Case Study: Rapid Urban Inundation Forecasting Technique Based on Quantitative Precipitation Forecast for Houston and Harris County Flood Control District

Abstract

This study explored the operational feasibility of an empirical approach to flood inundation forecasting using quantitative precipitation forecasting (QPF) from high-resolution numerical weather prediction models for the city of Houston and the Harris County Flood Control District (HCFCD). A proposed rapid-refresh technique for generating forecasted flood inundation maps was tested, wherein the processing time was limited by the time required for generating high-resolution QPF. Using the dense gauge network operated by the HCFCD, hurricane type storms were found to be generally more challenging for quantitative precipitation forecasting than the less intense and more frequent winter storm events. The investigation also indicated that forecasting inundation is possible based on rainfall forecasts using predeveloped rating curves between the observed rainfall and the expected increase in water level. The median of the relative root mean square error (RMSE) in percentage and the correlation of the forecasted water level at gauge locations are consistently below 10% and higher than 0.7, respectively, for up to a four-day lead time. In terms of spatial detection of flooded (non-flooded) areas, this technique yielded qualitative consistency during peak inundation episodes when the QPF was skillful.