Assessment of Rainfall-Runoff Dynamics and Reservoir Inflow Response under Rainfall Scenarios in the Yeleswaram Catchment Using HEC-HMS
Abstract
A continuous daily rainfall–runoff model of the Yeleswaram catchment (2378.03 km²), which drains into the Yeleru reservoir in Kakinada district, Andhra Pradesh, India, was developed using HEC-HMS version 4.2.1. Daily rainfall and meteorological data from NASA POWER, together with FAO-56 Penman–Monteith potential evapotranspiration, were used to drive the model for the period 2019–2024. The Soil Moisture Accounting (SMA) loss method and the Clark Unit Hydrograph transform were adopted, with model geometry and curve number characteristics derived from a Digital Elevation Model, land use/land cover and hydrologic soil group data in a GIS environment. The model was calibrated against observed daily reservoir inflow for 2019–2022 and validated for 2023–2024, achieving a Nash–Sutcliffe Efficiency (NSE) of 0.117 and 0.170, and a coefficient of determination (R²) of 0.287 and 0.286, respectively. The model reproduced the calibration-period peak reasonably well but underestimated the major validation-period peak and inflow depth, indicating limited predictive performance rather than strong forecasting skill. Five hypothetical rainfall scenarios, obtained by uniformly scaling the baseline daily rainfall by −30%, −15%, 0%, +15% and +30%, were simulated using the calibrated model to assess the sensitivity of reservoir inflow to rainfall variability. The results show a nonlinear response, with peak reservoir inflow decreasing by 54.51% under the −30% scenario and increasing by 62.41% under the +30% scenario relative to the baseline peak of 723.9 m³/s, while the corresponding simulated inflow depth changed by −84.98% and +203.03%. The study demonstrates a workflow for GIS-based continuous HEC-HMS modelling of a data-limited reservoir catchment and highlights the usefulness of rainfall-scenario simulation for examining potential changes in reservoir inflow relevant to water resources planning.