Towards the optimal fusion of high-resolution Digital Elevation Models for detailed urban flood assessment

Abstract

Newly available, more detailed and accurate elevation data sets, such as Digital Elevation Models (DEMs) generated on the basis of imagery from terrestrial LiDAR (Light Detection and Ranging) systems or Unmanned Aerial Vehicles (UAVs), can be used to improve flood-model input data and consequently increase the accuracy of the flood modelling results. This paper presents the first application of the MBlend merging method and assesses the impact of combining different DEMs on flood modelling results. It was demonstrated that different raster merging methods can have different and substantial impacts on these results. In addition to the influence associated with the method used to merge the original DEMs, the magnitude of the impact also depends on (i) the systematic horizontal and vertical differences of the DEMs, and (ii) the orientation between the DEM boundary and the terrain slope. The greater water depth and flow velocity differences between the flood modelling results obtained using the reference DEM and the merged DEMs ranged from −9.845 to 0.002 m, and from 0.003 to 0.024 m s−1 respectively; these differences can have a significant impact on flood hazard estimates. In most of the cases investigated in this study, the differences from the reference DEM results were smaller for the MBlend method than for the results of the two conventional methods. This study highlighted the importance of DEM merging when conducting flood modelling and provided hints on the best DEM merging methods to use.