Influence of Spatial Variability on Flood Prediction

Accurate flood forecasting remains a significant challenge within hydrology, primarily due to the pronounced spatial heterogeneity of rainfall and catchment characteristics. This challenge is particularly evident in large river basins influenced by monsoons, where traditional lumped models frequently fail to capture localized hydrological responses effectively. This research examines the role of spatial variability in influencing flood prediction accuracy in the Kabini River Basin, India, by systematically evaluating the impact of sub-basin resolution in hydrological modeling. Four Hydrologic Engineering Center – Hydrologic Modeling System (HEC-HMS) models were developed using 4, 8, 16 and 32 sub-basin delineations derived from 30-meter resolution Digital Elevation Model (DEM) data. Rainfall observations were collected from NASA datasets, and Model parameter calibration was undertaken using observed streamflow data from the GEOGloWS Hydroviewer. The hydrological modeling framework included the SCS Curve Number method for estimating losses, the SCS Unit Hydrograph for generating runoff, and Muskingum routing for the propagation of flow. Model performance was assessed using Nash-Sutcliffe Efficiency (NSE), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). The results indicated a clear enhancement in predictive accuracy with increasing spatial discretization. The configuration with 32 sub-basins yielded the best performance (NSE = 0.882, RMSE = 136.8, MAE = 91.1), while the model with 4 sub-basins demonstrated significantly lower accuracy (NSE = 0.563). These findings confirm that a finer sub-basin representation significantly improves the model's ability to capture spatial rainfall variability and the hydrological response of the basin. The study concludes that incorporating spatial heterogeneity through optimized sub-basin delineation markedly enhances the reliability of flood forecasting. This research contributes a practical methodological framework for balancing model precision and computational efficiency, thereby supporting improved flood risk assessment and water resources planning in data-limited river basins influenced by monsoons across the globe.