Assessment of runoff variability under various factors using SWAT model in a Tunisian semi-arid watershed

Abstract

This research aims to establish a correlation between prolonged shifts in precipitation patterns and runoff and to delineate spatial surface runoff distribution within the Wadi Rmel watershed using the Soil and Water Assessment Tool (SWAT) model. The calibration and validation processes reveal good agreement between measured and simulated flows with a monthly time step. Significantly, the Nash-Sutcliffe Efficiency (NSE) values of 0.61 and 0.83, accompanied by the coefficient of determination (R2) values of 0.66 and 0.85 for calibration and validation, respectively. It clearly indicates a strong agreement between the observed and simulated streamflow data. Subsequently, an exploration of the spatiotemporal variability of runoff underscores the synchronous relationship between runoff and monthly and annual precipitation fluctuations between 2000 and 2020. The values of R2 were 0.7 and 0.79 for monthly and annual studies, respectively, confirming a strong correlation between the two cited variables. Additionally, anthropogenic impacts, particularly Soil and Water Conservation (SWC) measures, were investigated. A high R² coefficients (0.91 for linear relationship) were obtained and showed the high relation between SWC and runoff. In fact, the combination of SWC techniques, vegetation cover, topography and precipitation led to a reduction of 44% of surface runoff, during 2015-2020. Spatially, higher runoff occurred in central and eastern parts due to agriculture and hilly terrain, while forested areas and gentle slopes exhibited lower runoff. In conclusion, this study underscores the model's utility in characterizing arid catchment surface runoff. Its capacity to simulate and assess runoff enriches water resource management comprehension in such contexts.