Warn-On-Forecast in the Watch to Warning Gap

Warn-On-Forecast in the Watch to Warning Gap

Daniel Reese, daniel.reese@noaa.gov

National Weather Service, Topeka KS

1116 NE Strait Avenue

Topeka, KS 66616

The Warn-on-Forecast System (WoFS) is a unique experimental forecast ensemble designed to improve forecast confidence and enhance lead-time during short-term severe thunderstorm events. WoFS is a high resolution model that assimilates real-time radar and satellite data and updates every 30 minutes in order to enhance the accuracy of initial conditions. The 18 member ensemble contains enough diversity in order for high-resolution probabilistic guidance to be created for the occurrence of specific hazards. An initial assessment of WoFS forecasts shows that the ensemble performs best after convective initiation (CI) occurs, and radar data can be assimilated into the model. However, the period before CI is a critical period, typically falling in the watch to warning gap where core partners are searching for information on how the severe weather threat will evolve.

Spring 2024 saw an active period of severe weather across the Plains, including the Topeka county warning area (CWA). Within the Topeka CWA, there were 12 severe weather events from March through June with at least five reports of severe weather. Of these events, seven events had at least some degree of WoFS coverage. For these events, forecasters in the Topeka office made use of the WoFS data that was available. While WoFS data was used during radar operations, it was examined most extensively in the watch-to-warning gap, when time and duty constraints were comparatively less. During this window, WoFS forecasts were used along with mesoanalysis to communicate to core partners and the public where thunderstorms were expected to first develop and what the initial hazards would be.

During the course of the spring, while WoFS did indeed perform best once convection was ongoing, it did have useful and accurate signals prior to thunderstorm development. This signal was more pronounced in some events compared to others, for both CI and storm hazards. With regard to CI, some events displayed a high probability of storm development within a concentrated area. These high probabilities typically verified. Other events had weaker probabilities spread across a larger area, indicating that there was greater uncertainty in CI. Regardless of signal strength, the probabilistic nature of the ensemble was of more use to forecasters than deterministic runs such as the High Resolution Rapid Refresh (HRRR). Furthermore, a subjective assessment of accuracy also rated WoFS forecasts as similar to or slightly better than deterministic high-resolution models such as the HRRR, as well as other ensemble forecasts such as the High Resolution Ensemble Forecast and Short Range Ensemble Forecast. This allowed forecasters to better communicate forecast confidence to partners and the public, and on several instances was critical in giving extra lead-time of tornadoes and other severe thunderstorm hazards.