Optimizing SAR-based Flood Extent Assimilation for Improved Hydraulic Flood Inundation Forecasts

Accurate forecasts of flood inundation are vital to effective flood rescue, response, and resource allocation. However, uncertainty in inputs, boundaries, and parameters necessitate the use of independent observations to constrain flood predictions. Radar remote sensing allows the synoptic and systematic coverage of flooded areas and is thus a valuable resource for more accurate flood forecasts when combined with models. Accordingly, this thesis first improved the satellite-based probabilistic flood extent observation, and then designed a novel likelihood function to integrate such observations with flood model estimates yielding improved flood inundation forecasts.