Examining high resolution ensemble QPF for case studies of flash flooding

In 2010, extreme flash floods occurred at the Albert Pike campground in Arkansas resulting in the deaths of twenty campers, while the following week there was an urban flash flood in Oklahoma City, OK, which fortunately resulted in no deaths but had millions of dollars of property damage. This study evaluates the skill of quantitative precipitation forecast (QPF) products from the Center for Analysis and Prediction of Storms (CAPS) Storm-Scale Ensemble Forecast (SSEF). The SSEF products are evaluated from three different perspectives: 1) traditional grid-to-grid comparison of the QPFs to NSSL's NMQ/Q2 quantitative precipitation estimate (QPE); 2) traditional hydrologic approach of comparing time series of basin-averaged QPFs to QPEs; and 3) a contemporary approach using a distributed hydrologic model available from the Flooded Locations And Simulated Hydrographs (FLASH) system to simulate return period flows conditioned on forcing from QPFs and QPEs. Results from this study show that all model inputs have poor skill when used in a deterministic fashion due to spatial errors in forecast precipitation. However, this study also shows that the CAPS SSEF can be used to derive precipitation forecasts with better representations of spatial locations of precipitation. Finally, the CAPS SSEF connected to the FLASH system accurately predicted flash flooding for the extreme Oklahoma City event with 12 hours of lead-time.