Predictability issues with the East Coast Winter storm of 14 March 2017: Slushmageddon
Rich Grumm, NOAA/NWS/Forecast Office, State College, PA
Seth Saslo, The Pennsylvania State University
Steven Greybush, The Pennsylvania State University


Abstract
A rapidly developing cyclone brought heavy snow from Pennsylvania into New England on 14 March 2017. A wide area from northeastern Pennsylvania into central New York received from 20 to 40 inches of snow. Initial forecasts implied that the heavy snow would affect most of the major cities along the East Coast from Boston to Washington, DC. However as forecast length decreased, the confidence in snowfall along the coastal plain began to decrease in the numeric guidance.The NCEP GEFS was used to show the forecast issues related to the location of the precipitation shield and the areas where heavy snow was most likely to fall. Predictability horizon diagrams were used to show the relatively high confidence in the quantitative precipitation forecasts verse less confidence in the forecasts of precipitation type. Islip, NY is used to show how high confidence in snow at days 2-4 became low confidence forecasts as the event horizon decreased. The NCAR 3km ensemble is used to show how convective allowing ensembles may provide clues to aid in forecasting precipitation type at shorter ranges. The NCAR ensembles initialized at 0000 and 1200 UTC 13 March clearly showed the uncertain precipitation type along the coast near the frontal boundary. The key findings included the value of high resolution ensembles to better illustrate uncertainty in mesoscale phenomena, the value of these forecasts systems at shorter ranges relative to coarser global ensemble forecasts systems, and the value of predictability horizon diagrams in quickly showing run-to-run uncertainty of multiple forecasts