The Seventh Annual Flash Flood and Intense Rainfall (FFaIR) Experiment. Part I: An Overview of the Subjective Verification of the Experimental Products Used in FFaIR 2019

Meteorological data, model guidance, forecasting methods and the way risk is communicated to the public are constantly evolving. To help assess the influx of new information and processes, the Flash Flood and Intense Rainfall Experiment (FFaIR) Experiment, operated by the Hydrometeorology Testbed (HMT) at the Weather Prediction Center (WPC), has been run annually since 2013. The goal of FFaIR is to analyze the utility of experimental forecasting models, products, and tools in assessing the threat of heavy rainfall and flooding and to determine if they should be transitioned into the National Weather Service’s (NWS) operations.

Following the NWS’s goal to create a Unified Forecast System , a number of the experimental deterministic and ensemble model guidance that were evaluated in the 2019 FFaIR experiment utilized the finite volume cubed-sphere dynamic (FV3) core. For instance, the Environmental Modeling Center provided deterministic runs of their convective allowing models (CAM) that use the FV3 core; the FV3-Nest and the FV3 stand-alone regional model. Likewise, the University of Oklahoma’s Center for Analysis and Prediction of Storms research team provided their CAM ensemble, which comprised of 14 members, all with the FV3 core. These, and other guidance with FV3 cores, were evaluated alongside other deterministic and ensemble model guidance with various dynamical cores, such as the experimental High Resolution Rapid Refresh and the experimental High Resolution Ensemble Forecast version 3.

A mix of subjective and objective verification methods were used to evaluate experimental guidance pertaining to heavy rainfall and flooding forecasts. This presentation focuses on the results from the subjective verification portion of the experiment. Subjective verification was performed by inviting forecasters and researchers in shifts of a week over four weeks in June and July. During this time, participants used the products in a real-time setting to produce four experimental forecasts daily, while providing feedback about the tools. They also compared guidance to observations, evaluating their utility both numerically and in the form of a group discussion.