Performance Evaluation of Rain-on-Grid Flood Simulations with a Collocated Co-kriging Reference in a Mamminasata Urban Watershed, Indonesia

Abstract

Abstract Land use change and intensifying extreme rainfall have increased flood risk in the Mamminasata metropolitan region, while flood depth information remains scarce. This study evaluates Rain-on-Grid (RoG) flood simulations using a reference depth surface constructed with Collocated Co-Kriging (CCoK) and develops an event-based hazard index for the 11 February 2025 event. RoG was driven by bias-corrected GPM IMERG rainfall, several digital elevation models (DEMs), and surface parameters derived from land cover. The CCoK-based reference surface in the Tallo Watershed was generated from field-surveyed flood depths and topographic covariates and outperformed IDW, ordinary kriging, and kriging with external drift, with RMSE = 0.491 m and MAE = 0.379 m (absolute error ≤ 0.5 m for 73.9% of flooded points). The RoG simulation using the enhanced DEM and calibrated Manning and Curve Number parameters achieved the best agreement with CCoK reference (POD of 0.86, RMSE of 0.57 m and MAE of 0.46 m) and reduced unrealistic extreme depths (max depth 6.54 m versus 9.91–21.81 m for the other DEM candidates). The hazard index that combines depth, depth multiplied by velocity, and duration reveals widespread shallow flooding and localized pockets of higher, long-lasting hazard on downstream floodplains. These findings indicate that integrating CCoK and RoG is suitable for model evaluation and event-based hazard mapping in data-scarce basins and can support disaster management and early warning for flood events.

Access to Document

Other files and links

• Link to publication in Scopus
• Open Access Version Available

Fingerprint