Wednesday, 31 January 2024: 2:30 PM
339 (The Baltimore Convention Center)
High-resolution historical flood inundation reanalysis can be used to assess regional flood risks and devise long-term mitigation and resilience strategies. Accurate representation of historical inundation demands solving the full 2D shallow water equations at locally relevant spatial resolutions. This study introduces an ongoing effort using TRITON (https://triton.ornl.gov/), a GPU-accelerated 2D hydrodynamic model, for the creation of a conterminous United States (CONUS) wide flood inundation reanalysis dataset. The hydrologic inputs for TRITON modeling are derived from historical runoff and streamflow data. These data are generated by a well-calibrated VIC-RAPID hydrologic model, which is driven by the National Center for Environmental Prediction Stage IV Hourly Quantitative Precipitation Estimates spanning the years 2002–2018. The baseline topography is the 10m resolution National Elevation Dataset that also serves as the computational grids of TRITON. The long-term climatic mean runoff and streamflow are used to calculate the steady-state water depth and velocity as the initial conditions for event-based inundation simulation. The inundation reanalysis is developed across the CONUS for each of the 4-digit Hydrologic Unit Code (HUC4) sub-regions. The accuracy of the simulated flood inundation is evaluated using various static benchmarks like high-water marks, remote sensing-derived maps, and available high-fidelity flood simulation maps. Other data such as daily/hourly gauge observations serve as a reference to assess the temporal evolution of simulated flood inundation. Finally, we share insights into the challenges of large-scale and high-resolution hydrodynamic inundation modeling and discuss necessary efforts to enhance flood representation and potential for real-time forecasting.

