A Multi-Sourced Flood Inventory in Contiguous United States During TRMM Era

Floods are among the most prominent and catastrophic natural disasters in the world that account for about one-third of all geophysical hazards globally. A reliable flood event inventory that reflects the occurrence and evolution of past floods is an important component for future flood prediction and its change assessment. However, existed flood inventories are based on single-sourced data such as news, reports, gauge- or remote sensing-based observations, thus tend to miss under-reported or less impactful flood events, resulting in incomplete records in both time and space. Despite that floods happen with strong spatial-temporal dynamics and show spatially distributed impacts directly on local social-economics and environment along river networks, traditional flood event archives store flood events only at sparse locations of the globe on geographic point basis, which significantly limits their applicability such as damage and risk assessment. Assessing flood change in a changing global environment and quantitatively attributing flood change to climate warming require different benchmarks of historical flood databases which are both lacking and caught little attention. Besides, none of the publicly accessible flood event inventories includes numerical model results yet, despite their arguable strength in spatial-temporal coverage.

In order to provide comprehensive documentation of floods occurred in past to serve various studies, we compile two 16-year multi-sourced, spatially distributed flood inventories for all river basins across the Contiguous United States from 1998 to 2013, by integrating flood information from in-situ observation, remote sensing and hydrological model simulation driven by three-hourly, 1/8^th degree spatial resolution precipitation information from the Phase 2 of the North American Land Data Assimilation System (NLDAS-2). One emphasizes the recording of all floods that actually occurred in the past reality with human-regulated landscape and flood protection infrastructures (valuable for global environment change studies), while the other emphasizes the reconstruction of the historic flood events caused by climate variation (no flood protection). Statistics from both inventories are also provided with comparison.