Oliver Wing (Fathom, United Kingdom); Colin Gleason (University of Massachusetts, Amherst, USA); Paul Bates (University of Bristol, United Kingdom); Jeffrey Neal (University of Bristol, United Kingdom); James Savage (Fathom, United Kingdom)

Evaluating the accuracy of flood inundation models at global scales is a challenging endeavor given the highly transient nature of flood events, the limited availability of ground truth observations (e.g. high-water marks) and the presence of cloud cover preventing the use of optical imagery. Significant progress in determining flood extent using Synthetic Aperture Radar (SAR) and Optical Imagery has been made over the last decade, with the proliferation of automated water detection threshold mapping techniques (e.g. Copernicus Global Flood Monitoring service) and the combination of complimentary satellite constellations (NASA DSWx OPERA product) to reduce revisit times, increasing the chances of capturing the flood event peak. However, the inability to capture the flood event elevation (and therefore depth) in areas where ground truth observations are not available, is a key inhibitor of assessing global flood model accuracy. Given that flood depth is a key determinant of flood vulnerability and the economic impact of a given flood event on infrastructure, more extension observations of flood height are critical for thorough model evaluation.

SWOT, with its novel swath KaRin instrumentation, has provided comprehensive monitoring of both flood extent and surface elevation every 21-days since its launch in December 2022. This offers the chance, for the first time, to evaluate a global flood inundation model's extent and depth simultaneously. Here, we use the NASA SWOT High Rate Raster Product (L2_HR_Raster) to compare with the full set of Return Periods (RP’s) from Fathom’s (www.fathom.global) industry leading Global Flood Map V3 (Wing et al, 2023), for a set of global flood events since the launch of the mission. The L2_HR_Raster is ideally suited for this purpose, with its focus on ease of use and application of higher level quality filtering. This presentation will firstly assess SWOT’s capability in observing flood event extent and elevation, including how to best discern fluvial flood events from other observed water surfaces (such as wet fields and surface water flooding). Secondly, we will discuss how the quality filters can be used to remove spurious observations and provide the most physically comparable set of observations with Fathom’s V3 flood hazard layers. Finally, we will show a comparison between SWOT observations and the flood hazard maps, highlighting the areas of agreement and differences between the two; with a focus on how both datasets can compliment each other.

References

Wing, O. E. J., Bates, P. D., Quinn, N. D., Savage, J. T. S., Uhe, P. F., Cooper, A., et al. (2024). A 30 m global flood inundation model for any climate scenario. Water Resources Research, 60, e2023WR036460. https://doi.org/10.1029/ 2023WR036460