Towards a global flood frequency map from SAR data

The main objective of this study is to generate inundation maps of past flood events based on an archive of Synthetic Aperture Radar (SAR) data. Within a hierarchical image splitting framework, the flood mapping algorithm uses a histogram thresholding operation and a region growing process to delineate the flood extent. This algorithm is applied to an archive of SAR images in order to generate a flood frequency map. We define the flood frequency of a specific area as the ratio between the number of images where the area was detected as flooded and the total number of images within the employed data collection. SAR water-like ambiguities (e.g. urban areas, crops or shadow regions) are filtered out using auxiliary data sources such as the Height Above Nearest Drainage (HAND) index or land cover maps. The proposed methodology is applied to an ENVISAT ASAR image archive over the UK area. Results presented in this article demonstrate the effectiveness of this methodology.