Enhancing Digital Elevation Model Accuracy for Flood Modelling – A Case Study of the Ciberes River in Cirebon Indonesia

Abstract

Topographic conditions represented by the Digital Elevation Model (DEM) are essential in flood inundation models. The DEM which is categorised as a Digital Surface Model (DSM) stores the height information, be-sides the ground and non-ground elevation required for preprocessing before being employed in hydrologic applications, particularly in relation to flood modelling by removing non-ground elevation along the flood plain and river channels. The improvement in the accuracy of flood inundation modelling is crucial in reduc-ing the impact of flood disasters. This study aims to compare the accuracy level of the DEM based on Ter-raSAR-X data with the filtering process using slope-based filtering and combining the cross-sectional river profile from the field measurement to the filtered DEM. The result confirms that the accuracy of the DEM product is improved via filtering to remove non-ground elevations and a significant improvement in accuracy by means of fused river profile information to filtered DEM. The results of adding river information to the DEM could provide a representation closer to the cross-sectional profile of the river based on field measure-ments within the accuracy level Mean Absolute Error 2.51 m, 2.72 m and 1.91 m in the left overbank, right overbank and centre of the river, respectively. The performance results of the 2-dimensional flood hydrody-namic modelling using HEC-RAS derived from the DEM before filtering, after filtering, and the addition of river information show increasing accuracy in flood depth at each stage of the DEM processing. There is an improvement in accuracy in flood depth of approximately 11.67% using the filtered DEM, besides an in-crease in accuracy in flood depth by 24.98% utilising the filtered DEM with added river channel information.