An Analysis of RPM, NCEP and ECMWF Numerical Weather Prediction Forecasts of the 22-23 January 2016 Eastern US Blizzard

The snowstorm of 22-23 January 2016 in the northeastern United States produced one of the largest snowfalls ever recorded in the metropolitan corridor that extends from Washington, DC to New York, NY. The indication for major cyclogenesis with a potential significant east coast snowfall was predicted 12 days and 6 days in advance by the GFS and the ECMWF models, respectively. Early and consistent model forecasts led to high confidence in predicting very heavy snowfall for Virginia, Maryland and southeastern Pennsylvania including the cities of Washington, Baltimore, and Philadelphia. Further, the models consistently indicated that there would be a very strong gradient in snowfall accumulation near the northern extent of the precipitation shield, which was predicted to fall within 100km north or south of the New York City metropolitan region. Variations in the northern extent of snowfall across the operational and ensemble model suites led to significant uncertainty in snowfall predictions for New York City. Just 12 hours before the onset of precipitation in New York City, deterministic model forecasts varied greatly with snowfall forecasts from the GFS, ECMWF, NAM, and WSI's RPM being 12, 11, 26, and 29 inches respectively. Further, NCEP's short-range ensemble forecasts (SREF) had even larger spread with ranges from 1" to over 32" of snow.

Retrospective analysis of global model forecasts and of NCEP's short-range ensemble forecasts system (SREF) revealed that forecasts based on the WRF-ARW dynamical core (including WSI RPM) were much more consistent and accurate at predicting the northern extent of heavy precipitation than forecasts based on other dynamical cores (NAM, GFS, and ECMWF). This presentation will provide both an analysis of the available real-time numerical weather prediction models, as well as an analysis of retrospective runs of the RPM system (using its WRF-ARW core) to reveal what components of the WRF-ARW contributed to the accurate snowfall forecasts, especially in the northern part of this snowstorm.