A Coupled SWAT-MCDM Framework for Delineating Potential Rainwater Harvesting Zones in a Tropical Semi-Arid Basin

In semi-arid tropical regions, water scarcity poses a formidable challenge to agricultural productivity and regional water security. For this, Rainwater Harvesting (RWH) could be a better alternative. However, the conventional approaches of identifying the best RWH sites often overlook the complex spatio-temporal dynamics of hydrological processes and critical socio-economic constraints. To deal with this limitation, this study presents a framework that synergistically integrates the Soil and Water Assessment Tool (SWAT) hydrological model with a geospatial Multi-Criteria Decision-Making (MCDM) approach. The advocated approach has been verified in the Daund watershed (11,205 km²) in western India, as a test case. In reproducing the observed daily streamflow hydrographs at the basin outlet, SWAT is first calibrated with the coefficient of determination (R²) and Nash-Sutcliffe efficiency (NSE) of 0.70 and 0.67, respectively; which are of R² = 0.66 and NSE = 0.63 during validation. Subsequently, using the Analytic Hierarchy Process framework, thematic layers of ten critical biophysical parameters, viz. rainfall, slope, elevation, soil texture, soil depth, land use/land cover, drainage density, geomorphology, curvature, and SWAT-derived runoff coefficients are used to create a comprehensive potential RWH zoning map. This potential map is further refined by incorporating socio-economic exclusion criteria, such as buffer zones around drainage networks, roads, urban centres, and geological fault lines, ensuring the proposed structures' practical feasibility and safety. The final RWH potential zonation revealed that approximately 29% of the watershed area is highly suitable, 47% moderately suitable, and 24% poorly suitable for RWH interventions. The predictive robustness of the advocated framework has been rigorously validated against the locations of surveyed 494 RWH structures in the watershed, achieving a Receiver Operating Characteristic (ROC) Area Under the Curve (AUC) of 0.77, signifying high accuracy. This research unequivocally demonstrates that integrating a hydrological model like SWAT with the MCDM framework could enhance the reliability of potential RWH mapping that could be upscaled to other tropical basins worldwide confronting similar hydro-climatic challenges.