COMPARISON OF HYDROLOGICAL SIMULATION GROUNDWATER RECHARGE PREDICTIONS

Abstract

Groundwater recharge is a critical component of water resource management, yet accurately estimating it requires the use of robust hydrological models. This study compares three widely used hydrological simulation models—SWAT, MODFLOW, and HEC-HMS—to evaluate their performance in predicting annual groundwater recharge. The methodology involves selecting appropriate models based on their capabilities, calibrating them using local data, and performing simulations for the period 2015–2019. The study area consists of a region with varied land use and climate conditions. Input data, including precipitation, temperature, and soil properties, were collected and preprocessed to fit the requirements of each model. Calibration was performed to minimize the error between simulated and observed groundwater levels and streamflow, followed by model validation for 2020–2022. Results show that MODFLOW provided the most accurate predictions of groundwater recharge, with a Nash-Sutcliffe Efficiency (NSE) of 0.84 and Root Mean Square Error (RMSE) of 18.2 mm/year. SWAT also performed well, accurately reflecting the impact of land use on recharge. However, HEC-HMS showed the lowest performance among the models, particularly in estimating recharge. Sensitivity analysis identified soil permeability as the most influential parameter in MODFLOW, while land use change had the strongest effect on HEC-HMS. The study concludes that while MODFLOW is best suited for groundwater recharge prediction in this study area, a multi-model approach could enhance model reliability and capture various hydrological processes, especially in regions with dynamic land use and climate variability.