Examining a Southward Bias in Lake-Effect Snow Band Forecasts in the Northeast Regional Ensemble

Justin Arnott
NOAA/National Weather Service
Weather Forecast Office
Gaylord, Michigan

ABSTRACT

     In this study, simulations from the northeast regional Weather Research and Forecasting (WRF) model ensemble of two lake-effect snow events from the 2007-2008 cool season are examined. In these simulations of lake-parallel, lake-effect snow bands downwind of Lake Ontario, a systematic southward bias in forecast snow band location is found with the Advanced Research WRF (WRF-ARW) members of the ensemble, consistent with previous research on mesoscale modeling of lake-effect snow and with qualitative forecaster assessments of other events during the 2007-2008 cool season. The bias is found to degrade the usefulness of the northeast regional ensemble.

     A series of sensitivity simulations is performed to help diagnose the cause of the southward bias. These simulations revealed that the WRF-ARW ensemble members underestimated the frictional slowing of wind, downwind of the central and eastern Great Lakes, when compared to a Rapid Update Cycle (RUC) analysis and WRF- Nonhydrostatic Mesoscale Model (WRF-NMM) simulations. The underestimation served to reposition the mesoscale convergence boundaries associated with the lake-effect snow bands farther south. These simulations also clearly indicate that model core rather than model physics and physical parameterizations is the primary source of the erroneous boundary layer flow. A final set of sensitivity simulations suggest that version 3 of the WRF-ARW improves upon this bias with forecast accuracy more comparable to the WRF-NMM shown in one case study.

     Given the results from this study, the feasibility of an operational mesoscale ensemble is discussed. It is shown that while this type of forecast tool shows promise, performing verification studies of each ensemble member is essential to its success.



View the full paper

Back to Digest Index