Flooding at the Fringe: A Reduced-physics Model for Assessing Compound Flooding from Pluvial, Fluvial, and Coastal Hazards

Developing scalable multi-hazard flood modeling approaches that balance efficiency and accuracy are critical to flood risk management in coastal watersheds. In this study, we evaluate the performance of the reduced-physics, hydrodynamic model SFINCS to simulate multiple drivers of flooding from Hurricane Florence (2018) in the Carolinas. SFINCS is a two-dimensional, raster-based flood inundation model that is constructed using scripting workflows and a data catalog setup by the user. We use the SFINCS-LIE (e.g., Local Inertial Equations) formulation with wind drag, spatially varying infiltration, and the subgrid feature which enables water levels to be updated based on high-resolution, topobathymetric data (e.g., <10m) while fluxes are computed at a coarser grid scale (e.g., 100m). Water levels from the storm surge model ADCIRC and USGS river gages are interpolated to the boundary cells along the coast and inland, respectively. Hourly gridded radar-rainfall and wind fields are applied directly to the model. We undertake a comprehensive validation of the model from inland to the coast, comparing outputs against USGS and NOAA gages, high-water marks, windshield surveys, and NFIP policy and claims data. Preliminary results demonstrate that the model accurately predicts flooding from pluvial, fluvial, and coastal drivers. SFINCS is scalable and can be quickly setup for various domain sizes (e.g., community, regional) and grid resolutions making it useful for rapidly generating flood hazard information for tropical cyclones. Our results show how important it is to consider model performance at the fringe of the floodplain where pluvial processes can dominate.