Authors: Louis Rosas and Jim Nelson – Brigham Young University
Title: Global Riverine Flood Inundation Maps for the GEOGLOWS River Forecast System Using ARC and Curve2Flood with Ensembles of DEMs
Presentation Type: Poster Presentation
Abstract: We present ongoing work toward developing global flood inundation maps to be generated in conjunction with the GEOGLOWS River Forecast System (RFS), a global hydrologic model that provides streamflow time series data for 6.8 million river segments from 1940 to near present based on ERA5 and daily forecasts using runoff from the ECMWF medium-range forecast ensemble. RFS uses the TanDEM-X Hydro stream and catchment hydrography derived from proprietary TanDEM-X 12 meter global DEM. The recently developed Automated Rating Curve (ARC) and Curve2Flood tools leverage RFS hydrography and streamflow to estimate riverine flood inundation. ARC is a simple but fast elevation raster-based tool that generates a database of simulated hydraulic characteristics and rating curves using DEMs, hydrography, and land use datasets. At each DEM cell that intersects a river centerline in the hydrography, ARC generates a cross-section, defines bathymetry, and calculates water depth for a range of discharges using Manning’s equation. The values for Manning’s n are determined experimentally using classes from the 10m European Space Agency WorldCover dataset. Curve2Flood uses ARC hydraulic database or rating curves and the corresponding DEM to create flood inundation maps for a specified set of discharges. We performed global parameter estimation of low flow events by comparing to the JRC permanent water bodies dataset. Additionally, we investigated how ensembles of elevation datasets from differing sources affect the model’s skill, both by comparing to the JRC dataset for low flow events and comparing against USGS surveyed and reconstructed higher-flow events across diverse U.S. watersheds. Ensemble members included the 30m FABDEM, the SRTM DEM, the 30m COPDEM, the 30m ASTER DEM, and the 10m USGS 3DEP DEM, among others. These experiments support the long-term vision of producing operational global riverine flood inundation maps powered by GEOGLOWS, and provide a pathway for extending early warning and risk assessment to regions where observational data is sparse or unavailable.