A Climatology of Significant Winter-Type Weather Events in the Contiguous United States, 1982–94

Abstract

A dataset of all significant winter-type weather events reported in Storm Data within the contiguous United States from 1982 through 1994 has been compiled. Statistical analysis of this dataset is used to determine climatological frequencies of winter weather, including seasonal fluctuations, relative frequencies of different hazards (e.g., heavy snow vs freezing precipitation), duration variations, and size (areal coverage) distributions. Findings are intended to assist in development of operational forecasting issues such as staffing requirements, forecast formats and lead times, and local and regional winter weather preparedness activities. In particular, the database is intended to assist in these issues as they apply to the Storm Prediction Center (SPC), which will begin operational short-term forecasting of hazardous winter weather in fall 1997. From the 13-yr database of 1663 events and 2075 event days, significant winter weather is shown to occur somewhere within the contiguous United States almost daily from mid-November through March. High frequencies of winter weather persist into the springtime climatological period of high severe convection frequencies, suggesting the potential need for extra manpower at the SPC during periods of concurrent multiple hazards. Size distributions indicate a marked prevalence of small-scale events, suggesting the importance of mesoscale processes and the need to develop further understanding thereof before effective short-term forecasting of winter weather can improve. Heavy snow was reported in more than 80% of all events, while significant freezing precipitation was reported in about one-fourth of all events. A second goal is to determine the reliability of Storm Data as an observational database for winter weather forecast verification. Examination of Storm Data over the 13-yr period reveals several shortcomings, including a lack of consistency in the way winter weather events are reported from state to state. Although it is shown that a useful database can be constructed from Storm Data, it also is shown that improvements are needed before such a database can be used routinely for forecast verification. Suggestions are offered for establishing greater consistency in Storm Data, as well as for improving real-time verification and ground-truth information by expanding the role of real-time local storm reports to include winter weather in addition to severe convective reports.