A scenario-based analysis of selected best management practices for reduced sediment and nutrient yield in the watershed located in the Shivalik hills, India

Runoff-induced soil erosion and its subsequent disposal into the river network through watershed outlets is a serious and continuous global environmental problem affecting the quality of water, soil and land resources (Bhattacharyya et al. 2015; Pandey et al. 2021). It is the process by which surface soil is entrained from one place and accumulates in a new location due to the impact of rainfall and runoff. The soil erosion and its transportation process cause degrading surface soil quality, a regime problem in the stream channels, a decrease in the capacity of stream channels, lakes and reservoirs by deposition of sediments, and an increase in the costs of maintaining water conveyance structures. It is, therefore, required to take some preventive measures to save this runoff and control erosion during the rainy season. The best management practices (BMPs) are strategies for conserving soil and water, as well as social and cultural activities, that have been identified as effective and feasible ways of reducing and controlling runoff-induced sediment and nutrient transport in a watershed (Debo & Reese 2003). Evaluating the efficacy of a BMP by collecting monitoring data from the fields is both expensive and time consuming especially for large watersheds with varying land use and soil features. In this regard, watershed models stand out as valuable tools as they have the capability to analyze watershed processes and their interactions, as well as simulate BMPs at a relatively low cost and without investing the time required to install and monitor BMPs in the field (He 2003; Liu et al. 2017). According to Khalid et al. (2016), SWAT is a well-established, physically based distributed model among other hydrological models to analyze the effectiveness of land management practices and BMP implementation in large and complex watersheds (Chaubey et al. 2010; Tyagi et al. 2014). As a result, the SWAT model has emerged as one of the popularly used river basin scale models for hydrologic and environmental evaluations. However, the performance of a BMP is basically affected by several factors, which differ significantly from location to location within any watershed. Therefore, it is difficult to develop a particular set of BMPs that is best for all watersheds (Maeda et al. 2018). Rather, it is necessary to determine the existing condition of the watershed and choose BMPs that will be effective to control the sediment and nutrient transport within the basin and have reasonable implementation and maintenance costs.

Sushanth et al. (2019) assessed critical erosion-prone areas in Patiala-ki-Rao watershed using geographic information system (GIS) and water erosion prediction project (WEPP) models. The results of their study showed that built-up, agriculture and fallow lands with high slopes and sand content are more prone to soil erosion. Simulating the sediment yield (SY) at hillslope level provided an idea about the critical erosion-prone areas within the watershed, and thus appropriate management strategies can be planned for the sustainable livelihoods of the inhabitants in the watershed. Yousuf et al. (2021) used the WEPP model to simulate the impact of check dams and sedimentation basins on runoff and SY. They concluded that channel erosion is predominant in the watershed and check dams are more efficient in controlling runoff and SY than sedimentation basins. However, the coupled implementation of both interventions was much more effective than the individual implementation of each intervention.

For a small watershed Patiala-Ki-Rao in Shivalik Hills in Northern India, which is selected for the present study, a few BMPs are already in place as ongoing research. These BMPs have been institutionalized because one of the biggest concerns in the watershed is its erodible nature, which becomes prominent during monsoon season. A large portion of monsoon rainfall (35–45%) is lost as runoff in the torrents causes large-scale erosion in the area. A survey has revealed that 90% of the villagers have agriculture as their primary occupation and the key constraints on agriculture production in these villages are deep groundwater, shortage of surface water, low water & nutrient holding capacity of the soil, degraded and undulating lands and poor socio-economic conditions of the residents. It has been found that large amounts of rainwater go as runoff in a small duration and create erosion and flooding in lower-lying areas, resulting in water loss and land degradation. Along with the eroded soils, pesticides and fertilizers from the agricultural lands are also transported and find their way into the natural streams causing deterioration of downstream water quality Therefore, a soil and water conservation plan is needed in the area, in which management decisions are based on physical principles and scientific concepts. However, before implementing management practices, it is also necessary to assess their effectiveness in controlling soil loss and nonpoint source (NPS) pollutants such as nutrients. Therefore, a methodology is needed to evaluate the effectiveness of the various soil water conservation measures in the selected watershed. The present paper thus attempts to (1) evaluate the impact of different BMPs as well as a few hypothetically implemented BMPs on reducing SY and nutrients loss using SWAT model and, (2) evaluate the most efficient BMPs using cost–benefit analysis (CBA) to reduce the input cost for sustainable watershed management.

SWAT is a continuous-time, physically based, semi-distributed river basin or watershed scale model (Arnold et al. 1998). It is a watershed hydrological transport model comprising weather, surface runoff, return flow, percolation, evapotranspiration, transmission losses, pond and reservoir storage, crop growth and irrigation, groundwater flow, reach routing, nutrient and pesticide loading, and water transfer components (Ronco et al. 2017). The model works on hydrological response units (HRUs) and hydrological processes are simulated by using water balance equation.

The first 5 years (2007–2011) of the 9-year period were used for calibration, and the subsequent 4 years (2012–2015) were used for model validation. The model was calibrated by using the values of 20 sensitive parameters that were identified during the sensitivity analysis. The model was calibrated for stream flow and SY simultaneously for four gauged micro-watersheds from 2007 to 2012 at a monthly time step. The performance evaluation was done on the basis of three performance indices: coefficient of determination (R²), Nash–Sutcliffe efficiency (NSE) and Percentage Bias (PBIAS), where NSE was used as the major objective function. It was observed in the study that, among all the four gauged micro-watersheds, micro-watersheds-IV predicted better results for stream flow and SY. The R², NSE and PBIAS values for the microwatershed-IV were 0.94, 0.91 and 13.9 for the calibration period and 0.92, 0.90 and 9.5 for the validation period, respectively, for stream flow. Whereas for SY, the R², NSE and PBIAS values were 0.92, 0.88, and 27.8 for calibration period and 0.88, 0.80 and 28.4 for validation period, respectively. The statistical analysis clearly predicts that the model performance for both calibration and validation for monthly stream flow and SY was good as per performance rating criteria of hydrological models (Premanand et al. 2018). Therefore, the predicted flow and SY of the best-chosen microwatershed-IV was used to validate the stream flow and SY for selected ungauged watershed and sub-watersheds.

In the present study, the SWAT model has already been calibrated and validated for stream flow and SY for the selected area and for the four gauged micro-watersheds. The data on water quality in terms of nutrients were not available for the selected watershed. In a similar condition, literature provides several studies (e.g., refer Nelson et al. 2006; Barlund et al. 2007; Abouabdillah et al. 2014; Smith et al. 2014; Himanshu et al. 2019; Busico et al. 2020; Abu-Zreig & Hani 2021) where efficiency of management practices on variables other than the variable for the model is calibrated due to the lack of observed data. Hence, although sediment is the pollutant of concern in the study area, the effectiveness of management practices on nutrients was also evaluated to test the SWAT model's applicability in evaluating the efficacy of management decisions. The simulated SY and nutrients, i.e., organic nitrogen (ON), nitrate as nitrogen (NO₃), total nitrogen (TN), organic phosphorus (OP) and total phosphorus (TP) for micro-watersheds with BMPs were compared to micro-watersheds and sub-watersheds without BMPs to evaluate the efficiency of BMPs in reducing the sediments and nutrients entering into the stream water.

Water quality parameters under two different scenarios with the combinations of the gauged-controlled (GC)/gauged-uncontrolled (GUC) and the ungauged-controlled (UGC)/ungauged-uncontrolled (UGUC) microwatersheds and sub-watersheds were simulated using SWAT to assess the impact of management practices. Among four gauged micro-watersheds MW-I, MW-II and MW-III implemented with BMPs as given in Table 2 are controlled micro-watersheds, whereas gauged microwatershed MW-IV and ungauged sub-watersheds SW-I and SW-II are uncontrolled microwatersheds and sub-watersheds where no control measure is implemented to control the runoff and pollutants. The existing BMPs were evaluated by directly changing the parameter (USLE_P) within the calibrated SWAT model. The effectiveness of existing BMPs analysed by comparing the outputs of sediment and nutrient yield under these two scenarios. In either of the two scenarios, simulated monthly sediment and nutrient loads were compared.

Scenario 1: GC/GUC

In the first combination, two gauged microwatersheds nested within the selected ungauged watershed were chosen and compared. The gauged microwatershed-I was applied with engineering, vegetation and fencing gauged and microwatershed-IV was not implemented with any control measure.

Scenario 2: UGC/UGUC

In the second combination, the selected ungauged sub-watershed with no control measures was compared with microwatershed-I (MW-I), where control measures are applied and hypothesized as ungauged.

In the selected watershed, check dam, plantation and fencing are the existing BMPs used to control runoff, soil loss and NPS pollution. Evaluation of these BMPs predicted their effectiveness to control runoff, soil loss and NPS pollution using SWAT model. Taking into account the efficacy of existing BMPs, some potential BMPs were hypothetically simulated to check their effectiveness. In this study, the impacts of contouring, terracing, grassed waterways and filter strip were hypothesized to check their effectiveness in reducing sediment and nutrient yield. The study thus aims to evaluate the impact of these soil and water conservation interventions prior to their application in the watershed to develop sustainable irrigation strategies. SWAT was run on a monthly basis from 2004 to 2015 (excluding the initial 3 years from 2004 to 2006 as a warm-up period). The BMPs, i.e., contouring, terracing, filter strips, and grass waterways, were hypothetically implemented using the scheduled management operations (.ops) file. The impacts of these BMPs were evaluated by compiling monthly SY, ON, nitrate as nitrogen (NO₃-N), TN, OP and TP losses for all the 9 years.

The simulated monthly loadings for sediment and nutrients when no BMPs were implemented were used as the baseline loading conditions to which the simulated loads from hypothetically implemented BMP scenarios were compared. The monthly load difference between a hypothetically implemented BMP scenario and the baseline scenario was used to determine the load reduction achieved by BMP implementation. The effectiveness of hypothetically implemented BMPs was carried out by comparing the monthly SY, ON, nitrate as nitrogen (NO₃), TN, OP, and TP under these two scenarios as:

Scenario 1 (Baseline scenario): The baseline simulation was considered when no BMP was applied in the watershed.

Scenario 2 (Hypothetically implemented BMPs Scenario): In this scenario, simulations were modeled hypothetically by implementing management practices to quantify the impact of contouring, terracing, filter strips, grass waterways on sediment and nutrient loadings.

The results of SY and nutrient load percentage reduction with the BMP application are shown in Table 3 for the two selected scenarios. The study shows the effectiveness of management practices when existing BMPs were evaluated. The results predicted that the percentage of suspended sediment and the nutrient load was low in the micro-watersheds where management practices were implemented compared to untreated micro-watersheds and sub-watersheds.

The percentage sediment and nutrient reduction with the hypothetical application of contouring, terracing, grassed waterways and filter strips at the outlet of selected ungauged watershed as shown in Table 4 predicts that contouring and terracing reduced the average sediments yield by 49.9%, while grassed waterways reduced the average sediments by 75.9% and filter strips reduced the average sediments by 11.23% from baseline simulation. Whereas, the percentage average decrease in TN and TP with the implementation of contouring, terracing and grassed waterways was 34.12, 29.97, 94.6, 40.13, 35.76, and 98.1%, respectively.

The results of the present study clearly predict that grassed waterway is the most effective BMP for managing sediments and nutrients entering into the stream. It was also observed that filter strips have the least significant impact on reducing sediments and nutrients whereas contouring and terracing also resulted in a decline in sediments and nutrient yield in the selected watershed. These estimates can help the decision-makers to determine how effective these BMPs might be when implemented in the untreated and ungauged watershed. Taking into account the effectiveness of the selected BMPs, soil water conservation techniques can be planned in such a way as to control land degradation, and improve water availability and crop production by implementing them the most economically acceptable way.

It is clear from the evaluation of existing and hypothetical BMPs implementation that the percentage of suspended sediment and the nutrient load was high in the micro-watersheds and sub-watersheds where management practices were not applied, as compared to treated micro-watersheds. The study indicated that conservation practices are needed in the selected watershed to reduce soil erosion and nutrients for the sustainability of the selected watershed on the basis of priority. Furthermore, before implementing conservation practices, there is a need to evaluate the most efficient BMPs using CBA to reduce the input cost and resources.

In the present study, the reduction of sediment and nutrients yield with the hypothetical application of BMPs was simulated. The results predicted that the most effective BMP is the grassed waterways followed by contour farming, terracing and filter strips to reduce the intrusion of sediments and nutrients into the stream. However, the implementation of terracing is labour-intensive and expensive as compared to contouring, and the simulated results also predicted that contouring was more effective in reducing sediment and nutrients as compared to terracing, as shown in Table 4. Therefore, grass waterways and contouring are more effective and economical in reducing the soil and nutrient loss among the four hypothetical BMPs evaluated for the selected watershed. Whereas to calculate the amount of TN and total phosphorous (TP) that gets reduced with the application of these BMPs in monetary terms, the average retail prices of the commonly used fertilizers urea and Di-Ammonium Phosphate (DAP) used by local farmers were obtained through market surveys. The price for the elemental forms of nitrogen and phosphorus was calculated, and from the calculated average prices of elemental nitrogen and phosphorus, input costs per nutrient for inorganic fertilizer were assessed. However, the actual cost for inorganic fertilizer will be somewhat higher than that of the elemental nutrients due to manufacturing and transportation.

In the absence of conservation practices, approximately 234.81 kg/ha of TN and 86.38 kg/ha of TP were required to replenish the nutrient loss. It was estimated that around 54.48 USD would be spent to replace the lost macronutrients nitrogen (N) and phosphorous (P) with the inorganic fertilizers, as shown in Table 5. The on-site cost of erosion would, however, be higher than this estimate. This is because the cost of macro nutrient potassium (K) (not simulated in SWAT model), the cost of conserving soil and water and the cost of compensating for the reduced crop production have not been included in this analysis. These losses lead to substantial environmental degradation. The consequences of the ensuing soil erosion along with nutrients decrease soil productivity, which makes crop production unsustainable.

The Upper Patiala-Ki-Rao watershed in Shivalik hills located in Northern India is economically poor land area that is degrading over a period of time. This is because of its highly erodible nature, loss of soil fertility and nonpoint pollution during the monsoon months when its rivulets carry the seasonal flow. A few BMPs are already in place to control runoff and soil loss in four gauged micro-watersheds nested within the selected ungauged sub-watersheds. However, the study on the efficacy of these BMPs in controlling erosion and nutrient loss is not reported. Hence, in an attempt to propose sustainable watershed measures, a scenario-based evaluation of BMPs was performed. The SWAT model was first calibrated and validated for sub-watersheds that are gauged following which the model was also calibrated for ungauged sub-watersheds using the area ratio method of regionalization. Once calibrated, the effectiveness of existing BMPs, such as check dams, plantation and vegetation, in four gauged micro-watersheds was predicted. Taking into account the efficacy of existing BMPs, some potential hypothetical BMPs, such as contouring, terracing, grassed waterways and filter strips were also simulated. Based on CBA, it was concluded that the grassed waterways and contouring were found to be effective and economical in reducing sediments and nutrients among the four hypothetical BMPs. The results predicted that the percentage of suspended sediment and the nutrient load was low in the micro-watersheds where management practices were implemented compared to untreated micro-watersheds and sub-watersheds. In the absence of conservation techniques, 234.81 kg/ha of TN and 86.38 kg/ha of TP were required to restore nutrient loss. The cost of replacing the lost macronutrients N and P with inorganic fertilizers was calculated to be 55 USD. The study indicated that conservation practices are needed in the selected watershed to reduce soil erosion and nutrients for the sustainability of the selected watershed.

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

The authors declare there is no conflict.