A segmentation and classification approach of IKONOS-2 imagery for land cover mapping to assist flood risk and flood damage assessment

Abstract

Various regions in Europe have suffered from severe flooding over the last decennium. Earth observation techniques can contribute toward more accurate flood hazard modelling and they can be used to assess damage to residential properties, infrastructure and agricultural crops. For this study, detailed land cover maps were created by using IKONOS-2 high spatial resolution satellite imagery. The IKONOS-2 image was first divided into segments and the land cover was classified by using spectral, spatial and contextual information with an overall classification accuracy of 74%. In spite of the high spatial resolution of the image, classes such as residential areas and roads are still fairly difficult to identify. The IKONOS-2-derived land cover map was used as input for the flood simulation model LISFLOOD-FP to produce a Manning roughness factor map of inundated areas. This map provides a more accurate spatial distribution of Manning’s roughness factor than maps derived from land cover datasets such as the EU CORINE land cover dataset. CORINE-derived roughness maps provide only averaged, lumped values of roughness factors for each mapping unit and are hence less accurate. Next, a method to produce a property damage map after flooding is presented. The detailed land cover map, water depth estimates resulting from the LISFLOOD-FP model, and known relations between water depth and property damage yielded a map of estimated property damage for the 1995 flood which affected the villages of Itteren and Borgharen in the southern part of The Netherlands. Such a map is useful information for decision makers and insurance companies.