Numerical model forecast sensitivity to microphysical parameterization for lake effect snow

The sensitivity of numerical forecasts of lake effect snow to the choice of microphysical parameterization is tested using eight different schemes available in the Advanced Weather Research and Forecasting model. The schemes include WSM6, Goddard, Ferrier, Thompson, SBU-Lin, WDM6, Milbrandt-Yau, and Morrison. The WSM6, WDM6, Ferrier, and Milbrandt-Yau forecasts are more convective in appearance with narrow snow bands that wobble back and forth near the coastline. The 24-h accumulated precipitation maxima in these forecasts is comparatively high (up to 94 mm). In the remaining forecasts, the snow bands are broader and 24-h maxima are more modest (ranging from 43 to 50 mm). There is no obvious distinction between single- and double-moment schemes. Rather, the apparent cause for differences rests in the comparatively high production of graupel (or heavily rimed particles) by the WSM6, WDM6, Ferrier, and Milbrandt-Yau schemes. The faster fall speeds of these particles allows for decreased downstream advection and higher local accumulations. Differences in the amount of sublimation-cooling also appear to be important for generating stronger downdrafts and the more convective appearance in these forecasts.