Flood inundation simulation models are widely used for simulating severe events of flood, generating hazard maps, risk assessment, and to identify flood vulnerable locations. It is important to assess the degree of accuracy of flood model results as these results may be one of the triggering parameters considered in developing flood hazard maps, flood mitigation policies, and land using planning where multi-criteria analysis is approached. In the present study, an algorithm is developed in order to know the performance of flood models by validating it with flood footprints extracted from synthetic aperture radar (SAR) images using multi-segmentation and Otsu's thresholding technique. Evaluation of the performance of the model is based on two best fit criteria called F1 and F2. For this, HEC-RAS model is used for simulating the severe event of flood witnessed in Mahanadi River in Odisha stretching between Tikarpara and Mundali during September 2008.Three simulations were made by considering three different Manning's roughness for river and floodplain. The model gives appreciable results and best fit F1 = 0.85 and F2 = 0.74 was found for Manning's roughness 0.020.
The study showcases development of Algorithms for extraction of flood disaster footprints without any manual intervention which can help disaster respondents to respond to the event in near real time and development of algorithm for evaluation performance of flood models which help flood modellers to evaluate the used flood model performance in validation with satellite based observed flood.