Short Range Probabilistic Winter Precipitation Forecasts at the Weather Prediction Center - Verification and Myth-Busting!

Probabilistic Winter Precipitation Forecasts (PWPF) have been produced at the Weather Prediction Center (WPC) since 2010. These are “human-influenced” but automatically-generated gridded datasets that contain probability of exceedence values as well as percentile amounts of both snow and ice. It is based on the human forecaster's deterministic forecast but uses a multi-model and ensemble membership to form the variance around the “most likely” solution. Verification from recent winter seasons has shown that the human-produced deterministic forecast guidance shows more skill than any single deterministic model, ensemble mean, or automated multi-model/ensemble guidance. The probabilistic guidance has shown to be statistically reliable. Modifications have been made regarding the spatial and temporal resolution as well as the ensemble membership between this season (2014-15) and last winter (2013-14). Effects from these changes as well as verification from this past winter season will be presented. In addition, forecasts from the New England Blizzard of 26-27 January, 2015 will be discussed.

With the rise in the popularity of these WPC graphics, there is an increased risk of confusion and misinterpretation of this information. Common recurring questions include: Why isn't the 50th percentile the most likely scenario? How much influence does the WPC forecaster have in forming the probabilities? Is it fair to verify a singular probabilistic forecast? Myths about what the PWPF is and isn't will be addressed.