Applying local research to National Weather Service operations – forecasting heavy mountain snowfalls in Vermont and Northern New York

Heavy snowfalls, greater than 6 inches in 12 hours, that are confined mainly to the Green Mountains of Vermont and the Adirondacks of northern New York have presented a local forecasting challenge to the National Weather Service (NWS) forecasters in Burlington, Vermont (KBTV). NWS KBTV assumed long-fused winter weather warning and zone forecast responsibilities in May 1999 and was faced with its first season of winter forecasting in the winter of 1999-2000.

Two cases were examined, 15-16 November 1999 and 2-3 March 2000, both with similar synoptic patterns that yielded different results. In the November 1999 case, a heavy snowfall occurred in the mountains that was not forecast by the operational numerical weather prediction (NWP). Over a foot of snow fell in many mountain locations and only a trace of snow at KBTV, located in the Champlain Valley. Verification of the NWS KBTV warnings for event yielded a Probability of Detection (POD) of 0.67 and a False Alarm Ratio (FAR) of 0.20 and a warning lead time of 10 hours. At that time, some informal efforts were made to identify upper level and surface patterns that led to the heavy snow fall event. For the March 2000 case, Winter Storm Watches and Warnings were issued for the mountainous areas in anticipation of the event. The event did produce snowfall of 1 to as much as 4 inches, but did not reach warning criteria of more than 6 inches in 12 hours. For this event the POD was 0.00 and FAR was 1.00 with no lead time.

These two events were hypothesized to be largely driven by persistent northwesterly lower tropospheric flow of moisture that was largely orthogonal to the mountains resulting in heavy snowfall over the higher terrain. It was evident by the challenge of accurately forecasting these two events, that local research was needed to develop the knowledge and tools necessary to forecast events of this type in the future.

In 2001, NWS KBTV in association with the University at Albany/SUNY (UA) and other NWS offices, through the Collaborative Science, Technology, and Applied Research Program (CSTAR), began research on this local forecasting challenge. This local research effort was undertaken to identify synoptic-scale and mesoscale patterns associated with this phenomena, determine if advances in NWP could predict these events, and provide forecasters at NWS KBTV techniques and tools to assist in the prediction of these events. The results of this research were applied to a heavy mountain snowfall on 13-14 November 2003 in Vermont and northern New York. This event produced over 12 inches of snowfall in the mountain and only a trace of snow in the Champlain Valley. Using a forecast checklist and Advanced Weather Interactive Processing System (AWIPS) procedures developed for upslope snow events, along with high resolution national and local NWP, NWS KBTV was able to successfully forecast the event several days in advance. A Winter Storm outlook was issued more than 2 days (55 hr) in advance to alert the public of the potential for heavy snow in the mountains. A Winter Storm Watch preceded the event by 43 hours. Preliminary verification of the event yielded a POD of 1.00 and a False Alarm Ratio of 0.10, and a warning Lead Time of nearly 19 hours.

NWS offices performing research in collaboration with Universities can lead to improved predictions of local weather phenomena. Synoptic-scale and mesoscale patterns associated with these phenomena are identified. Advances in local NWP are shown to assist the forecaster in the prediction of these events. Local forecast techniques and tools developed for upslope snow events and used by the NWS KBTV forecasters are illustrated. The transfer of the research results to NWS operational forecasters is also discussed. Warning event statistics for these upslope snow events are shown, since they were useful in measuring the successful transfer of the research and technology to operational weather forecasting.