Hydrological Modelling and Sustainable Management of Murredu and Peddavagu Watersheds in India Using Geospatical and Artificial Intelligence Techniques

Abstract

Water is a vital resource for life, essential for drinking, agriculture, industry, and newlineecosystems. This study focuses on the Murredu and Peddavagu tributaries of the Godavari newlineRiver, aiming to improve water efficiency and protect these watersheds for future generations. newlineUsing the soil and water assessment tool (SWAT), artificial intelligence (AI), and other newlinemethods, the research analysed rainfall-runoff behavior, identified artificial recharge sites, and newlineprioritised sub-watersheds for sustainable management. The study employed the SWAT model newlinefor monthly rainfall-runoff modelling in the Murredu (1996 2005) and Peddavagu (1987 newline2010) watersheds, identifying 17 and 18 sensitive parameters, respectively, using the soil and newlinewater assessment tool-calibration and uncertainty programming (SWAT-CUP). For the newlineMurredu watershed, the model demonstrated strong performance during calibration newline(coefficient of determination (R²) = 0.77, Nash-Sutcliffe efficiency (NSE) = 0.77, root-meansquare newlineerror (RMSE) = 7.59 m³/s, mean absolute error (MAE) = 4.4 m³/s, RMSE-observations newlinestandard deviation ratio (RSR) = 0.48, percent bias (PBIAS) = -13.34) and validation (R² = newline0.84, NSE = 0.80, RMSE = 10.67 m³/s, MAE = 6.83 m³/s, RSR = 0.44, PBIAS = -13.3), with newlineevapotranspiration (612.3 mm/year) accounting for 49.85% of water loss and surface runoff newlineaveraging 178.5 mm/year. Similarly, in the Peddavagu watershed, calibration (R² = 0.63, NSE newline= 0.62, RMSE = 50.64 m³/s, MAE = 19.65 m³/s, RSR = 0.61, PBIAS = -0.10) and validation newline(R² = 0.75, NSE = 0.71, RMSE = 27.59 m³/s, MAE = 12.14 m³/s, RSR = 0.53, PBIAS = -11.11) newlineresults indicated evapotranspiration (571.6 mm/year) as the largest water balance component newline(49.24% loss), with surface runoff averaging 224.13 mm/year. The SWAT model effectively newlinepredicted low flows but struggled with peak flow estimation in both watersheds, underscoring newlinethe importance of evapotranspiration and surface runoff variability for sustainable water newlineresource management. newlineThe SWAT model, while effective for rainfall-runoff model