RELIABILITY ASSESSMENT OF CALIBRATED CHIRPS-BASED WATER BALANCE MODELING ACROSS TWO TROPICAL RESERVOIR CATCHMENTS

Abstract

Seasonal rainfall variability in tropical irrigation systems complicates reservoir operation by affecting inflow reliability and long-term storage sustainability. Although CHIRPS is useful for data-scarce basins, most studies remain limited to single-catchment evaluation. This study assesses reliability in calibrated CHIRPS-based water balance modeling in two tropical reservoir catchments characterized by contrasting spatial scales and irrigation requirements. CHIRPS rainfall data for 2004-2023 were statistically calibrated against observed data and assessed using the correlation coefficient (R), Nash-Sutcliffe Efficiency (NSE), and RSR. Corrected rainfall inputs were employed in the F.J. Mock rainfall-runoff model with Genetic Algorithm parameter optimization. Dependable discharge (Q20, Q50, and Q80) was estimated using Weibull probability analysis, while long-term surplus-deficit dynamics were evaluated through water balance modeling. The findings indicate that both catchments have steady discharge modeling performance (NSE > 0.70) and strong calibration reliability (R > 0.75). Although the overall frequencies of surplus deficits were comparable, watershed size and irrigation demand influenced dry-season resilience and the seasonal concentration of water deficits. These results show that satellite-based hydrological modeling combined with demand-aware reservoir management can support seasonal water security and adaptive reservoir operation in tropical data-scarce environments. Beyond site-specific performance evaluation, this study proposes a cross-catchment reliability assessment framework to examine the transferability of calibrated CHIRPS-based water balance modeling across heterogeneous tropical reservoir systems.