DEM Generation from Multi-View Satellite Images in Sub-Sahel Region

Abstract. Floods are causing a significant loss of human lives and valuable resources in West Africa. In particular, Niger and Burkina Faso were highly affected areas in past years. In order to predict flood, an accurate Digital elevation model (DEM) is required for flood mapping. At the studied area in Niger, up to this date, the LiDAR DEMs are scarcely available, and the only available DEMs are global DEMs like global SRTM DEMs with a resolution of 10m. These global DEMs are not accurate enough to be used for flood mapping. So, in this context, this study investigates the potential of multidate, multi-view stereo pairs PlanetScope images for the generation of DEM. Three DEMs were generated from images with slightly different view angles to see the effect of view angles of images on 3D modelling. One of the DEM generated by PlanetScope images was compared with DEM generated by high-resolution drone imagery and shows the normalized Median of Absolute Deviation (NMAD) of the elevation differences of 10m. Results show that planetScope images are useful assets for generating multiple DEMs due to their high temporal resolution. Such DEMs could be extremely useful for studying dynamic phenomena or monitoring disaster events like floods.