H216-06
Multi-Sourced Flood Inventories over the Contiguous United States for Actual and Natural Conditions
Wednesday, 16 December 2020: 20:50
Virtual
Zhijun Huang1, Huan Wu2, Robert F Adler3, Guy Schumann4,5, Jonathan J Gourley6, Albert Kettner7 and Nergui Nanding1, (1)Sun Yat-Sen University, Guangzhou, China, (2)Sun Yat-Sen University, School of Atmospheric Sciences, Guangzhou, China, (3)University of Maryland, College Park, MD, United States, (4)University of Bristol, School of Geographical Sciences, Bristol, United Kingdom, (5)Remote Sensing Solutions, Inc., Pasadena, CA, United States, (6)National Severe Storms Lab, Oklahoma City, OK, United States, (7)University of Colorado Boulder, INSTAAR, Boulder, CO, United States
Abstract:
A reliable flood event inventory that reflects the occurrence and evolution of past floods is important for studies of flood hazards, hydroclimate extremes, and future flood projection. However, currently-available flood inventories are based on single-sourced data such as news articles, reports, gauge- or remote sensing-based observations, and often neglect under-reported or less impactful flood events. Thus, they cannot be considered as complete records of historical flood events in both time and space. Furthermore, despite the high spatial and temporal variability of floods, existing archives typically store flood events only as sparse geographic points, which significantly limits their applicability and, ultimately, their value. Also, few publicly accessible inventories have focused on the reconstruction of historic flood events which would have happened without flood defenses.
In order to provide a comprehensive documentation of historic flood events, we construct two types of multi-sourced flood inventories (MFI) for all river basins across the contiguous United States covering the period 1998-2013. These archives integrate flood information from in-situ observations, remote sensing observations, hydrological model simulations, and five different precipitation products. The first inventory (MFI-Actual) includes all actual floods that occurred in the past in the presence of flood protection infrastructures, while the second, “natural (undefended)” inventory (MFI-Natural) reconstructs the possible “historical” floods without flood protection, which could be more directly influenced by climate variation. In the proposed two inventories, 2,755 and 4,661 flood events were estimated, respectively. MFI-Natural reconstructed 1,597 floods in ungauged basins, and recovered 608 extreme streamflow events in gauged sub-catchments where floods would have happened if there were no flood protection.