patterns (between 1961 and 1990) in the mean sea level pressure, temperature, and precipitation [56].

Supplementary Material S1 (S1) provides a chronological summary of the CMIP5 GCMs. Climate models need to be compatible with other modeling methods to predict climate susceptibility and climatic parameters.

The downscaling approach has been applied to various climate scenarios in GCMs [57]. GCM data are generally of low resolution and lack the spatiotemporal precision required for point-by-point provincial scrutiny. In numerous cases, they present errors that make it difficult to reproduce present-day atmospheric conditions [58]. GCMs are fragile for forecasting future climate information; however, they can provide reasonable precision with respect to the vast range of large-scale characteristics and other dissimilarities resulting from climatic factors [59]. Every climate model has several uncertainties. Therefore, Cuculeanu et al. (2002) posited that the use of multiple climate models (i.e., an ensemble approach) provides a more accurate projection for dealing with climate model challenges [60]. Thus, future studies of GCMs should focus on improving their sensitivity and assessing the inputs of variables that affect the results of climate models. The authors used various meteorological models provided by different countries in the CMIP5 project for the Indian context. Moreover, researchers used CORDEX data in the SWAT model for future scenarios in river basins.

S2 (Supplementary Material S2) presents the outputs of 17 CORDEX RCMs in the CMIP5 driving experiment with different contributing modeling centers. These were used for different river basins. Different GCMs were included: IITM-RegCM4 (six ensemble members), SMHI-RCA4 (ten ensemble members), and MPI-CSC-REMO2009 (one ensemble member). The CCCR, IITM, India, developed the driving GCMs (CanESM2, GFDL-ESM2M, CNRM-CM5, MPI-ESM-MR, IPSL-CM5A-LR, and CSIRO-Mk3.6); the Rossby Center, SMHI, Sweden, developed EC-EARTH, MIROC5, NorESM1-M, HadGEM2-ES, CanESM2, GFDL-ESM2M, CNRM-CM5, MPI-ESM-LR, IPSL-CM5A-MR, and CSIRO-Mk3.6; and CSC, Germany, developed the MPI-ESM-LR RCM.

2.3. Bias Correction of Climate Projections

GCMs and RCMs are essential tools for forecasting probable future scenarios for climatic parameters. These models identify measurement errors as less critical when the models are applied for the assessment of climate change impacts over large provinces. However, although the correction of these errors is more important for model data, it is also important when used for a hydrologic model at the basin scale [61,62]. A bias correction method is primarily used to emphasize the statistical parameters of historical data in the outputs of GCMs and RCMs, and different bias correction techniques have been developed.

Bias correction methods that have been used to correct climate model data include the delta change method [63], multiple linear regression method [64], local intensity scaling [65], monthly mean correction [66], gamma–gamma transformation [67], analog methods [68], fitted histogram equalization [69], and quantile mapping [70,71]. Some studies have used different bias correction methods in different river basins to correct precipitation and temperature data, i.e., mean bias removal (MBR), linear scaling (LS), parametric quantile mapping using scaling transformation (QMS), parametric quantile mapping using power transformation (QMP), parametric quantile mapping with linear transformation (QML), parametric quantile mapping using exponential asymptote transformation (QMEA), nonparametric empirical quantile mapping (QE), nonparametric quantile mapping by local linear least square regression (QR), and nonparametric quantile mapping via the smoothing splines method (QSPL) [62,72,73,74]. The methods used for time-averaged corrections and higher-distribution moments can also be used for other climatic variables, such as solar radiation, relative humidity, and wind [75,76,77]. Haerter et al. (2015) used two types of bias correction, namely model resolution and gauge resolution correction, for gridded and point precipitation data and further used station data to improve the correction of the gridded data [78]. Kumar Mishra and Herath (2015) applied a quantile–quantile bias correction to correct a GCM’s precipitation output for the Bagmati River Basin, Nepal, and studied the flood frequency in the basin [79]. Nyunt et al. (2016) used the three-step bias correction method to rectify bias in the GCMs of tetrad catchments in various climate regions and reported that the overall performance of the GCMs in the four regions, i.e., Kalu Ganga (Sri Lanka); Pampanga, Angat, and Kaliwa (Philippines); Yoshino (Japan); and Medjerda (Tunisia), was good [80].

In India, a few studies have focused on the bias correction of climatic data, which has been applied to hydrological research at the basin level. Bias-corrected PRECIS projections were used by Kumar et al. (2017) in the upper Kharun catchment in Chhattisgarh, India, to identify the impact of climate change; the results indicated increases in rainfall and temperature [81].The bias-corrected GFDL ESM2M was used to assess the streamflow and water balance of the Bharalu (an urban basin) and Basistha (a rural basin) river basins using the SWAT model [82]. Pandey et al. (2019) used the distribution mapping (quantile–quantile) method to eliminate the systematic bias correction of precipitation and temperature RCM data [83]. They also used these data in the SWAT model to assess the blue and green waters of the Upper Narmada River basin. The linear scaling technique was utilized to remove biases from RCMs (MPI, SMHI, and IITM) for RCP 4.5 and RCP 8.5. These models were used to calculate streamflow and sediment absorption in the Purna River basin, India [84]. Chanapathi et al. (2020) used the linear scaling method for four RCMs to assess the impact of climate change on water resources and crop yields in the Warangal district of Telangana, India [85]. Bias-corrected and downscaled scenarios have been shown to be useful inputs for various impact assessment studies. However, different bias correction and downscaling methods have their own strengths and weaknesses that need to be understood before they are applied [74].

3. SWAT-Based Climate Change Studies

Climate change significantly impacts the global water cycle, leading to alterations in precipitation patterns, temperature increases, and evapotranspiration. Variations in the amount of precipitation and its frequency and intensity directly influence the magnitude and timing of runoff and the severity of extreme events. The impact of environmental change on water availability is likely to strain water resource management due to increasing demand, leading to severe issues worldwide. Climate change poses a substantial threat to the global variability of the water supply [86]. Hanjra and Qureshi (2010) examined the global water supply and its link to food security, incorporating the effect of climate change in both components [87]. Over the past decades, several GCMs and RCMs have been developed and improved to predict climate variability and understand the elevated concentrations of greenhouse gases in the atmosphere. These GCMs and RCMs serve as excellent sources of input data for many impactful assessment studies in water resources planning and development for the future.

Researchers have used climate model scenarios for river basins to gain insight into changes in runoff under current and future scenarios in different parts of the world. These climate model scenarios include GISS and GFDL for the Great Lake Basin, North America [88]; CCCM, GFDL, and GISS + GWLF for the Tsengwen Creek watershed [89], Taiwan; DOE/NCAR PCM for the Columbia River Basin, North America [90]; HadCM3 for the Yellow River Basin, China [91]; and CGCM2, CSIRO Mk2, ECHAM4, and HadCM3 for the River Thames, UK [92]. In particular, the effects of temperature and precipitation on the runoff and streamflow in river basins were analyzed. Wurbs et al. (2005) developed a WAM system to study water availability and climate impacts in the Brazos River Basin in Texas and found that the streamflow is decreasing and water availability varies across the basin [93]. Moss et al. (2010) explained that climate change results from shifts in the Earth’s system as well as human activities, such as technological and economic advances, lifestyles, and policies [94]. White et al. (2013) used dynamical downscaling modeling and Pinto et al. (2016) used CORDEX data to predict extreme events related to the temperature and precipitation over a basin [95,96]. Different climate scenarios indicate elevated annual temperatures and decreased precipitation in the future. However, the SRES A2 climate scenario used by Akhtar et al. (2008) implies increasing temperature and precipitation toward the end of the 21st century in the Hindukush–Karakorum–Himalaya region [97]. Other studies, such as those by Diallo et al. (2012), Mote and Salathé (2010), Sharma and Babel (2013), used 21 AR4 scenarios, B1, A1B, and A2. The results showed an increase in the annual temperature and a decrease in the monsoon rainfall across river basins [98,99,100].

3.1. SWAT-Based Climate Change Studies in Different Parts of the World

Hydrological models have been used to assess the effects of climate change, land use, and meteorological variables to provide insight into the hydrologic environment of river basins [101]. Several papers on climate change impacts and SWAT model interactions are available. However, the results of these papers are highly dependent on current and future scenarios. Presently, we can only provide a brief overview of these studies. S3 (Supplementary Material S3) summarizes the SWAT literature related to hydrological models and climate change impacts studies worldwide that were conducted at the basin level between 2018 and 2019.

Zhang et al. (2012) used the SWAT model to estimate the decadal runoff in the Huifa River Basin, Northeast China, and observed a decreasing trend in the annual runoff in different years [102]. Marhaento et al. (2018) explained the impacts of future land use and climate change in the Samin catchment in Java, Indonesia, and concluded that changes in both parameters separately affected the water balance in the river basin during the 1983–2005 period and simulated similar results for 2030–2050 [103]. Emami and Koch (2019) applied the SWAT model to predict the water yield and surface runoff in the Zarrine River Basin (ZRB) in historical and future periods; they used different precipitation and temperature GCMs for RCP2.6, 4.5, and 8.5 [104]. Nazari-Sharabian et al. (2019) used downscaled daily projected climate data from BNU-ESM of the RCP4.5 and RCP8.5 scenarios to predict the runoff and water yields in the Mahabad Dam Reservoir, Iran, from 2020 to 2050 [105]. Shrestha et al. (2019) used a SWAT model of the Great Miami River Watershed, Ohio, to analyze the historical and future streamflow using ten downscaled CMIP5 climate models of RCP4.5 and 8.5 for three future periods, i.e., 2016–2043, 2044–2071, and 2072–2099 [4]. Sowjanya et al. (2020) predicted the variation in the streamflow in the Wardha watershed using RCM datasets of RCP4.5 and 8.5 and divided the analysis period into four 20-year spans, i.e., 2020–2039, 2040–2059, 2060–2079, and 2080–2099 [106]. Three RCMs were used by Nilawar and Waikar (2019) to analyze the impact of climate change on streamflow and sediment in the Purna River Basin, India [41]. They considered four future periods, i.e., P1 to P4 (2009–2099), with 30-year differences and predicted an increase in the average monthly streamflow. Yan et al. (2019) considered the streamflow response and total nitrogen loading in the Miyun Reservoir Basin using the SWAT model and projected the total nitrogen content and streamflow for the RCP4.5 and 8.5 scenarios from 2021 to 2035 and from 2051 to 2065 [10].

3.2. SWAT-Based Climate Change Studies in India

India was divided into 34 basins, 94 sub-basins, and 3448 watersheds (Figure 3), which is almost consistent with the basins delineated in the Watershed Atlas of India 1990 (AISLUS). The river basins were divided into several sub-basins, each of which was split into many watersheds (http://cgwb.gov.in/watershed/about-ws.html accessed on 23 August 2023). Information about the 34 basins and 94 sub-basins was obtained from the CGWB, Government of India, and is presented in

Table 1. Information about Indian river basins and sub-basins.

Basin	Sub-Basin
1. Barmer	The sand dunes of Barmer
2. Beas	Beas
3. Bhadar	Bhadar and other west-flowing rivers; Shetrunji and other east-flowing rivers
4. Bhatsol (Rivers from Sheravatito Tapi flowing into the Arabian Sea)	Bhatsol and others; Vasishti and others
5. Brahmani (from Mahanadi to Damodar)	Baitarni, Brahmani, and Subarnarekha
6. Brahmputra	Downstream of confluence with the Subansirito–Bangladesh Border (Lower Brahmaputra); upstream of confluence with the Subansiri (Upper Brahmaputra)
7. Cauvery	Lower Cauvery, Middle Cauvery, and Upper Cauvery
8. Chambal	Banas, Chambal upstream of Maharana Pratap Sagar (Upper Chambal), Kali Sindh and others up to confluence with Parbati, and Parbati and others (Lower Chambal)
9. Chenab	Chenab
10. Churu	Ephemeral streams of Churu
11. Ghaghar	Chautang and others; Ghagar and others
12. Godavari	Between Gaikwad and Pochampad (Middle Godavari), Indravati, Kolab and others (Lower Godavari), Manjra, Pranhita and others, upstream of Gaikwad (Upper Godavari), Wardha, and Weinganga
13. Imphal	Imphal and others, Mangpui Lui and others
14. Indus	Gilgit, Shyok, up to confluence with Shyok (Indus Lower), and upstream of Shyok confluence (Upper Indus)
15. Jhelum	Jhelum
16. Krishna	Lower Bhima, Upper Bhima, Lower Krishna, Middle Krishna, Upper Krishna, Lower Tungabhadra, and Upper Tungabhadra
17. Kutch	Drainage of Rann; Saraswati
18. Lower Ganga	Bhagirathi and others (Lower Ganga), Damodar, Gandakand others, and Sone
19. Luni	Lower Luni; Upper Luni
20. Mahanadi	Lower Mahanadi, Middle Mahanadi, and Upper Mahanadi
21. Mahi	Lower Mahi; Upper Mahi
22. Narmada	Lower Narmada, Middle Narmada, and Upper Narmada
23. Pennar (Cauvery To Krishna)	Musi and others, Palar and others, Pennar and others, and Ponnaiyar and others
24. Periyar (Rivers from Kanyakumari To Sharavati Flowing into Arabian Sea)	Netravati and others, Periyar and others, and Varrar and others
25. Qura-Qush	Shaksgam; Sulmar
26. Ravi	Ravi
27. Sabarmati	Lower Sabarmati; Upper Sabarmati
28. Surma (Drainage Flowing into Bangladesh)	Barak, Kynchiang, and other south-flowing rivers; Naoch Chara and others
29. Sutlej	Sutlej above Bhakra Dam (Upper Sutlej); Sutlej below Bhakra Dam (Lower Sutlej)
30. Tapi	Lower Tapi, Middle Tapi, and Upper Tapi
31. Upper Ganga	Above Ramganga confluence, Ghaghara, Ghaghara confluence to Gomti confluence, Gomti, Ramganga, Tons, and upstream of Gomti confluence to Muzaffarnagar
32. Vaippar (Kanyakumari to Cauvery)	Pamba and others; Vaippar and others
33. Vamsadhara (Godavari to Mahanadi)	Nagvati and others; Vamsadhara and others
34. Yamuna	Confluence with Ganga to confluence with Chambal (Lower Yamuna), confluence with Chambal to confluence with Hindon (Middle Yamuna), and upstream of confluence with Hindon (Upper Yamuna)

.

Some studies using climate scenarios have been conducted in Indian river basins. Sarthi et al. (2016) used multi-model simulations of CMIP3 and CMIP5 over the GP, India [107]. Bhatla et al. (2018) conducted CORDEX-SA experiments on the Indian subcontinent [108]. Mishra et al. (2020) used ICTP RegCM version 4.7 (RegCM4.7) in the Indian subcontinent (six Indian homogeneous rainfall zones, i.e., Northwest India (NWI), West Central India (WCI), Central Northeast India (CNI), Northeast India (NEI), South Peninsular India (SPI), hilly India (HI)) to observe reductions in precipitation over the utmost portions of India, which are attributed to convective precipitation [109]. In India, in most studies, the focus was placed on the streamflow in historical and future periods based on the use of different GCMs and RCMs. Kundu et al. (2017) used the HADCM3 and NCEP models for the Narmada River Basin, India [110]. Singh and Goyal (2017) used CMIP5 GCMs, such as ESM–2 M, CM2P1, and CM3, for the Teesta River catchment [111]. Islam et al. (2018) used the CORDEX South Asia for the Brahmaputra River Basin [112]. Chanapathi et al. (2018) used six climate models for the Krishna River Basin [39]. Pandey et al. (2019) used MIROC5, CNRM-CM5, and MPI-ESM-LR for the Upper Narmada Basin [83]. Nilawar and Waikar (2019) used the RCMs MPI-CSC-REMO2009, SMHI-RCA4, and IITM-RegCM4-1 for the Purna River Basin, India, to calculate the water yield and streamflow in the basin [84].

Table 2. Applications of SWAT model simulations (historical and future) in different Indian River Basins.

References	Applications of the SWAT Model	SWAT Output Variables	Simulation Year/Model Used (Historical)
Singh and Saravanan (2022) [127]	Wunna watershed, India	Combination of stream flow and soil moisture simulation	Historical period 1998–2016
Singh and Saravanan (2022) [128]	Three watersheds: Wunna, Bharathpuzha, and Mahanadi	Runoff and sediment	Historical period 2001–2016
Singh and Saravanan (2022) [129]	Bharathpuzha catchment, India	Streamflow	Historical period 1998–2016
Santra Mitra et al. (2021) [130]	Kangshabati River Basin of West Bengal	Runoff	Historical period 1982–2017
Swain et al. (2021) [131]	Brahmani and Baitarani River catchments	Streamflow	Historical period 1979–2018
Nune et al. (2021) [132]	Himayat Sagar (HS) catchment, India	Streamflow and groundwater levels	Historical period 1980–2007
Horan et al. (2021) [133]	Cauvery catchment	Streamflow	Historical period 1986–2003
Desai et al. (2021) [134]	Betwa River Basin	Streamflow	Historical period 2000–2011
Joseph et al. (2021) [135]	Son River	Flow regimes	Historical period June 1971 to 1975
Swain et al. (2020) [136]	Brahmani–Baitarani River Basin	Streamflow	Historical period 1990–2009
Patil and Nataraja (2020) [137]	Hiranyakeshi watershed	Runoff	Historical period 1996–2015
Shukla et al. (2020) [138]	Upper Ganga River Basin	Hydrological components	Historical period 1980 to 2012
Setti et al. (2020) [139]	Nagavalli River Basin	Streamflow	Historical period 1970–2012
Venkatesh et al. (2020) [140]	Tungabhadra River	Streamflow	Historical period 2002–2012
Padhiary et al. (2020) [141]	Baitarani River Basin	Streamflow	Historical period 2021–2095
Chauhan et al. (2020) [142]	Ghaggar River Basin	Streamflow	Historical period 1985–2015
Singh and Saravanan (2020) [143]	Ib River watershed in Mahanadi River Basin	Streamflow	Historical period 1993 to 2011
Kanishka and Eldho (2020) [144]	Godavari River Basin	Streamflow	Historical period 1995 to 2005
Merina et al. (2019) [145]	Upper Girna sub-basin, Nashik, India	Streamflow; net inflow of the Girna dam	Historical period 2000–2015
Visakh et al. (2019) [44]	Mahanadi River Basin, Brahmani–Baitarani River Basin, Hooghly River, and adjacent small river basins	Inter-comparison of water balance	Historical period 1987–2013
Singh et al. (2019) [146]	Teesta river catchment, a part of North Sikkim, Eastern Himalayas, India	Water yield and streamflow	Historical period 1981–2005
Budamala and Baburao Mahindrakar (2021) [147]	Kagna watershed of Krishna River Basin Telangana, India	Streamflow	Historical period 1987–2014
Bhattacharya et al. (2019) [148]	Beas River Basin of North Western Himalaya	Streamflow and sediment yield	Historical period 1979–2016
Paul et al. (2019) [149]	Baitarani River Basin, India	Streamflow simulation	Historical period 1977–2004
Adhikary et al. (2019) [150]	Major river basins of southern India (the Pennar Basin of India)	Streamflow	Historical period 1992–2004
Anshuman et al. (2019) [151]	Upper Godavari River Basin	Streamflow	Historical period 2000–2011
Adla et al. (2019) [152]	Punpun River Basin	Streamflow	Historical period 1979–1997
Ikhar et al. (2018) [153]	Jayakwadi reservoir stage I, Maharashtra, India	Inflows	Historical period 1981–2013
Saini et al. (2018) [154]	Kanva watershed, a rural catchment in Kaveri Basin, Southern India	Water yield, groundwater recharge, percolation, and evapotranspiration	Historical period 1992–2016
Tiwari et al. (2018) [155]	Satluj River Basin	Streamflow; runoff	Historical period 1982–2004
Sinha and Eldho (2018) [156]	Netravati river basin in the Western Ghats of India	Monthly streamflow and sediment yield	Historical period 1979–2010
Yaduvanshi et al. (2018) [157]	Subarnarekha River in India	Runoff response during extreme rain events	Historical period 1982–2011
Goswami et al. (2018) [158]	Narmada River Basin	Estimation of surface runoff	Two active phases of JJAS 2016
Anand et al. (2018) [38]	Ganga River Basin	Water balance due to change in land use	Historical period 1980–2013
Nilawar and Waikar (2018) [41]	Purna River Basin, India	Climate and land-use changes on streamflow and sediment concentration	Historical period 1980–2005
Chinnasamy et al. (2018) [40]	Ramganga Basin in India	Baseline hydrologic regime	Historical period 1999–2010
Nagraj et al. (2018) [159]	Malaprabha sub-basin, a sub-basin in the Krishna River Basin	Streamflow	Historical period 1969–2005
Shivhare et al. (2018) [160]	Part of the Ganga River Basin	Streamflow	Historical period 1996–2015
Dutta and Sen (2018) [125]	Hirakud Reservoir, Mahanadi River	Soil erosion and sediment yield	Historical period 1990–2012
Pati et al. (2018) [161]	Vansadhara River of the Mahanadi–Pennar Basin	Comparative analysis of the routing schemes	Historical period 2001–2012
Yaduvanshi et al. (2018) [162]	Ghatshila catchment; middle-lower part of Subarnarekha River	Streamflow	Historical period 1982–2005
Himanshu et al. (2018) [163]	Marol watershed, part of the Krishna River Basin	Runoff and sediment yield	Historical period 1998–2013
Setti et al. (2018) [164]	Nagavali River Basin	Streamflow	Historical period 1985–2000
Dutta et al. (2017) [124]	Tilaiya Reservoir, Jharkhand, India	Runoff and sediment yield	Historical period 1991–2000
Hasan and Pradhanang (2017) [165]	Karnali River	Streamflow	Historical period 1979–2007
Himanshu et al. (2017) [166]	Ken Basin, Central India	Runoff, sediment and water balance	Historical period 1982–2005
Makwana and Tiwari (2017) [167]	Limkheda watershed, Gujarat, India	Streamflow modeling	Historical period 2007–2012
Suryavanshi et al. (2017) [168]	Betwa River Basin, Central India	Water balance components	Historical period 1973–2001
Jothiprakash et al. (2017) [169]	Musi River, a tributary of the Krishna River	Streamflow	Historical period 2001–2012
Kumar et al. (2017) [118]	Tons River Basin	Streamflow	Historical period 1979 to 2011
Halefom et al. (2017) [170]	Indore City, Madhya Pradesh, India	Water balance in the catchment	Historical period 1979 to 2013
Kumar et al. (2017) [171]	Upper Kharun Catchment, Chhattisgarh, India	Land-use changes; water balance components	Historical period 1989–2011
Abeysingha et al. (2016) [172]	Gomti River Basin of India	Production, evapotranspiration, and irrigation requirements	Historical period 1982–2010
Alam et al. (2016) [173]	Brahmaputra River Basin	Estimation of future streamflow	Historical period 1981–2010
Patel and Nandhakumar (2016) [174]	Anjana Khadi watershed, part of the Lower Tapi Basin	Estimation of runoff potential	Historical period 2006–2007
Pandey et al. (2015) [116]	Mat River Basin	Streamflow	Historical periods 1988, 1991, and 1994
Babar and Ramesh, (2015) [175]	Nethravathi River Basin	Streamflow	Historical period 2000–2009
Abeysingha et al. (2015) [120]	Gomti River Basin in India	Water yield and evapotranspiration	Historical period 1985–2010
Singh et al. (2015) [176]	Sutlej River sub-basin (middle catchment)	Streamflow and the water balance of the sub-basin	Historical period 1970–2010
Uniyal et al. (2015) [177]	Upper Baitarani River Basin of Eastern India	Impact on water balance components	Historical period 1998–2005
Reddy and Reddy (2015) [178]	Kaddam watershed, the central part of the middle Godavari	Estimation of runoff and sediment yield	Historical period 1996–2010
Narsimlu et al. (2015) [179]	Kunwari River Basin, India	Estimation of streamflow	Historical period 1987–2005
Pervez and Henebry (2015) [117]	Brahmaputra River Basin	Assessing freshwater availability	Historical period 1988–2004
Verma and Jha (2015) [12]	Upper Baitarani River Basin, Eastern India	Streamflow and sediment yield	Historical period 1998–2005
Chandra et al. (2014) [115]	Upper Tapi Basin	Estimation of runoff and sediment yield	Historical period 1976–2005
Singh et al. (2014) [121]	Nagwa watershed in Jharkhand, India	Estimation of sediment yield	Historical period 1991–2007
Murty et al. (2014) [119]	Ken Basin, Central India	Estimation of water balance	Historical period 1985–2009
Reshmidevi and Nagesh Kumar (2014) [180]	Malaprabha River, North Karnataka, India	Streamflow	Historical period 1992–2003
Wagner et al. (2013) [181]	Mula and Mutha Rivers’ catchment upstream of Pune	Estimation of water balance	Historical period from 1989/1990 to 2009/2010
Santra and Das (2013) [123]	Watershed of the western catchment of Chilika Lake, India	Estimation of runoff	Historical period 1996–2006
Singh et al. (2013) [182]	Tungabhadra River	Estimation of streamflow	Historical period 1990–2002
Kushwaha and Jain (2013) [183]	Dabka watershed, Kumaon region Uttarakhand, India	Estimation of runoff	Historical January 2005–May 2007
Bhuvaneswari et al. (2013) [113]	Cauvery River Basin	Streamflow and rice productivity	Historical period 1970–2008
Garg et al. (2012) [184]	Upper Bhima River Basin	Agricultural water productivity	Historical period 1998–2005
Garg et al. (2012) [185]	Kothapally watershed, Southern India	Surface runoff, evapotranspiration, and agricultural water	Historical period 1978–2008
Perrin et al. (2012) [186]	Gajwel experimental watershed, India	Runoff and surface water storage; groundwater table fluctuations	Historical period 2000–2010
Singh et al. (2012) [122]	Nagwa watershed	Sediment yield	Historical period 1993–2007
References	Applications of the SWAT Model	SWAT Output Variables	Simulation Year/Model Used (Historical and Future)
Pandey et al. (2021) [187]	Upper Narmada Basin, India	Water balance	Historical period (1978 to 2005); future period (2011–2100)
Desai et al. (2021) [188]	Betwa River Basin	Water balance	Historical period (1961–1990); future periods (2010–2039, 2040–2069, and 2070–2099)
Thomas et al. (2021) [189]	Upper Narmada Basin	Streamflow	Historical period (1970–2005); future periods (2006–2040 (near-term), 2041–2070 (mid-term), and 2071–2099 (end-term))
Alam et al. (2021) [190]	Brahmaputra River Basin	Streamflow	Historical period (1981–2010); future periods (2011–2040, 2041–2070, and 2071–2100)
Das et al. (2021) [191]	Gomti River Basin	Water yield and surface runoff	Historical period (2002–2013), future period ((2017–2039), mid-century period (2040–2069), and end century (2070–2099))
Dash et al. (2021) [192]	Brahmani River Basin	Streamflow	Historical period (1970–1999); future period (2050)
Gaur et al. (2021) [193]	Subarnarekha Basin	Streamflow	Historical period (1981–2005); future period (2006–2049)
Abeysingha et al. (2020) [194]	Gomti River Basin	Flow regimes	Historical period (1982–2010); future periods (2020s, 2050s, and 2080s)
Sowjanya et al. (2020) [106]	Wardha watershed, India	Streamflow	Historical period (1975–2003); future period (2020–2099)
Sinha et al. (2023) [195]	Kadalundi River Basin, Western Ghats, India	Streamflow	Historical period (1981–2010); near (2011−2040), middle (2041−2070), and far (2071−2099) future periods
Nilawar and Waikar (2019) [84]	Purna River Basin, India	Streamflow and sediment concentration	Future periods P1 (2009–2031), P2 (2032–2053), P3 (2054–2075), and P4 (2076–2099)
Pandey and Palmate (2019) [126]	Betwa River Basin	Water yield and sediment yields	Baseline (1986–2005); future horizons (2020–2039, 2040–2059, 2060–2079, and 2080–2099)
Pandey et al. (2019) [83]	Upper Narmada Basin (UNB)	Water yield	Historical period (1970–2005); future period (2006–2100)
Chanapathi et al. (2018) [39]	Krishna River Basin	Rainfall extremes and water yield analysis	Historical period (1970–2005); future period (2006–2100)
Saharia and Sarma (2018) [82]	Bharalu (urban basin) and Basistha (rural basin) River Basins near the Brahmaputra River, India	Evaluate streamflow and water balance components variation	Historical period (1988 to 2012) and future periods (2046–2064 and 2081–2100)
Islam et al. (2018) [112]	Brahmaputra River Basin	Streamflow	Historical period (1980–2009); future periods in the 2020s (2010–2039), 2050s (2040–2069), and 2080s (2070–2099)
Saraf and Regulwar (2018) [196]	Upper Godavari River Basin, Maharashtra State, India	Runoff	Historical period (1985–2010); future periods (2011–2040 (2020s), 2041–2070 (2050s), and 2071–2099 (2080s))
Sahoo et al. (2018) [197]	Gandherswari River Basin, West Bengal, India	Streamflow	Historical period (1990–2016), future GCM (2030, 2050, and 2080) of the HadCM3 A2 and B2 scenarios
Kumar et al. (2018) [198]	Tons River Basin Madhya Pradesh, India	LULC changes on Hydrol. Process.	Historical period from 1985 to 2015 and future period from 2015 to 2035
Pandey et al. (2017) [199]	Armur watershed in Godavari River Basin, India	Estimate the water balance components	Baseline (1961–1990); future period (2071–2100); HadRM3 for the A2 and B2 scenarios
Kundu et al. (2017) [110]	Narmada River Basin, Madhya Pradesh, India	Water balance	Historical period (1961 to 2001); future periods in the 2020s (2011–2040), 2050s (2041–2070), and 2080s (2071–2099)
Singh and Goyal (2017) [111]	Teesta River catchment	Streamflow and water yield	Historical period (1980–2005); future periods (2011–2040, 2041–2070, and 2071–2100)
Mudbhatkal et al. (2017) [200]	Malaprabha River catchment and Netravathi River catchment	Streamflow	Historical period (1975–2004) and future period (2006–2070)
Singh and Goyal (2017) [201]	Teesta and Lachung Rivers	Streamflow, water depth, and precipitation	Historical period (1991–2005) and future period (2008–2100)
Mittal et al. (2016) [202]	Kangsabati River Basin	Flow regime	Historical period (1970–2008) and future period (2021–2050)
Kulkarni et al. (2014) [203]	Krishna River Basin	Surface flow, water yield, and ET and PET	Historical period (1961–1990); future periods in the 2020s (2011–2040), 2050s (2041–2070), and 2080s (2071–2098)
Mittal et al. (2014) [204]	Kangsabati River Basin	Flow regime	Historical periods (1970–1999 and 1989–2008); future period 2021–2050; and ECHAM5 and HadCM3 under the SRES A1B scenario
Narsimlu et al. (2013) [114]	Upper Sind River Basin, India	Estimate of the streamflow	Historical period (1961–1990), future period (2021–2050), and the end of the century (2071–2098)
Narula and Gosain (2013) [205]	Upper Yamuna watershed, North India	Streamflow, groundwater recharge, and nitrate load distributions in various components of the runoff	Historical period (1961–1990); future periods (2071–2100 and 2071–2098)

shows an assessment of the water yields, sediments, nutrients, climate, and land-use changes in the last decade (2012–2022). We included 110 publications to determine the correlations between the SWAT models used in different river basins, and most studies focused on estimating the streamflow. Most studies considered in this section discussed the effects of climate change on runoff, sediment, water yields, and water balance in different river basins. The SWAT has been used by various authors to estimate the basin streamflow, including Bhuvaneswari et al. (2013) for the Cauvery River Basin [113], Narsimlu et al. (2013) for the Upper Sind River Basin [114], Chandra et al. (2014) for the Upper Tapi Basin [115], Pandey et al. (2015) for the Mat River Basin [116], Pervez and Henebry (2015) for the Brahmaputra River Basin [117], and Kumar et al. (2017) for the Tons River Basin, India [118]. Murty et al. (2014) studied and predicted the water balance in the Ken Basin (India) using the SWAT model [119]. Abeysingha et al. (2015) showed the spatiotemporal correlations between evapotranspiration and the water yield in the Gomti River Basin, India [120]. Several studies have used the SWAT model to estimate the sediment yield across different regions in India, including the Nagwa watershed, Jharkhand [121,122]; the Dengei Pahad Watershed, Chilika Lake, Orissa [123]; the Upper Tapi Basin [115]; the Upper Baitarani River Basin in Eastern India [12]; the Tilaiya Reservoir, Jharkhand [124]; the Hirakud Reservoir, Mahanadi River [125]; and the Betwa River Basin [126].

3.3. The SWAT’s Performance in India

3.3.1. Sensitivity Analysis

The primary objective of a sensitivity analysis is to categorize the parameters affected during hydrological modeling by tracking and enumerating variations in the model’s inputs and outputs. Many researchers have utilized SWAT-CUP to automatically perform calculations based on input parameters to conduct sensitivity analyses [45]. S4 (Supplementary Material S4) provides the complete definitions of the calibration and validation parameters for the SWAT.

Among the 95 publications reviewed in the present study, sensitivity analyses were conducted in 87 publications (

Table 3. Calibration, validation, and statistical parameters used during the evaluation of the SWAT’s performance in different studies.

Study	SWAT-CUP Calibration and Validation Parameters	Statistical Parameters (Calibration/Validation)	SWAT Performance (Calibration/Validation)	Sensitivity Parameters (Rank)
Singh and Saravanan (2022) [127]	CH_N2, CN2, SOL_AWC, SOL_K, SOL_Z, ALPHA_BF, RCHRG_DP, REVAPMN, and SOL_BD	NSE; PBIAS	Satisfactory/very good	CN2; ALPHA_BF
Singh and Saravanan (2022) [128]	Wunna watershed, ALPHA_BF, CN2, CH_N2, SOL_BD SOL_Z, SOL_AWC, and CH_K2 and CANMX. Bharathpuzha, CN2, ALPHA_BF, SOL_BD, ESCO, REVAPMN, SOL_K, CH_N2, and CH_K2, GW_DELAY. Mahanadi, ALPHA_BF, CN2, SOL_BD, and GW_DELAY and SOL_K.	NSE; R2	Very good/Good	CN2, ALPHA_BF, and SOL_BD
Singh and Saravanan (2022) [129]	SOL_AWC, CANMX, CH_K2, RCHRG_DP, CH_N2, GW_DELAY, CN2, SURLAG, SOL_BD, REVAPMN, SOL_Z, ALPHA_BF, EPCO, ESCO, GW_REVAP, and SOL_K	R2, NSE, PBIAS, and KGE	Good	SOL_AWC
Pandey et al. (2021) [187]	CN2, ALPHA_BF, and GQ_DELAY, EPCO, ESCO	R2; NSE	Good/Very good	CN2, ALPHA_BF, and GQ_DELAY
Santra Mitra et al. (2021) [130]	SOL_AWC, CH_N2, GW_DELAY, CN2, REVAPMN, ALPHA_BF, ESCO, GW_REVAP, and GWQMN	R2; NSE	Satisfactory	CN2
Thomas et al. (2021) [189]	CN2, GWDELAY, GW_REVAP, GWQMN, SOL_AWC, and ALPHA_BF and ESCO	NSE, RMSE, RSR, and PBIAS and R2	Good/Very good	CN2
Swain et al. (2021) [131]	ALPHA_BF, ALPHA_BNK, CN2, CH_K2, CH_N2, GW_DELAY, GWQMN, GW_REVAP, REVAPMN, and SOL_AWC	NSE, PBIAS, and R2	Satisfactory	ALPHA_BF
Nune et al. (2021) [132]	SOL_AWC, SOL_K, CN, STRUCTURES_K, RESERVOIRS_K, ALPHA_BH, GW_DELAY, ESCO, and SOL_BD	NSE and R2	Good	Not Mentioned
Das et al. (2021) [191]	CN2, SOL_K, SOL_AWC, RCHRG_DP, SURLAG, REVAPMN, DEEPST, SOL_ALB, GWQMIN, GW_DELAY, ALPHA_BF, CANMX, and SOL_BD	NSE, PBIAS and R2	Good	CN2
Horan et al. (2021) [133]	SOL_BD, SOL_AWC, SOL_Z, SOL_K, CN2, GW_REVAP, REVAP_MN, GWQMN, GW_DELAY, SURLAG, ALPHA_BF, RES_K, and RES_K	NSE, PBIAS, and KGE	Good	Not Mentioned
Desai et al. (2021) [188]	CN2, SOL_AWC, GW_DELAY, REVAPMN, ESCO, GW_REVAP, GWQMN, CH_K2, ALPHA_BF, and EPCO	R2, NSE, RSR, and PBIAS	Very good/Satisfactory	CN2
Dash et al. (2021) [192]	RCHRG_DP, SOL_K, CH_N2, SOL_AWC, ALPHA_BF, SLSUBBSN, ALPHA_BNK, GW_SPYLD, GW_DELAY, and GWQMN	R2, NSE, and PBIAS	Good	RCHRG_DP
Gaur et al. (2021) [193]	CN2, SOL_K, GW_REVAP, CH_N2, GW_DELAY, ALPHA_BF, ESCO, GWQNM	R2, NSE, and PBIAS	Satisfactory	CN2
Joseph et al. (2021) [135]	CN2, AWC, Soil K, and ESCO	R2	Good	CN2
Alam et al. (2021) [190]	SOL_AWC, ALPHA_BF, GW_DELAY, GW_REVAP, CN2, SMTMP, ESCO, GWQMN, and REVAPMN	R2, NSE, P-factor, and R-factor	Very good	SOL_AWC
Swain et al. (2020) [136]	CN2, GW_DELAY, ALPHA_BF, GWQMN, CH_K2, CH_N2, ALPHA_BNK, SOL_AWC, REVAPMN, and GW_REVAP	R2, NSE, and PBIAS	Good	CN2; SOL_AWC
Patil and Nataraja (2020) [137]	CN2, SOL_AWC, ESCO, GW_DELAY, ALPHA_BF, GW_REVAP, and RECHRG_DP	R2; NSE	Satisfactory	CN2
Shukla et al. (2020) [138]	CN_2, TLAPS, SOL_AWC, SOL_K, ALPHA_BF, GW_DELAY, ESCO, SMTMP, SMFMN, SMFMX, SNO50COV, SFTMP, TIMP, and SNOCOVMX	R2, NSE, and PBIAS	Very good	CN2
Setti et al. (2020) [139]	HRU_SLP, LAT_TTIME, ALPHA_BNK, CH_K2, SLSUBBSN, ESCO, CN2, SOL_AWC, EPCO, OV_N, RCHRG_DP, SOL_K, GW_REVAP, GW_DELAY, ALPHA_BF, REVAPMN, GWQMN, SURLAG	NSE; PBIAS	Good	CN2
Venkatesh et al. (2020) [140]	CN2, ALPHA_BF, GW_DELAY, GWQMN, CH_N2, CH_K2, SOL_AWC, SOL_K, ESCO, GW_REVAP, REVAPMN, SLSUBBSN, SLSOIL, and ALPHA_BNK	R2, NSE, and PBIAS	Good	CN2
Padhiary et al. (2020) [141]	CN2, ALPHA_BF, GW_DELAY, GWQMN, CH_N2, CH_K2, SOL_AWC, SOL_K, ESCO, and SURLAG	R2, NSE, and PBIAS	Good	ALPHA_BF
Chauhan et al. (2020) [142]	CN2, ESCO, GWQMN, ALPHA_BF, GW_REVAP, GW_DELAY, SOL_K, and OV_N	R2, NSE, and PBIAS	Satisfactory	CN2
Singh and Saravanan (2020) [143]	ALPHA_BF, CN2, CH_N2, CH_K2, RCHRG_DP, SOL_AWC, SOL_K, SOL_Z, SOL_BD, GW_DELAY, REVAPMN, ESCO, GWQMN, GW_REVAP, CANMX, EPCO, and SURLAG	R2, NSE, and PBIAS	Good	ALPHA_BF
Abeysingha et al. (2020) [194]	EPCO, ALPHA_BF, SURLAG, CH_N2, SOL_AWC, GWQMN, SOL_K, ESCO, CN2, CH_K2, RCHRG_DP, CANMX, POT_VOLX, OV_N, SOL_BD, POT_FR, and GW_DELAY	NSE, R2, PBIAS, and RSR	Good	CN2
Sinha et al. (2023) [195]	SOL_AWC, SURLAG, CN2, ESCO, EPCO, ALPHA_BF, GW_DELAY, GW_REVAP, GWQMN, and RCHRG_DP	R2, NSE, and PBIAS	Good	CN2
Kanishka and Eldho (2020) [144]	CN2, SOL_AWC, ESCO, SLSUBBSN, OV_N, HRU_SLP, GW_REVAP, GWQMN, and REVAPMN	R2, NSE, PBIAS	Good	CN2
Sowjanya et al. (2020) [106]	CN2, ALPHA_BF, GW_DELAY, GWQMN, GW_REVAP, ESCO, SOL_K, ALPHA_BNK, SOL_AWC, REVAPMN, SOL_BD, OV_N, CH_K2, EPCO, HRU_SLP, CH_N2, and SLSUBBSN	R2; NSE	Very Good/Good	CN2
Merina et al. (2019) [145]	CN, ALPHA_BF, GW_DELAY, GWQMN, GW_REVAP, REVAPMN, ESCO, CH_K2, SOL_AWC, SOL_K, SLSUBBSN, and CH_N2	NS; R2	Good/Good	ALPHA_BF
Nilawar and Waikar (2019) [84]	CN2, ALPHA_BF, GW_DELAY, GWQMN, GW_REVAP, ESCO, CH_N2, CH_K2, ALPHA_BNK, SOL_AWC, SOL_K, SOL_BD, USLE_C.plant, CH_COV2, USLE_K, LAT_SED, CH_COV1, SPCON, and SPEXP	R2; NSE	Very Good/Very Good	ALPHA_BNK
Visakh et al. (2019) [44]	ALPHA_BF, CH_K2, CN2, SURLAG, ESCO, GW_DELAY, SOL_AWC, GW_REVAP, REVAPMN, GWQMN, and SLSUBBSN	NSE	Good/Good	CN2
Singh et al. (2019) [146]	CN2, ALPHA_BF, GW_DELAY, GWQMN, ESCO, SFTMP, SOL_AWC, SOL_K, CH_N2, CH_K2, ALPHA_BNK, SNOCOVMX, SNOCOVMN, SNO50COV, and PLR	R, R2, and RMSE	Good/Good	CN2
Budamala and Baburao Mahindrakar (2021) [147]	CN2, EPCO, SOL_AWC, GW_REVAP, REVAPMN, GWQMN, GW_DELAY, ALPHA_BF, RCHRG_DP, and CH_K2	NSE, PBIAS, R2	Very Good/Very Good	CN2
Bhattacharya et al. (2019) [148]	CN2, ESCO, SOL_AWC, ALPHA_BF, SURLAG, SOL_BD, ESCO, EPCO, TIMP, SFTMP, USLE_K, USLE_P, SPCON, SPEXP, ADJ_PKR	NSE; R2	Satisfactory/Satisfactory	CN2
Paul et al. (2019) [149]	CN2, GWQMN, GW_DELAY, ALPHA_BF, CH_N2, CH_K2, and ESCO	NSE, PBIAS, and R2	Satisfactory/Satisfactory	CN2
Adhikary et al. (2019) [150]	CN2, ESCO, GW_DELAY, SOL_AWC, RCHRG_DP, GW_REVAP, SOL_K, GWQMN, SOL_BD, ALPHA_BNK, ALPHA_BF, SLSUBBSN, EPCO, REVAPMN, and OV_N	R2, NSE, RSR, and PBIAS	Good/Very Good	CN2
Anshuman et al. (2019) [151]	CN2, SOL_AWC, SOL_K, SOL_CLAY, ESCO, SLSUBBSN, HRU_SLP, GWQMN, REVAPMN, GW_REVAP, ALPHA_BF, RCHRG_DP, ALPHA_BNK, CH_K2, and CH_N2	R2, NSE, and PBIAS	Very Good/Very Good	CN2
Pandey and Palmate (2019) [126]	CN2, SURLAG, ALPHA_BF, GDRAIN, RCHRG_DP, ESCO, GWQMN, GW_DELAY, SOL_AWC, USLE_K, CH_ERODMO, RES_STLR_CO, PRF, CH_COV1, ADJ_PKR, RES_SED, CH_COV2, USLE_P, LAT_SED, USLE_C, AGRL, RES_NSED, SPEXP, and USLE_C	R2, NSE, and PBIAS	Good/good	CN2
Adla et al. (2019) [152]	CH_K2, GW_DELAY, SOL_AWC, GW_REVAP, ESCO, CN2, ALPHA_BF, and SURLAG	P-factor, R-factor, R2, and NSE	Good/good	Not mentioned
Pandey et al. (2019) [83]	CN2, ALPHA_BF, GW_DELAY, GWQMN, ESCO, EPCO, GW_REVAP, REVAPMN, SOL_K, OV_N, SOL_AWC, CANMX, CH_N2, and HRU_SLP	R2; NSE	Good/Satisfactory	CN2
Ikhar et al. (2018) [153]	CN2, ESCO, ALPHA_BF, GW_DELAY, SOIL_AWC, GWQMN, and GW_REVAP	NS, R2, and RSR	Very Good/Very Good	CN2
Saharia and Sarma (2018) [82]	CN2, ALPHA_BF, ESCO, BLA, SOL_AWC, CANMX, SOL_Z, GW_DELAY, GW_REVAP, SLOPE, NPERCO, BIOMIX, SHALLST_N, SOL_NO3, and SOL_ORGN	RMSE, PBIAS, NSE, R2	Good/Good	CN2
Anand et al. (2018) [38]	CN2, SURLAG, GWQMN, GW_DELAY, ALPHA_BF, GW_REVAP, ESCO, OV_N, SFTMP, SMTMP, PLAPS, TLAPS, SOL_AWC, SMFMN, and SMFMX	NSE, R2, PBIAS, and RSR	Very Good/Very Good	Not Mentioned
Chanapathi et al. (2018) [39]	CN2, RCHRG_DP, GWREVAP, SOL_K, GWDELAY, CH_K2, CH_N2, ALPHA_BNK, REVAPMN, ALPHA_BF, GWQMN, and ESCO	R2; NSE	Good/Good	CN2
Nilawar and Waikar (2018) [41]	GWQMN, GW_REVAP, CN2, GW_DELAY, SOL_K, REVAPMN, ALPHA_BF, ESCO	R2; NSE	Good/Good	Not Mentioned
Sinha and Eldho (2018) [156]	CN2, USLE_P, SOL_AWC, USLE_K, ESCO, LAT_SED, EPCO, HRU_SLP, SUR_LAG, SLSUBBSN, ALPHA_BF, GW_DELAY, GW_REVAP, CH_N2, RCHRG_DP, CH_K2, GWQMN, CH_COV1, and CH_ERODMO	NS, R2, and PBIAS	Good/Good	CN2
Saini et al. (2018) [154]	SOL_AWC, CN2, SOL_K, ALPHA_BF, GWQMN, GW_REVAP, REVAPMN, RCHARG_DP, ESCO, SOL_AWC, SURLAG, and OV_N	R2, NSE, RSR, and PBIAS	Very Good/Very Good	Not Mentioned
Tiwari et al. (2018) [155]	CN2, ALPHA_BF, GW_DELAY, GWQMN, SFTMP, SMTMP, SMFMX, SMFMN, TIMP, SNOCOVMX, MSK_CO1, MSK_CO2	NS; R2	Good/Satisfactory	CN2
Islam et al. (2018) [112]	ALPHA_BF, ALPHA_BNL, CH_N2, CN2, EPCO, ESCO, GWQMN, GW_DELAY, GW_REVAP, HRU_SLP, OV_N, REVAPMN, SFTMP, SLUSUBBSN, SOL_AWC, SOL_BD, and SOL_K	R2, NSE, PBIAS, RMSE, and RSR	Good/Satisfactory	CN2
Nagraj et al. (2018) [159]	CN2, ESCO, SOL_AWC, GW_DELAY, and ALFA_BF	R2; NSE	Good/Good	CN2
Shivhare et al. (2018) [160]	GW_DELAY, SOL_K, ALPHA_BF, CH_N2, CH_K2, CN2, USLE_P, SOL_AWC, SURLAG, USLE_K	R2; NSE	Very Good/Very Good	CH_K2
Saraf and Regulwar (2018) [196]	CN2, ALPHA_BF, GW_DELAY, GWQMN, GW_REVAP, ESCO, CH_N2, CH_K2, SOL_AWC, SOL_K, SURLAG, CANMX	R2; NSE	Good/Good	CH-K2
Sahoo et al. (2018) [197]	CN2, GW_DELAY, ALPHA_BF, GWQMN, GW_REVAP, REVAPMN, SOL_AWC, ESCO, SOL_K, ALPHA_BNK, CH_K2, EPCO, HRU_SLP, CH_N2, OV_N, SLSUBBSN, SOL_BD, and SURLAG	R2, NSE, and PBIAS	Satisfactory/Satisfactory	RCHRG_DP, CN2
Yaduvanshi et al. (2018) [162]	CN2, ALPHA_BF, GW_DELAY, GWQMN, GW_REVAP, ESCO, CH_N2, CH_K2, and SOL_K	R2; NSE	Good/Good	CN2
Kumar et al. (2018) [198]	ALPHA_BF, SOL_K, ESCO, GW_DELAY, CH_K2, and SLSUBBSN	P and R factor, R2 NSE, RSR, and PBIAS	Good/Good	ALPHA_BF
Pati et al. (2018) [161]	SOL_BD, CN2, ALPHA_BF, GW_DELAY, GWQMN, SOL_K, SOL_AWC, LAT_TTIME, REVAPMN, SURLAG, CH_N2, ESCO, EPCO, CH_K2, GW_REVAP, and CANMX	R2, NSE, and MAPE	Very Good/Good	CH_N2
Yaduvanshi et al. (2018) [157]	CN2, ALPHA_BF, GW_DELAY, SOL_AWC, and ESCO	NSE;, P and R factors	Good/Good	CN2
Himanshu et al. (2018) [163]	ALPHA_BF, BIOMIX, CH_COV2, CH_K1, CH_K2, CH_N2, CH_S1, CN2, EPCO, ESCO, GW_DELAY, GWQMN, GW_REVAP, OV_N, RCHRG_DP, REVAPMN, SLSUBBSN, SOL_AWC, SURLAG, SOL_K, SURLAG, USLE_K, and USLE_P	NSE, PBIAS, and RSR	Very Good/Very Good	CH_N2
Hasan and Pradhanang (2017) [165]	CN2, OV_N, ALPHA_BF, GW_DELAY, GWQMN, GW_REVAP, RG_DP, REVAPMN, CH_N2, ESCO, EPCO, TLAPS, SOL_AWC, SMTMP, SMFMX, SMFMN, SFTMP, TIMP, and SURLAG	R2, NSE, and PBIAS	Very Good/Very Good	GWQMN,
Setti et al. (2018) [164]	LAT_TTIME, HRU_SLP, CN2, SOL_AWC, ESCO, SLSUBBSN, RCHRG_DP, EPCO, GWQMN, REVAPMN, OV_N, GW_DELAY, ALPHA_BF, and GW_REVAP	R2, NSC, PBIAS, and RSR	Very Good/Very Good	LAT_TTIME
Kundu et al. (2017) [110]	CN2, ESCO, SOL_ Z, SOL_AWC, CANMX, GWQ_MN, REVAPMN, BLAI, EPCO, GW_REVAP, SURLAG, ALPHA_BF, SOL_K, SLOPE, GW_DELAY, BIOMIX, SOL_ALB, SLSUBBSN, CH_N2, CH_K2, TLAPS, TIMP, SMTMP, SMFMX, SMFMN, and SFTMP	R2, NSE, PBIAS, and RSR	Very Good/Very Good	CN2
Singh and Goyal (2017) [111]	CN2, ALPHA_BF, GW_DELAY, GWQMN, ESCO, SFTMP, SOL_AWC, SOL_K, CH_N2, CH_K2, and ALPHA_BNK	R2; NSE	Good/Good	CN2
Suryavanshi et al. (2017) [168]	CN2, ESCO, SOL_AWC, SOL_Z, REVAPMN, GWQMN, CH_K2, ALPHA_BF, EPCO, CH_N2, GW_DELAY, SURLAG, SOL_K, and GW_REVAP	NSE, R2, PBIAS, and RMSE	Very Good/Very Good	CN2
Makwana and Tiwari (2017) [167]	Input parameters by (Arnold et al. 2012). ALPHA_BF, CH_K2, SURLAG, and GWQMN	R2, NSE, RMSE, and mean absolute error (MAE)	Satisfactory/Satisfactory	ALPHA_BF
Himanshu et al. (2017) [166]	ALPHA_BF, CH_N2, CH_K2, CN2, SURLAG, ESCO, SOL_Z, GW_DELAY, CANMX, SOL_AWC, GW_REVAP, SOL_K, REVAPMN, BLAI, GWQMN, SLSUBBSN, EPCO, SLOPE, SOL_ALB, BIOMIX, CH_EROD, SPEXP, RCHRG_DP, SPCON, CH_COV, USLE_C, USLE, and USLE_P	NSE, R2, PBIAS, and RMSE	Very Good/Good	SPCON
Kumar et al. (2017) [118]	CN2, ALPHA_BF, GWQMN, ESCO, CH_K2, CH_N2, REVAPMN, SOL_AWC, HRU_SLP, SOL_K, SOL_BD, SLSUBBSN, GW_REVAP, EPCO, GW_DELAY, SFTMP, ALPHA_BNK, SURLAG, and OV_N	R2, NSE, PBIAS, and RSR	Good/Good	ALPHA_BF
Pandey et al. (2017) [199]	CN2, ESCO, GWQMN, SOIL_AWC, SOIL_Z, ALPL_BF, GW_REVAP, BLAI, EPCO, CH_K2	R2, ENS, and PE	Good/Good	CN2
Mudbhatkal et al. (2017) [200]	CN2, SOL_K, SLOPE, SOL_AWC, ESCO, SOL_Z, CANMX, CH_K2, SURLAG, ALPHA_BF, EPCO, CH_N, GW_DELAY, GW_REVAP	R2; NSE	Good/Good	CN2
Jothiprakash et al. (2017) [169]	CH_K2, CN2, OV_N, SURLAG, ESCO, RCHRG_DP, SOL_K, GW_DELAY, GWQMN, GW_REVAP, EPCO, ALPHA_BF, SOL_AWC, and SOL_BD	R2, NSE, MAE, and PBIAS	Satisfactory/Satisfactory	CH_K2
Singh and Goyal (2017) [201]	CN2, GW_DELAY, SOL_K, HRU_SLP, and ALPHA_BF	R2; NSE	Very Good/Very Good	CN2
Kumar et al. (2017) [171]	CN2, ALPHA_BF, GWQMN, ESCO, CH_K2, CH_N2, REVAPMN, SOL_AWC, HRU_SLP, SOL_K, SOL_BD, SLSUBBSN, GW_REVAP, EPCO, GW_DELAY, SFTMP, ALPHA, _BNK, and SURLAG, OV_N	P and R factor, R2, NSE, PBIAS, and RSR	Good/Good	ALPHA_BF
Dutta and Sen (2018) [125]	CN2, SURLAG, CH_N, ESCO, SLSUBBSN, GWQMN, EPCO, GW_REVAP, SOL_AWC, SOL_ALB, GW_DELAY, ALPHA_BF, REVAPMN, SOL_K, and TLAPS	R2, NSE, and RSR	Good/Good	CN2
Kumar et al. (2017) [81]	SOL_Z, EPCO, GW_REVAP, SOL_BD, GWQMN, ALPHA_BF, RCHRG_DP, SURLAG, SOL_K, SOL_AWC, CH_K2, ESCO, GW_DELAY, and CN2	R2; NSE	Very Good/Very Good	CN2
Abeysingha et al. (2016) [172]	ALPHA_BF, SOL_AWC, EPCO, GW_DELAY, RCHRG_DP, GW_REVAP, ESCO, CN2, GWQMIN, REVAPMIN, SOL_K, OV_N, CANMAX, CH_K2, and SOL_BD	NSE, R2, RMSE (RSR), and PBIAS	Good/Good and Satisfactory/Satisfactory	ALPHA_BF
Mittal et al. (2016) [202]	ALPHA_BF, GW_DELAY, GWQMN, GW_REVAP, ESCO, CH_N2, CH_K2, CN2, SOL_AWC, SOL_K, and SOL_BD	PBIAS, NSE, and R2	Good/Good	CN2
Pandey et al. (2015) [116]	ALPHA_BF, GWQMN, ESCO, CH_N2, REVAPMN, and SOL_AWC	R2; NSE	Very Good/Very Good	ALPHA_BF
Narsimlu et al. (2015) [179]	CN2, ALPHA_BF, GWQMN, ESCO, CH_K2, CH_N2, REVAPMN, SOL_AWC, HRU_SLP, SOL_K, SOL_BD, SLSUBBSN, GW_REVAP, EPCO, GW_DELAY, SFTMP, ALPHA_BNK, and OV_N	NSE, R2, PBIAS, and RSR	Very Good/Good	ALPHA_BNK, ESCO
Uniyal et al. (2015) [177]	ALPHA_BF, CH_K2, ESCO, SOL_K, SOL_AWC, CN2, GW_DELAY, SURLAG, CH_N2, and GWQMN	R2, MAE, and NSE	Very Good/Very Good	SOL_AWC
Abeysingha et al. (2015) [120]	RCHRG_DP, SOL_AWC, SOL_K, GW_DELAY, EPCO, OV_N, ESCO, CN2, GW_REVAP, ALPHA_BF, GWQMIN, REVAPMIN, CANMAX, and CH_K2	NSE, RSR, and PBIAS	Satisfactory/Satisfactory	Not Mentioned
Pervez and Henebry (2015) [117]	CN2, ESCO, ALPHA_BF, PLAPS, TLAPS, SLSUBBSN, GWQMN, REVAPMN, GW_REVAP, and EPCO	PBIAS, NSE, R2, and RMSE	Very Good/Very Good	CN2
Babar and Ramesh (2015) [175]	ALPHA_BF, CANMX, CH_K2, CN_N2, CN2, ESCO, GWQMN, REVAPMN, GW_DELAY, SOL_K, and SURLAG	NSE, R2, and RMSE	Very Good/Very Good	ALPHA_BF
Verma and Jha (2015) [12]	ALPHA_BF, CH_N2, SURLAG, CN2, CH_K2, GWQMN, ESCO, GW_DELAY, SOL_K, SOL_AWC, SOL_Z, GW_REVAP, EPCO, REVAPMIN, CANMX, SOL_ALB, BLAI, and BIOMIX	NSE, RSR, PBIAS, and R2	Good/Good	ALPHA_BF
Murty et al. (2014) [119]	CN, SOL_AWC, SOL_Z, ESCO, GWQMN, REVAPMN, BLAI, CANMX, CH_K2, EPCO, GW_REVAP, and SOL_K	R2, PBIAS, NSE, RSR, and d index	Very Good/Very Good	GWQMN
Chandra et al. (2014) [115]	ALPHA_BF, CANMAX, CH_K2, CH_N2, CN2, EPCO, ESCO, GW_DELAY, GW_REVAP, GWQMN, REVAPMN, SOL_ALB, SOL_AWC, SOL_K, SOL_Z, SURLAG, RCHRG_DP, LAT_TIME, MANNING’S, “N” VALUE FOR OVERLAND FLOW, SOL_BD, SPCON, CH_EROD, USLE_P, and USLE_C	NSE; RSR	Satisfactory/Satisfactory	Not Mentioned
Singh et al. (2014) [121]	USLE_P, CH_EROD, SPCON, CH_COV, SPEXP, USLE_C	R2, NSE, and PBIAS	Good/Good	USLE_P
Reshmidevi and Nagesh Kumar (2014) [180]	ALPHA_BF, CH_K2, CH_N2, CN2, EPCO, ESCO, GW_DELAY, GW_REVAP, GWQMN, RCHRG_DP, REVAPMN, SOL_AWC, SOL_K, and SURLAG	R2, NSE, and RMSE	Very Good/Very Good	RCHRG_DP
Kushwaha and Jain (2013) [183]	CN2, ESCO, SOL_Z, SOL_K, SOL_AWC, CH_N, GWDELAY, GW_REVAP, REVAPMN, GWQMN, ALPHA_BF, RCHRG_DP, and SURLAG	R2, NSE, and RMSE	Good/Good	CN2
Singh et al. (2013) [182]	CN2, ALPHA_BF, GW_DELAY, GWQMN, GW_REVAP, ESCO, CH_N2, CH_K2, ALPHA_BNK, and SOL_K	R2; NSE	Good/Good	CH_K2
Narsimlu et al. (2013) [114]	CN2, ALPHA_BF, GW_DELAY, GWQMN, GW_REVAP, ESCO, EPCO, CH_N2, CH_K2, ALPHA BNK, SOL_AWC, SOL_K, SOL_BD, and SFTMP	R2; NSE	Very Good/Very Good	SOL_AWC
Santra and Das (2013) [123]	ALPHA_BF, CH_K2, CH_N, CN2, ESCO, GW_DELAY, GWQMN, and SURLAG	NSE; RMSE	Good/Good	CH_K2
Narula and Gosain (2013) [205]	SMTMP, SMFMX, SMFMN, TIMP, SFTMP, CN2, ESCO, AWC, SOL_K, CH_K2, GW_DELAY, and SURLAG	R2; NSE	Very Good/Very Good	TIMP
Garg et al. (2012) [184]	RCHRG_DP, CN2, ALPHA_BF, ESCO, SOL_AWC, SOL_Z, GW_DELAY, GW_REVAP, SURLAG, and REVAP_MN	R; NSE	Satisfactory/Satisfactory	RCHRG_DP
Singh et al. (2012) [122]	USLE_P, CH_EROD, SPCON, CH_COV, SPEXP, and USLE_C	R2; NSE	Good/Good	USLE_P

). The parameters of the sensitivity analyses are presented in Table 3. Different sensitivity parameters were used in various studies. We found that the streamflow calculation of the SWAT model was the most sensitive to changes in the initial SCS curve number (CN). This parameter is linked to the runoff from a precipitation event in a specific area. Overall, the most significant parameters in the modeling processes were CN2, ALPHA_BNK, SPCON, ALPHA_BF, SOL_AWC, GWQMN, SOL_AWC, RCHRG_DP, CH-K2, CH-N2, LAT_TTIME, USLE_P, and TIMP (Table 3). In the 95 review papers, many sensitivity parameters were identified for different river basins; CN2 appeared 49 times as a more sensitive parameter in Indian river basins, followed by ALPHA_BF in 12 articles and CH_K2 in five articles; 10 articles did not mention any sensitive parameters (Figure 4). The CN2 covered 50% of the sensitivity analysis and showed greater sensitivity for streamflow analyses in different climate scenarios.

3.3.2. Streamflow Performance

The streamflow performance depends on the discharge of the stream and varies with time and space. Excessive streamflow can lead to hazards such as flooding. Narsimlu et al. (2013) calibrated a SWAT model in the Upper Sind River Basin of India using monthly streamflow data and predicted drastic changes in future streamflow during the monsoon season and decreases during non-monsoon seasons for selected future periods [114]. Narsimlu et al. (2015) also used the SWAT and SWAT-CUP for calibration, sensitivity, and uncertainty analyses in the Kunwari River Basin (KRB) [179]. The authors used monthly streamflow data and the sequential uncertainty domain parameter-fitting algorithm (SUFI-2) technique, reporting that the model effectively simulated the observed streamflow in the KRB. Another study estimated the effect of climate change on the water balance component of the Upper Baitarani River Basin, using daily streamflow data for model calibration and validation, and they reported that their results were comparable with those of the NSE and mean absolute error (MAE) [177]. Verma and Jha (2015) used the SWAT model for the Upper Baitarani River to evaluate the streamflow and sediment yield at monthly (NSE = 0.91 and 0.90) and daily (NSE = 0.62 and 0.59) time steps [12]. The performances of the calibrated and validated models in the monthly simulation were good. The most sensitive parameter of the SWAT model has been identified as ALPHA_BF, followed by SOL_K, SFTMP, SLSUBBSN, and SOL_AWC, and these parameters considerably affect the streamflow in the Tons River Basin, India [118]. Singh et al. (2013) estimated the streamflow using SUFI-2 and GLUE in SWAT-CUP [182]. Their results showed excellent agreement throughout the monthly calibration; however, in the case of the Tungabhadra River, it was moderately good during the daily calibration. Himanshu et al. (2017) evaluated the sediment yield and water balance of the Ken Basin in Central India [166]. The daily streamflow simulation was good, the monthly simulation was very good, and the sediment concentration in the study area was accurately simulated. Makwana and Tiwari (2017) estimated the streamflow by using a combination of the SWAT model and new approaches, such as neural networks (NNs), for the Limkheda watershed, Gujarat, India [167]. Based on the calibration and validation results, the simulation using NNs was better than the SWAT’s simulations of surface runoff. Nilawar and Waikar (2019) quantified the effects of climate change on streamflow and sediment yield in the Purna River Basin, India [84]. The authors used RCMs with RCP4.5, RCP8.5, and P1 to P4 (2009–2099) and projected that the average monthly streamflow would increase in both scenarios.

4. Challenges and Future Directions

The SWAT model has been used in a limited number of studies to determine the effects of climate and land-use changes on the streamflow and sediment yield in Indian river basins. The model has not been implemented in many regions of India, and its application is limited to a few basins. In the future, studies should focus on other aspects, including uncertainty analyses, crop assessments, ecosystem services, and extreme events. In India, SWAT workshops and training should be promoted to understand SWAT applications in different fields to meet user interests. The results of several studies showed that climate change negatively affects the streamflow of basins in climatic scenarios, and several studies have emphasized the importance of land-use modifications in simulations [206,207].

In India, various government agencies provide weather, river discharge, and other relevant data that must be processed to ensure the reliability of the data before they are used as inputs in SWAT models. Climate data are freely available from the Indian Meteorological Department, while the availability of discharge data is limited to a few agencies. The SWAT website (https://swat.tamu.edu/data/india-dataset/ accessed on 23 August 2023) provides free input data for SWAT applications in India. Therefore, future studies should be carried out using different emission scenarios to predict the streamflow of river basins using the SWAT model. These scenarios have been developed by considering the effects of climate or socioeconomic changes, deforestation, urban development, and reductions in agricultural land. Consequently, the susceptibility of water resources in various future scenarios can be highlighted by grouping climate and land use/land cover (LULC) change scenarios. Previous studies did not describe crop factors, and it remains unclear how the SWAT model represents the land cover in the basin. Additionally, none of the publications discussed agriculture-related outputs such as biomass, crop yield, or leaf area index. Calibration and validation processes indicate the good and very good performance of the SWAT in Indian river basins. Streamflow, dam information, sediment, and water quality data at the sub-basin or basin level have been used in studies to optimize the model’s efficiency in different geological areas. In some studies, validation and calibration details remain unclear, while in others, excellent SWAT model performance has been reported [84,168,177]. Gassman et al. (2007) provided a comprehensive description of the strengths, weaknesses, and exploration requirements of the SWAT model [21]. In many SWAT studies in India, the model has been considered sufficiently reliable for use with statistical indexes such as the NSE, R², PBIAS, and RSR for model calibration and validation at the basin level.

The SWAT model has limitations in estimating streamflow, particularly in cases of extreme precipitation [23,208,209]. These drawbacks affect the performance of SWAT applications during various extreme events that result in significant environmental damage and influence flow simulations. Therefore, improvements to the model are needed to better represent runoff responses to extreme precipitation. In addition, several other models, including SWIM, SWAT-G, SWAT-MODFLOW, ESWAT, and SWATgrid, have been utilized for various applications such as climate and land-use impact assessments, groundwater modeling, and grid-based analyses [210,211,212,213]. Bieger et al. (2017) described the SWAT+ version, which was completely restructured from the SWAT model; they improved the algorithm and modular design [214]. SWAT+ provides a more flexible spatial representation of interactions and processes within a watershed. The addition of landscape elements, flow, and pollutant routing throughout the terrain is the most significant modification in the SWAT+. It is also more flexible than the SWAT model, allowing for the definition of management schedules, routing components, and connections between achieved movement systems and the natural stream network. These changes in the SWAT application may be helpful in the future and may improve the accuracy of streamflow simulations in the Indian context.

5. Conclusions

The SWAT model is a highly regarded, integrated tool utilized for conducting multidisciplinary studies at the regional scale across various physiographic and meteorological conditions. Despite being continuously developed, it stands as the most widely employed watershed-scale model globally. Notably, it has found applications in simulating hydrologic regimes within the context of India. The outcomes of these applications have primarily revolved around the estimation of streamflow, water balance, sediment yield, and runoff within diverse river basins. This comprehensive review seeks to provide in-depth insights into the hydrological and climatic systems of Indian rivers, thereby contributing to addressing both local and global concerns related to the adverse impacts of a changing climate on hydrology, agriculture, and the environment.

Several studies have delved into historical changes in hydrological aspects and the prediction of various climatic scenarios, utilizing data derived from river basins, meticulously simulated with the SWAT model. The model has exhibited commendable performance, guided by effective calibration and validation procedures. Consequently, its adoption has witnessed significant growth in India, as evidenced by the increasing number of research articles being published. These studies have primarily emphasized climate change assessments and the analysis of the model’s capabilities. Notably, the results from various climatic models have been harnessed to predict streamflow in river basins, albeit with variations depending on the specific basin and climatic model employed. It is worth noting that no single climatic model has comprehensively covered an entire river basin, leading to variations in the predicted water balance for each basin.

Some of the main concerns of SWAT users in India are the accessibility and reliability of data. Recently, however, the model has gained credibility in Indian river basins due to its reliance on statistical indicators for performance evaluations, including the NSE, R², PBIAS, and RSR. Descriptions of parameters and the calibration and validation methods presented in the studies are vital for a comprehensive understanding of the SWAT model’s performance analysis. Nevertheless, it is important to note that while several studies have been conducted, only a few have incorporated internal calibration points or other distributed data sources, such as remote sensing, tracers, and groundwater data, to validate distributed predictions. This review represents a multidisciplinary effort aimed at improving our understanding of hydrological processes and contributing to the development of more effective water management strategies, particularly in light of evolving climate conditions. In summary, the SWAT model stands as a valuable tool for examining river basins and devising sustainable strategies to mitigate the impacts of climate change and shifts in land use.