Evaluating the satellite-derived DEM accuracy with rain-on-grid modeling for flood hydrograph prediction of Katulampa Watershed, Indonesia

Abstract

ABSTRACT In this study, we investigated the applicability of five open-access, satellite-derived DEMs with resolutions of ∼90 m (MERIT-Hydro and TanDEM-X), ∼30 m (SRTM and ALOS), and ∼8.1 m (DEMNAS) to rain-on-grid modeling (with HEC-RAS) for flood hydrograph prediction in the Katulampa watershed, Indonesia. Additionally, the influence of the Manning coefficient and infiltration values on the modeling results was also investigated based on three types of land use (forest, agriculture, and urban area). It was shown that ALOS (∼30 m) was the most accurate data among other DEMs to predict flood hydrographs. Interestingly, MERIT-Hydro (∼90 m) significantly outperformed SRTM (∼30 m) and DEMNAS (∼8.1 m), indicating that DEMs with finer-resolution are not necessarily more accurate than coarser-resolution DEMs for rain-on-grid modeling. Furthermore, we observed that the modeling results were strongly influenced by the Manning coefficient and infiltration values, indicating the importance of calibration for such values. Based on 279 combinations carried out, only agriculture was sensitive to both Manning coefficient and infiltration values, while forest and urban were not. The Manning coefficients calibrated for each land use data were similar for different flood events. However, only the infiltration values calibrated for agriculture were distinct for different flood events, thus emphasizing the need for individual calibrations of infiltration values for each event. Our findings are useful to demonstrate how rain-on-grid modeling – with proper DEM data and the calibrated Manning coefficient and infiltration values – can be utilized for accurate flood hydrograph predictions.