Multivariate analysis of lake-effect snowstorms in western New York

Abstract

The Great Lakes region of the United States is subjected to a wintertime convective phenomenon known as lake-effect snow (LES). These events are capable of producing significant quantities of snow over localized areas by developing elongated bands that tap into heat and moisture exchanges between the warm lake surface and the overlying continental polar air. Several factors are believed to affect the snowfall intensity associated with LES events, including the ice coverage over the lake where the snowbands originate. Improvements in the quality of snowfall forecasts associated with these LES events require an approach that uses a larger number of events than used in previously investigated case studies. The intensity of 91 LES events was assessed using the total volume of snow (snow depth multiplied by snow coverage area) per day in western New York as reported by the National Weather Service in Buffalo, New York, from 1998–2011. These events were cross-referenced against the fetch, capping inversion height, thermodynamic instability, wind speed, and ice coverage over Lake Erie during each event. A multivariate regression revealed that only fetch, inversion height, and ice coverage were statistically significant in determining the intensity of the LES event, with ice coverage having the greatest impact. These parameters accounted for approximately 30% of the overall snowfall volume variability in our LES events. An examination of two events demonstrated that other environmental controls—such as orography and low-level moisture—may have affected the quality of the snowfall predicted by our regression.