Research: Improving probabilistic forecasts of extreme rainfall through intelligent processing of high-resolution ensemble predictions

This research is sponsored by NOAA's Joint Technology Transfer Initiative, and builds upon our analysis of extreme rainfall events using historical "return periods" for rainfall amounts. In Stevenson and Schumacher (2014), we developed a database of events in which the rainfall amounts exceeded the "2% annual exceedence probability," more commonly referred to as a "50-year rain event" for various accumulation periods, and Herman and Schumacher (2016) extended this to analyze numerical model forecasts. In this work, we are using the same recurrence interval framework, along with NOAA's global ensemble reforecast dataset and machine-learning algorithms, to produce probabilistic forecasts of extreme rainfall for days 2 and 3 (i.e., 36-60 and 60-84 hour forecast periods). During the Flash Flood and Intense Rainfall (FFaIR) experiment at the Hydrometeorology Testbed and WPC in June-July 2017, this model was used as a "first guess'' for forecasters in their creation of excessive rainfall outlooks (indicating the likelihood of flooding rains), and we continue to work with WPC to potentially transition this into forecast operations. The experimental forecasts can be found here: . This work was featured on 9News in Denver in July 2017.