Geomorphometric based catchment runoff characterization and assessment of temporal landuse change impact on runoff inflow to an ungauged reservoir under different climatic scenarios using geospatial techniques and SWAT model

Landuse and landcover (LULC) has altered many ecosystems rapidly and extensively over the last five decades because of the growing demand for the ever-increasing population for land and other natural resources (Sushanth & Bhardwaj 2019b). Understanding the interconnections between human activities and the environment requires knowledge about the LULC change of a catchment. The change in LULC influences many hydrologic processes and it modifies the rate and the volume of runoff, sediments and their transport towards the catchment outlet, thereby causing siltation of reservoirs and water conveyance systems in the downstream areas (Walling 2009). It ultimately affects the quantity and quality of water supply to the users. The knowledge of LULC is significant for planning conservation interventions in a catchment. Moreover, it is a predominant component for modelling catchment hydrological processes such as runoff. Frequent assessment of LULC changes in catchments is pivotal for better understanding the present-day situations of many reservoirs worldwide (Verburg et al. 2004).

In north-western India, an area around 3.33 Mha in the Shivalik foot-hills that comprises the states of Himachal Pradesh, Uttarakhand, Jammu and Kashmir, Punjab and Haryana, locally known as the Kandi region, is considered to be one of the most degraded and fragile agro-ecosystem in the country (Bhardwaj & Kaushal 2009; Yadav et al. 2015). This area's average annual erosion rate is around 80 Mg/ha/year; in some of the watersheds, it is as high as 244 Mg/ha/year (Bhardwaj & Kaushal 2009). According to Sur et al. (1999), reservoirs of many small- and medium-size dams constructed in this region are losing their gross storage capacity at a rate of 1–1.3% per annum, which necessitates a frequent assessment of temporal landuse change in these catchments for taking corrective conservation measures.

The soil and water assessment tool (SWAT) is a river basin or watershed scale model, which is presently one of the world's leading spatially distributed physically based continuous-time hydrological model (Arnold & Allen 1996; Arnold et al. 1998). It divides a catchment/basin into smaller separate calculation units for which the spatial variation of the major physical properties is limited and hydrological processes can be treated as being homogeneous. The total catchment behaviour is a net result of several small sub-catchments. The soil and LULC maps within the sub-catchment boundaries are used to generate a homogeneous physical property, i.e., Hydrological Response Unit (HRU). The water balance for HRUs is computed on a daily time step. The SWAT model divides the catchment into sub-units that have similar characteristics in soil, LULC, slope and that are situated in the same sub-catchment. The SWAT model has been used successfully worldwide (Arnold & Allen 1996; Srinivasan et al. 1998; Spruill et al. 2000; Rostamian et al. 2008; Briak et al. 2016; Puno et al. 2019; Islam et al. 2021; Kumar et al. 2022; Tarekegn et al. 2022) for estimating runoff, sediment yield and other hydrological processes for different LULC, crop and conditions. Several studies were also conducted in India for quantification of runoff and sediment yield using SWAT model (Tripathi et al. 2004; Jain et al. 2010; Bhatt et al. 2016) wherein the model calibration, validation and sensitivity analysis (Rossi et al. 2008; Bhatt et al. 2016) were also done using different historical hydrological datasets with various slope conditions of the catchments.

The satellite and aerial remote sensing (RS) offers rich spatial information about ever-changing dynamic earth surfaces and natural resources. The geospatial techniques, RS and geographic information system (GIS), are the most time-saving and cost-effective tools to study temporal landuse changes at the catchment scale and provide inputs to the hydrologic models. With the emergence of high-resolution satellite imagery in conjunction with GIS, it has become extremely easy to monitor even minute changes in the catchment landuse and provide an excellent platform for data analysis, processing and recovery in less time, at a lower cost, and with high precision (Kachhwala 1985; Cihlar 2000). The SWAT is a hydrological model mostly used for the simulation of catchment runoff, sediment yield and other various hydrological processes; and is widely accepted for simulation of hydrological processes and for planning land management practices all over the world (Arnold et al. 1998; Jain et al. 2010; Bhatt et al. 2016; Puno et al. 2019).

The Saleran dam catchment situated in the Shivalik foot-hills of Punjab in Hoshiarpur district is facing the serious problem of soil erosion and silt deposition into the reservoir of the dam constructed at its outlet. The reservoir is losing its storage capacity swiftly due to the high rate of the sediment inflow. There is an enormous amount of silt inflow to the Saleran reservoir. According to a report, the Saleran reservoir has lost its gross storage capacity by 24.44% till the year 2008 (IPRI 2008). Therefore, it becomes imperative to study the landuse dynamics of Saleran reservoir catchment and its impact on runoff inflow, as runoff is a major transporter of sediments into the reservoir situated at the outlet of the catchment. This would help in planning suitable conservation measures to reduce soil erosion in the catchment and hence sediment inflow to the reservoir thereby enhancing its useful life.

In the past, many studies related to landuse change and soil erosion were conducted by researchers to understand the watershed conditions situated in the Shivalik foot-hills of Punjab (Bhardwaj & Kaushal 2009; Sushanth et al. 2019). Sushanth & Bhardwaj (2019a) conducted a study in Patiala-Ki-Rao watershed located in Shivalik foot-hills of Punjab for geomorphologic characterization and demarcation of runoff potential zones using RS and GIS. The results of the study illustrate that runoff generation in the watershed was moderate to high, and at the watershed outlet, it was low to moderate (Sushanth & Bhardwaj 2019a). Kushwaha et al. (2022) performed a comparative study on morphometric analysis and RUSLE-based approaches for micro-watershed prioritization in a watershed situated in Shivalik foot-hills using RS and GIS and found that both the methods are potentially viable for prioritizing the micro-watersheds for planning and implementation of land and water conservation strategies in this region. Digra et al. (2022) recently performed a study on temporal landuse/landcover change analysis in Kotla sub-watershed of Rupnagar district (Punjab) using RS and GIS and found decreasing trend in the values of NDVI (Normalized Difference Vegetation Index) which depicts the overall decrease in vegetation in this region. Sushanth & Bhardwaj (2019b) assessed the impact of landuse change on runoff and sediment yield of Patiala-Ki-Rao watershed and found that forest is the most affected landuse among all other landuses and noticed that landuse change in the past 10 years has critically influenced the hydrology of the watershed (Sushanth & Bhardwaj 2019b).

In the study region (Shivalik foot-hills of Punjab), a large number of small- and medium-size dams/reservoirs were constructed for flood control and irrigation water storage (IPRI 2008) but to date, no study has been conducted to understand the present situation of these reservoirs. Keeping the above information in view, the present study is undertaken to (1) characterize the runoff pattern of the Saleran reservoir catchment based on geomorphometric parameters, and (2) study the impact of catchment landuse change on runoff generation and inflow to the Saleran reservoir located in Shivalik foot-hills of north-west India. The study is performed concerning the designed storage capacity of Saleran reservoir, under different climatic scenarios using geospatial techniques and SWAT hydrologic model.

The Saleran dam is one of the few small dams constructed in the Shivalik foot-hills of Punjab under the Kandi Watershed Development Project to prevent floods and store water for irrigation purposes (IPRI 2008). The construction of dam was completed in the year 1995. The salient features of the Saleran dam and its reservoir are given in Table 1.

The topographical sheet no. 44M/14 and 53A/2 on a 1:50,000 scale of the study area was collected from the office of Survey of India (SOI), Chandigarh, India. The satellite imageries of Landsat-5 and Landsat-8 for 2 March 1999, 13 March 2009 and 9 March 2019, with a spatial resolution of 30 m, were downloaded from the USGS website https://earthexplorer.usgs.gov/. The ALOS PALSAR digital elevation model (DEM) with a high spatial resolution of 12.5 m was downloaded from the Alaska satellite facility website https://search.asf.alaska.edu/#/?dataset=ALOS, which are both open access GIS data providers. The meteorological data, namely daily rainfall, maximum and minimum temperature, relative humidity, wind speed, sunshine hours and also data on annual runoff (1992–2001) of the training watershed was collected from PAU Regional Research Station, Ballowal Saunkhri and the Department of Water Resources, Hoshiarpur, Government of Punjab. The soil samples were collected from the catchment and analysed in the laboratory for particle size distribution to determine various soil parameters of the Saleran catchment.

The geomorphometric analysis of the Saleran catchment was performed by studying linear, areal and relief aspects of the catchment. The measurement of basic morphometric parameters, namely stream order, stream number, bifurcation ratio, stream length, length of overland flow, basin shape, drainage density, drainage frequency, circulatory ratio, slope and relief, etc., were computed by using ArcGIS and standard empirical formulae as shown in Table 2 using an MS Excel sheet.

The catchment landuse maps were prepared using ArcGIS software for the years 1999, 2009 and 2019. The maximum likelihood algorithm-based supervised classification technique was applied for classification of landuse classes in the Saleran catchment using Landsat satellite imageries. The ground truth data were collected from the catchment to rectify geometric and human errors in the landuse maps (Prasad & Bhardwaj 2021). After final classification, the classified images were converted to polygon format and the area under each landuse class were calculated in the GIS environment. All classified images were mapped after successful landuse classification using converted polygons (from raster to polygon) of different landuse classes in ArcGIS. The detailed information regarding the landuse of the Saleran catchment used in this study is available in Prasad & Bhardwaj (2021).

The SWAT is a catchment scale, continuous-time and spatially distributed physically based model developed to predict the impact of land management practices on water, sediment and agricultural chemical yields in complex catchments with varying soils, landuse and management conditions over long periods (Arnold et al. 1998; Anand et al. 2018; Islam et al. 2021; Kumar et al. 2022; Tarekegn et al. 2022). It is mainly developed to assess the impacts of management practices on hydrology, sediment and water quality for ungauged catchments (Risal & Parajuli 2022). The model operates on a daily time step and allows a catchment to be sub-divided into several homogeneous hydrologic response units (HRUs) with unique soil and landuse characteristics. Major components of the hydrologic cycle and their interactions considered in SWAT are surface runoff, lateral flow in the soil profile, groundwater flow, evapotranspiration, channel routing, pond and reservoir storage (Neitsch et al. 2005). The SWAT uses the NRCS (CN) curve number method to calculate surface runoff from daily rainfall. The peak runoff rate is estimated using the modified rational method (Neitsch et al. 2011) and the NRCS TR-55 method (SCS 1986). There are three methods available for the simulation of evapotranspiration, i.e., Penman–Monteith (Monteith 1965; Allen 1986; Allen et al. 1989), Hargreaves (Hargreaves & Samani 1985) and Priestley–Taylor (Priestley & Taylor 1972) methods. Infiltrated water is re-distributed into the soil profile by the SWAT percolation component, which uses a storage routing technique combined with a crack flow method to predict flow through each soil layer. The water that percolates below the soil profile is assumed to recharge to the shallow aquifer. This water can either flow laterally towards a stream/percolates into the deeper aquifer and is considered lost from the simulated system.

The Muskingum method (Overton 1966; Brakensiek 1967) was applied for runoff routing in the present study. It is based on the continuity equation and the empirical linear storage equations.

The subscripts ‘1’ and ‘2’ refer to the start and end of the routing time interval ⁠, respectively.

The catchment landuse change impact on runoff inflow to the Saleran reservoir was analysed by comparing the results obtained from the SWAT model under different landuse and climatic scenarios. The landuse change impact analysis on runoff inflow was done on annual basis. For the determination of landuse change impact, three different climate scenarios, i.e., 1999, 2009 and 2019 were used. The climate scenario 1999 represents weather conditions for the years 1995–1999 (5 years), and the scenarios 2009 and 2019 for the years 2000–2009 (10 years) and 2010–2019 (10 years), respectively. The duration of the scenario 1999 was considered keeping in view that the Saleran dam and its reservoir were built in 1995. The calibrated and validated model was applied to simulate runoff flowing into Saleran reservoir under these three different climatic scenarios for three landuses of the catchment for the years 1999, 2009 and 2019. The runoff was simulated annually to determine whether the reservoir has enough capacity to store inflowing rainwater and to scrutinize the designed reservoir storage capacity in present-day conditions. The impact of landuse change on the storage capacity of Saleran reservoir was performed by comparing annual runoff inflow into the reservoir with the designed storage capacity of Saleran reservoir. Figure 3 shows the flow chart of the methodology adopted for simulating the landuse change impact on runoff inflow to the Saleran dam reservoir.

The aerial aspects of the catchment such as drainage density, stream frequency, drainage texture, form factor and circulatory ratio were calculated. The form factor (0.39) and the circulatory ratio (0.31) indicate that the catchment is less elongated and has moderate to high runoff discharge. The same is corroborated by the elongation ratio (0.70). This type of catchment may produce a high peaked hydrograph, causing the problem of soil erosion. The compactness coefficient (1.80) also indicates that the catchment is moderately vulnerable to soil erosion risk. The drainage density in the catchment (4.55 km/km²) is quite high, indicating impermeable subsurface materials, sparse vegetation and mountainous relief, resulting in high runoff (Strahler 1964). The drainage pattern in the catchment is sub-dendritic to dendritic. The drainage texture (3.05 no./km) falls under the category of coarse drainage texture. The value of stream frequency (9.36 no./km²) is high which indicates that the topography of the catchment is not plain and that runoff water would flow quickly to the outlet.

The average slope of the catchment is around 19.5%, with 82% catchment area having a slope greater than 10% (Table 5). The relief of the catchment (220 m) for a basin length of 4.35 km indicates that the catchment runoff may have high flow velocity, low infiltration and high runoff peak. The value of the relief ratio (0.05) and moderately steep slope of the catchment may result in a moderately high velocity of runoff flow in the catchment, thereby quickly reaching the reservoir along with high amounts of sediments (Schumm 1956; Rai et al. 2017), causing sedimentation in the reservoir.

The landuse mapping results clearly show that the major portion of the catchment was covered by mixed forest followed by degraded land, water bodies and streams. The area under mixed forest accounts for 74.87, 63.18 and 65.53% of the catchment during 1999, 2009 and 2019, respectively. The area under degraded land increased from 19.31 to 30.97% of the total area (737.34 ha) during 1999–2009, while it remained almost constant during the next 10 years, i.e., 29.99% in 2009–2019. A small decrease in the area under water bodies was observed. The area under streams remained almost constant throughout the study period, which accounts for 1.57, 1.61 and 1.79% of the total area (737.43 ha) of catchment during the years 1999, 2009 and 2019, respectively. It was found that significant changes in landuse of Saleran catchment have taken place during the past 20 years (1999–2019), particularly in the case of mixed forests and degraded land. The previous studies also reported the similar effects as observed in this study such as Digra et al. (2022) studied temporal landuse/landcover change analysis in Kotla sub-watershed of Rupnagar district (Punjab) situated in the Shivalik foot-hills using RS and GIS and recorded decreasing trend in the values of NDVI, which depicts an overall decrease in vegetation in this region. Sushanth & Bhardwaj (2019b) assessed the impact of landuse change on runoff and sediment yield of Patiala-Ki-Rao watershed and found that forest is the most affected landuse among all other landuses. They concluded that landuse change in the past 10 years has critically influenced the hydrology of the watershed (Sushanth & Bhardwaj 2019b).

The training of the SWAT model was carried out in three steps, namely sensitivity analysis, calibration and validation, on a small gauged watershed (40.28 ha) located adjoining to Saleran catchment. The elevation of this training watershed varies from 342 to 416 m above the mean sea level. The slope within this watershed is divided into five classes. The maximum slope in the watershed is 48%. The major type of soil in the watershed is loamy sand. It is predominantly a forest watershed similar to the Saleran catchment.

Nineteen parameters were tested for their sensitivity to runoff simulation. According to the results obtained from global sensitivity analysis, the curve number (CN2) was found to be the most sensitive parameter, followed by surface runoff lag time (SURLAG), soil depth (SOL_Z), the available water content of soil (SOL_AWC) and saturated hydraulic conductivity of soil layers (SOL_K).

The summary statistics of observed and simulated annual runoff for the calibration period shows the values of the root mean square error (RSR) of 0.54, the coefficient of determination (R²) of 0.72, percent bias (PBIAS) of 8.58% and model efficiency (E_NS) of 71.12% were found. For the calibration period, statistics and graphs indicate a reasonably accurate simulation of surface runoff using the SWAT model.

Once the model is calibrated, proper validation is essential for model testing before it can be applied for the simulation of any hydrological process in a catchment. In this study, runoff data collected at the outlet of the training watershed from 1998 to 2001 was used for validation.

After proper calibration and validation of the SWAT model, it was applied to the main Saleran reservoir catchment for simulation of runoff inflow into the reservoir on an annual basis.

In this study, the SWAT model was used to simulate annual runoff inflow under three different landuses of the years 1999, 2009 and 2019 for the respective climatic duration, i.e., 1995–1999, 2000–2009 and 2010–2019, respectively. Table 6 shows the annual runoff inflow into the Saleran reservoir under different landuses for 25 years, i.e., from 1995 to 2019. The annual excess/deficit runoff inflow to the reservoir was also determined by comparing the designed live storage capacity of 90.12 ha-m of the Saleran reservoir.

The average annual rainfall from 1995 to 1999 was 1,409.9 mm, and the average number of rainfall events was 74. The simulated average annual runoff inflow to the reservoir from its catchment in 5 years was 200.20 mm, which amounts to 13.31% of the average annual rainfall of that period. Table 6 also indicates that the catchment generated an average of 14.2 mm of runoff per 100 mm of rainfall from 1995 to 1999. From 1995 to 1999, an excess runoff was generated in 1995, 1996 and 1998, while deficit runoff was generated in the years 1997 and 1999 as compared to the designed live storage (90.12 ha-m) capacity of the Saleran reservoir. That means 60% of the year generated enough runoff to fill the reservoir to its full capacity. The year 1995 was recorded as extreme rainfall, during which 442.72 mm of runoff was generated, which amounts to 25.68% of that year's annual rainfall, resulting in 236.32 ha-m (262.22%) excess runoff volume as compared to the designed live storage capacity of the reservoir. In the year 1997, the runoff inflow was lowest in 5 years, i.e., 48.67 mm, which caused a runoff deficit of 60.18%.

During 2000–2009, the average annual rainfall was 1,106.6 mm, and the average number of rainfall events was 57. The simulated average annual runoff inflow to the reservoir from its catchment in 10 years was 146.1 mm, which amounts to 11.51% of the average annual rainfall of that period. The catchment generated on an average 13.2 mm of runoff per 100 mm of rainfall during 2000–2009. During this period, the excess runoff was generated in 50% of the years compared to the designed live storage capacity of the reservoir. Under the landuse-2009, 2008 was recorded as extreme rainfall, during which 412.59 mm of runoff amounts to 23.92% of the annual rainfall of that year. It resulted in 214.10 ha-m (237.57%) of excess runoff inflow volume compared to the live storage capacity of the reservoir. In 2004, the lowest amount of runoff inflow was simulated, i.e., 9.91 mm, resulting in a runoff deficit of 91.89%.

The average annual rainfall during 2010–2019 was 1,020.2 mm, and the average number of rainfall events was 64. The simulated average annual runoff inflow to the reservoir from its catchment in 10 years was 120.03 mm, accounting for 11.26% of the average annual rainfall of that period. The catchment generated an average of 11.8 mm runoff per 100 mm rainfall. During this period, the excess runoff was generated in 4 years, i.e., 2011, 2012, 2018 and 2019, while deficit runoff was generated in 6 years, i.e., 2010, 2013, 2014, 2015, 2016 and 2017, compared to the live storage capacity of the reservoir. That means 40% of the years generated enough runoff to fill the reservoir to its full capacity. Under the landuse-2019, the year 2019 was the highest runoff-producing year, during which 235.41 mm of runoff was generated, accounting for 18.90% of the annual rainfall of that year. It resulted in 83.46 ha-m (92.61%) of excess runoff volume. In 2010, the lowest runoff inflow was generated (37.4 mm), resulting in a 69.40% deficit runoff to fill the reservoir (Table 6).

The results of the study indicate that the catchment generated an average of 14.2, 13.2 and 11.8 mm of runoff annually, which is equivalent to 10.47, 9.73 and 8.67 ha-m volume of water per 100 mm of rainfall during 1995–1999, 2000–2009 and 2010–2019 for the landuses 1999, 2009 and 2019, respectively that entered to Saleran reservoir. This shows that the amount of runoff inflow to the reservoir continuously decreased due to changes in landuse with time. The mean runoff inflow during 1995–2019 was 146.48 mm, 11.77% of the mean annual rainfall of 1,132.7 mm, respectively. It is also clear from Table 6 that the reservoir filled to its full capacity for 60, 50 and 40% of the years under landuse in 1999, 2009 and 2019, respectively, which also indicate a reduction in water inflow to the Saleran reservoir. There was a maximum runoff inflow deficit of around 92% in 2004 and an excess of 262% in 1995, indicating a lot of temporal variation in rainfall and hence the amount of runoff during the study period of 25 years (1995–2019). However, the mean volume of runoff inflow was 108 ha-m (146.48 mm), much higher than the designed live storage capacity of the reservoir.

In this study, the impact of landuse change on runoff inflow to the reservoir has been assessed under different climatic scenarios. To assess the catchment landuse change impact on runoff inflow to Saleran reservoir under different climatic scenarios, the model runs were made to simulate runoff inflow for the landuse 1999, 2009 and 2019 under the climate 1995–1999, 2000–2009 and 2010–2019, respectively.

Under the climate 2000–2009, the average annual rainfall of the catchment was 1,106.6 mm. The simulated average annual runoff inflow to Saleran reservoir from its catchment in 10 years was 144.85, 146.06 and 159.26 mm, and the average runoff generated per 100 mm of rainfall during this period was 13.1, 13.2 and 14.4 mm for the landuse 1999, 2009 and 2019, respectively. During this period/climate, the catchment generated 13.1, 13.2 and 14.4% of average annual rainfall as runoff flowed into the reservoir. This shows that the amount of runoff inflow continuously increases with the change in landuse as in the previous climatic duration. There was a small increase of 0.84% in a runoff for the landuse-2009 over the landuse-1999, but the increase was significant (9.95%) for the landuse-2019 over the landuse-1999. Figure 8 shows the variation in average annual runoff inflow into the reservoir for different landuses under the climate 2000–2009.

Under the climate 2010–2019, the average annual rainfall in the catchment was 1,020.2 mm. The variation in average runoff inflow for different landuses under the climate 2010–2019 is shown in Figure 8. The simulated average annual runoff inflow to Saleran reservoir from the catchment during these 10 years was 107.48, 108.56 and 120.03 mm (Table 7), which was 10.5, 10.6 and 11.8% of the average annual rainfall equivalent to 10.5, 10.6 and 11.8 mm per 100 mm of rainfall for the landuse 1999, 2009 and 2019, respectively (Table 7). This indicates that the amount of runoff inflow to the Saleran reservoir continuously increases with the change in landuse, similar to previous climatic duration. There was a marginal increment of 1% in the total runoff for the landuse-2009 over the landuse-1999, but the increase was significant (11.67%) for the landuse-2019 over the landuse-1999 under the climate 2010–2019.

In the past, Sushanth & Bhardwaj (2019b) applied Water Erosion Prediction Project (WEPP) model to assess the impact of landuse change on runoff and sediment yield in Patiala-Ki-Rao watershed situated in the Shivalik foot-hills of Punjab. They reported that forest is the most affected landuse among all other landuses in the watershed. This similar pattern is observed in the present study. The impact of landuse change on annual runoff in Patiala-Ki-Rao watershed was assesed for the two landuses. i.e., 2006 and 2016 under two climate scenarios, i.e., Climate-2006 and Climate-2016. The results of the study showed that the annual runoff increased by 20.3 to 28.6% under Climate-2016 as compared with that under Climate-2006 was due to higher rainfall. It was conlcuded that landuse change in the past 10 years has critically influenced the hydrological processes of the watershed (Sushanth & Bhardwaj 2019b). Similarly, in the present study, we observed that mixed forest is major landuse affected in the last 20 years (1999–2019) which causes significant variation in the amount of runoff inflow to the Saleran reservoir. The increase in annual runoff inflow is significant for the landuse-2019 over the landuse-1999 as observed under all the three climatic scenarios which are quite similar to the result obtained by Sushanth & Bhardwaj (2019b). It is also noticed that rainfall is a critical factor in runoff simulation because there is linearity in the rainfall–runoff relationship (Ben-Zvi 2020). Therefore, the study of landuse change along with the climate has significant meaning in the planning of watershed management and natural resource conservation.

The current study is performed in a single reservoir catchment, i.e., Saleran. However, in this region (Shivalik foot-hills of Punjab), there are more than 29 small to medium-size reservoirs/dams constructed for flood control and irrigation, and many of them are facing the critical issues similar to Saleran dam catchment. Furthermore, there is a necessity of large dataset such as runoff and sediment yield to validate a model effectively which is a major limitation to apply models like SWAT in this region. Therefore, other hydrologic models can also need to be tested. Apart from that, there is a need to comprehensively study the reservoirs' catchmentd so that proper planning can be done for the management of natural resources. Also, there is a need to develop comphensive database about the present situations of these dams/reservoirs and their catchments, so a proper decision can be taken for effective functioning of these reservoirs and prioritization would be done based on the severity. In the past, many researchers conducted studies on land degradation, morphology and impact of landuse change in this region as discussed in this paper but none of them attempted to study reservoir catchments. Therefore, this study will give a path to focus on critical issues being faced by several dams/reservoirs in this region.

The geomorphometric analysis exhibits that the Saleran catchment is less elongated in shape, having a coarse drainage texture (3.05 no./km), stream network having quite a high stream frequency (9.36 no./km²), drainage density (4.55 km/km²) and high relief (220 m). The average slope of the catchment is 19.5%, with 82% catchment having a slope higher than 10%. However, soils in the catchment are light in texture (sandy loam to loamy sand), indicating that the catchment would generate a moderate amount of runoff with a relatively moderate peak and base width of the hydrograph. The landuse analysis results revealed that the major portion of the catchment was covered under mixed forests followed by degraded land, water bodies and streams throughout the study period. Still, there is a significant reduction in the area under mixed forests and an increase in the area under degraded land. The simulated results show that the catchment generated an average of 14.2, 13.2 and 11.8 mm of runoff annually which is equivalent to 10.47, 9.73 and 8.67 ha-m volume of water per 100 mm of rainfall during 1995–1999, 2000–2009 and 2010–2019 for the landuses 1999, 2009 and 2019, respectively that entered into the reservoir. This shows that the amount of runoff inflow to the reservoir continuously decreased due to changes in monsoonal rainfall with time. The mean runoff inflow during 1995–2019 was 146.48 mm, 11.77% of the mean annual rainfall of 1,132.7 mm. The reservoir filled to its full capacity for 60, 50 and 40% of the years under landuse in 1999, 2009 and 2019, respectively, indicating a reduction in water inflow to the Saleran reservoir. There was a maximum runoff inflow deficit of around 92% in 2004 and an excess of 262% in 1995, indicating a lot of temporal variation in rainfall and hence the amount of runoff during the study period of 25 years (1995–2019). Hence, an appropriate conservation strategy needs to be developed and adopted in the Shivalik foot-hills for managing catchment landuse for sustainable reservoir water supply.

Both the authors contributed to the study's conception and design. Vishnu Prasad performed material preparation, data collection and analysis under the supervision of Anil Bhardwaj. Vishnu Prasad wrote the first draft of the manuscript and it was reviewed by Anil Bhardwaj. Both the authors commented on previous versions of the manuscript. Both authors read and approved the final version of the manuscript.

The corresponding author received National Talent Scholarship (ICAR-NTS) from the Indian Council of Agricultural Research, New Delhi, India, to conduct this study as a part of master's degree program at Punjab Agricultural University, Ludhiana, Punjab, India.

The authors acknowledge the Head, Department of Soil and Water Engineering, Punjab Agricultural University, Ludhiana, for providing necessary research facilities. The authors are highly grateful to the Director, Punjab Remote Sensing Centre, Ludhiana, for providing the required analytical facilities for carrying out this work. The authors are also thankful to the Regional Research Station (Punjab Agricultural University), Ballowal Saunkhri and the Department of Water Resources, Government of Punjab, India, for providing necessary data and support for the timely completion of this work.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.