Evaluation of Convection-Allowing Ensemble Forecasts for Central U.S. Banded Snowfall Events

Banded snowfall events are associated with snowfall rates as high as 8 cm h-1, accumulations of snow exceeding 30 cm, and gradients in snowfall as large as 1 cm km-1. Because of these large gradients, large errors in forecast snowfall amounts can occur. Convection-allowing models offer the potential to improve the prediction of mesoscale bands of snow, as models with coarser grid spacing lack the ability to simulate the strength and width of the vertical circulation responsible for snowfall production. Ensemble numerical weather prediction forecasts at grid spacings of 3 km or less have not yet been employed operationally. In this study, forecasts from NCAR's real-time 3 km 10-member ensemble are evaluated for two long-duration Central U.S. snowfall events that featured very prominent banding. Comparisons with currently operational ensemble and deterministic forecast systems are made, highlighting the improved capabilities of next-generation modeling systems for this phenomenon.