Integrating Probabilistic Data into Operations for a Deterministic Forecast Process

The National Weather Service (NWS) is on a path towards migrating to probabilistic services, as highlighted by "Ken's 10 Probabilistic IDSS'' initiative. While this process occurs, the NWS must continue to provide its deterministic services, such as the National Digital Forecast Database (NDFD). To create these forecasts, most NWS offices have migrated to a program called ForecastBuilder. ForecastBuilder is a Common Forecast Management Tool that brings together data from the National Blend of Models (NBM) and National Centers, a suite of crowdsourced vetted tools and a common process in building the NDFD. Recently ForecastBuilder underwent a significant change in the creation of precipitation type, utilizing the National Blend of Models (NBM) explicit precipitation type forecasts. This offered up an opportunity to change how forecasts maintain meteorological integrity. For example, rather than the deterministic temperature dictating what precipitation type is allowed, ForecastBuilder looks at the 10th and 90th percentile temperatures from the NBM. This approach would allow a forecast of only rain mentioned with sub-freezing deterministic temperatures. Traditionally this kind of a forecast would seem physically impossible and meteorologically wrong. However, it is a common occurrence when the ensemble data in the NBM is dominated by cold and dry scenarios, thus resulting in the deterministic temperature below freezing, yet a portion of the ensemble data indicates rain in a warmer scenario.

The presentation will dig deeper into how ForecastBuilder incorporates probabilistic temperature data and other kinds of probabilistic data into a largely deterministic forecast process. Unsurprisingly, this mix of probabilistic and deterministic datasets creates communication challenges for NWS forecasters during non-traditional meteorological situations and a discussion of those will also be presented.