Forecasting of winter precipitation remains a challenging problem. Diagnosing the onset and duration of precipitation, as well as precipitation type and amount, is complicated by uncertainty in cyclone tracks and in the evolving local thermodynamic environment. Models utilizing high vertical and horizontal resolution offer greater detail and realistic precipitation structures. However, uncertainty in model physical parameterizations and initial and boundary conditions complicate the forecast problem. In cool-season forecasting, the result may be incorrect placement and timing of the mesoscale regions of mixed precipitation and of heavy snowfall and blizzard conditions.

Researchers at the Department of Atmospheric Sciences at the University of Illinois and forecasters at the National Weather Service Office in Lincoln, IL have undertaken a collaborative research effort to improve and better understand Midwest cool-season precipitation forecasting. New diagnostic products are being developed to allow thorough interrogation of model output for prediction of winter precipitation onset, duration, type and amount. In support of this effort, an ensemble of high-resolution numerical models (MM5, WRF, Workstation Eta and NCEP Regional Spectral Model) is being run daily for a 10-state Midwest region.

These tools are applied to a significant winter storm which struck parts of the Plains, Midwest and Great Lakes from January 29-31, 2002. 6-18" of snow and up to 3" of ice affected these areas, leaving hundreds of thousands of homes and businesses without power (CIMSS, NWS reports, 2002). Regional model simulations made during the event show the evolution of a narrow region of mixed precipitation, with significant snowfall also indicated. Examples of individual and ensemble prediction products from the suite of model simulations, the verification of these products, and operational forecasts will be discussed.