Time-Dependent Reservoir-Coefficient Framework for Forecasting Sedimentation in Data-Scarce Hill Reservoirs

Sedimentation is a major factor that threatens the sustainability of water infrastructure by compromising reservoir storage capacity and dam multifunctionality. Therefore, accurate prediction of sediment distribution is critical for reservoir management. However, limited information is available on the comparative performance of empirical approaches and their suitability under different reservoir conditions. This study aimed to establish a framework for long-term environmental monitoring of water infrastructure in semiarid regions, which are characterized by limited data availability. Six empirical methods were evaluated to predict the sedimentation dynamics in a Type-III (hill) reservoir. Model calibration relied on the 1970 survey, with the 2013 and 2022 surveys retained for independent validation. The methods were compared to identify the most accurate approach for forecasting the reservoir’s future useful capacity in 2030, 2050, and 2100. The results showed substantial differences in accuracy among the methods: root mean squared error (RMSE) ranged from 3.38 to 17.51 Mm 3 (2013) and from 3.28 to 20.07 Mm 3 (2022), while Nash–Sutcliffe efficiency (NSE) ranged from 0.924 to 0.997 (2013) and from 0.88 to 0.997 (2022). The modified noniterative approach incorporating a time-dependent reservoir coefficient achieved high accuracy, with a relative error within 3%. Projections under three sedimentation scenarios indicate a considerable reduction in useful storage capacity, reaching 51% to 70% by 2050. These findings highlight the need for dynamic modeling tools under increasing erosion rates, and provide a decision-support framework for water resource managers facing data scarcity challenges in semiarid hill reservoirs.