In collaboration with the National Severe Storms Laboratory (NSSL) and the Earth System Research Laboratory (ESRL), the Hydrometeorological Testbed at the Weather Prediction Center (HMT-WPC) hosted the first annual Flash Flood and Intense Rainfall Experiment (FFaIR). The experiment focused on exploring the challenges associated with short-term Quantitative Precipitation Forecasting (QPF) and flash flood forecasting during the warm season.
During the experiment, which was held from 8-26 July 2013, 26 participants (including 8 participating remotely) issued a series of short-term QPF and flash flood forecasts. These forecasts included a 12-hr probability of precipitation > 1", a 6-hr probability of flash flooding, and a 12-hr overnight probability of flash flooding. Participants had access to a variety of high resolution experimental model guidance including ESRL's High Resolution Rapid Refresh (HRRR) and Experimental Regional Ensemble Forecasting System (ExREF). In addition to issuing experimental forecasts, participants explored several flash flood diagnostics including Flooded Locations and Simulated Hydrographs (FLASH), a high resolution rapidly updating (1 km, 5 min) distributed hydrologic model developed by NSSL.
One of the new forecasting techniques explored during the experiment was the use of neighborhood probabilities. Neighborhood probabilities search for the occurrence of an event (ex: QPF>1") within a defined radius (ex: 20 or 40 km), which helps account for some of the spatial errors in high-resolution model forecasts. During FFaIR, participants explored the utility of both the neighborhood probability of QPF>1" and the neighborhood probability of QPF exceeding flash flood guidance (QPF>FFG) and found that the neighborhood probabilities often provided a more realistic depiction of the threat areas than standard point probabilities.
FFaIR highlighted the strengths and limitations of the current state of flash flood forecasting as participants saw firsthand that meteorological confidence does not always equate to hydrologic confidence given the uncertainties in the hydrologic response. While models will not correctly predict every event every time, the experiment revealed that high resolution model guidance can provide valuable information about the potential for a flash flood event before precipitation develops on radar. Participants also found that QPF>FFG probabilities provided valuable forecast guidance despite the limitations of flash flood guidance.
The 2013 Flash Flood and Intense Rainfall Experiment provided a unique opportunity to bring the operational forecasting, research, and model development communities together to explore the challenge of flash flood forecasting. The experiment provided valuable feedback about the utility of the experimental models and forecasting tools to WPC, NSSL, and ESRL, and HMT-WPC continues to work with both NSSL and ESRL to optimize this guidance for use in WPC operations. In addition, HMT-WPC is working with HMT-SEPS to further explore a flash flood event in western North Carolina that occurred at the end of the experiment period.