Fine resolution flood forecasts and hindcast of Hurricane Harvey for Harris County, TX using PRIMo

Urban flooding from extreme precipitation has emerged as a major threat to metropolitan regions of the United States. There is a pressing need for flood inundation forecasts capable of meeting emergency response needs for public safety and damage reduction from the household scale upwards to the entire region. Leveraging the computational efficiency of the numerical hydrodynamic model PRIMo (University of California Irvine), we present a flood inundation forecasting system that ingests short-range quantitative precipitation forecast data provided by the National Weather Service, 3 m resolution topographic data available from the 3D Elevation Program of the United States Geologic Survey, and other data to simulate flooding at fine-resolution (3 m) at the county/regional scale, and inform emergency management at household scales. With an application to Metropolitan Houston and 5-day simulations of Hurricane Harvey, the forecast system is shown to capture flood peaks with a mean absolute error (MAE) of 0.86 m, hourly streamflow stage is captured with a median MAE and Nash Sutcliffe Efficiency (NSE) of 0.90 m and 0.41, respectively, and high water marks (HWMs) are captured with a MAE of 0.77 m. The forecast system also achieves hit rates of 90% and 73% predicting 3-1-1- distress calls and FEMA property damage claims, respectively, based on simulated flood depth. Based on fast model speeds made possible by upscaling and parallel computing, these results demonstrate the potential to operationally forecast flood inundation and its impacts in the U.S. for emergency management. This dataset contains a Hurricane Harvey hindcast as well as three forecast scenarios at 3 m linear resolution using different levels of PRIMo's computational grid upscaling technology. The data represents maximum accumulated flood depths over the period August 26 2017 - August 30 2017 as a result of heavy precpitation and storm surge. The spatial extent of the flood hazard layers is approximately 9,000 km2 encompassing Harris County, Texas.Included GeoTIFF files (.tif extension) may be opened directly in any geospatial data visualization software supporting rasterized data including but not limited to ArcGIS and QGIS. The metadata standard includes spatial projection of the flood hazard rasters for overlay with other spatial information. To access the data contained in the zipped Geo-database (.gdb.zip extension), first decompress the archive file (e.g. using 7Zip). Subsequently the geodatabase may be opened directly in most GIS software including ArcGIS and QGIS.Funding provided by: UCI Beal Center for Applied Innovation*Crossref Funder Registry ID: Award Number