Combining moored observations and SAR images in validating compound flood models

Estuarine compound flooding is a major concern for many communities worldwide. Terrestrial and space observations can provide high-resolution data to assess the accuracy of flood models by capturing the temporal and spatial scales of hazards. In this study, we applied in situ measurements and Sentinel-1 SAR data to calibrate and validate the LISFLOOD-FP hydrodynamic model in simulating compound flood events. The model parameterization of bottom roughness was calibrated using a land cover map, with in-estuary water levels during two flooding events compared to observations from moored pressure sensors. Furthermore, a raw Sentinel-1 SAR image that captured the flooding extent was pre-processed to remove noise and correct distortions. Image classification techniques were applied to isolate surface water from dry land in the SAR image, with the un-supervised classifier showing better results than the supervised. The processed image was compared with the modelled flood extent using eight comparative scenarios, using overlapping indices. Matching was good in the upper estuary, where compound flooding was strongest, but less accurate in other locations, likely due to unresolved drainage channels on the digital elevation model (DEM). The study has shown that processed SAR images combined with traditional in situ measurements can provide robust spatiotemporal validation for flood inundation models.