The "Millennium" Snowstorm of 30-31 December 2000 brought extremely heavy snow from central New Jersey northeast to western New England, with over two feet reported in portions of northwest New Jersey and southeastern New York. Eta model quantitative precipitation forecasts (QPF's) gave a good indication of the threat in areas that received the heavy snow, but the model also predicted significant snow for the Washington-Baltimore corridor that did not occur. As a result, winter storm warnings calling for significant accumulations in this region were issued, but no snow fell.

This paper examines the successes of the forecast further north, but it focuses on the problems in the Washington area. The model tracked the storm too close to the coast and deepened it too quickly, resulting in the forecast of a major precipitation shield wrapping westward across the mid-Atlantic. A combination of initialization of upper-air features and sea surface temperatures (SST's) appears to have been responsible for the poor forecasts. In particular, the role of the failure of the SST analysis to capture the cold shelf water along the coast leading to an activation of the deep convective scheme in the model will be examined.