Evaluating the performance of a hydrodynamic model using SAR images

The evaluation of the performance of flood models is a difficult task, mainly due to the lack of in-situ data. In this context, the contribution of Remote Sensing is crucial. Flood simulation, incorporating Remote Sensing techniques, constitutes a subject which has been widely discussed (e.g., Yan et al. 2013; Patro et al. 2009; Di Baldassarre et al. 2011) demonstrating the high potential of using satellite data to evaluate the flood model results. Specifically, the ability of microwave irradiation to penetrate cloud cover, and thus, effectively detect floods has rendered the use of radar imagery decisive in flood mapping (Horritt 2000). The present study employs satellite Synthetic Aperture Radar (SAR) images to test the performance of a hydrodynamic model. The methodology was applied in a real-world case study, namely a 25-km reach of Pineios River, Thessaly, Greece. Two Sentinel-1 SAR images depicting the area of interest before and at the last day of a specific precipitation event were selected. After appropriate processing of the SAR images, the extent of the flooded area was extracted. The flood extent was also determined by simulating the flood flow on a selected reach of the main river, using the well-known HEC-RAS software. Inundation extents as derived from both processes were compared based on an index indicating the measure of fit between the results predicted by the model and the satellite data (Horritt 2000).