. . . In synoptically-driven winter storms, quantitative snow predictions can be made with reasonable ease and acceptable accuracy (e.g., Chaston, 1989). Model output generally does a satisfactory job of handling such systems (e.g., Sanders, 1987;Smith and Mullen, 1993; or Grumm, 1993) -- at least in regions where mesoscale factors provide little or no effect. However, the mesoscale can significantly alter standard QPF totals. Lake effect snow (LES) events are notoriously capricious, bringing local accumulations an order-of-magnitude greater than that which occurs on the broader scale. These events focus snowfall into narrow bands such that a region receiving 6-8" of snow per hour -- or more -- can be adjacent to an area with light snow, or sometimes even clear skies (Niziol, 1987).
. . . In LES events, there can be a significant risk of overwarning. In that case, not only are credibility issues at stake, but there can be important, negative economic impacts as well. For example, large payrolls will be wasted if dozens of road clearing crews sit parked in their plows alongside highways, waiting for a snowstorm that does not materialize. The general public, too, can suffer dollar loss, as well as deep frustration, through such false alarms.
. . . The following discussion centers on an LES event which took place in northern Michigan over a three-day period from 08 Dec through the late evening of 11 Dec 1995. It developed within a period of general snowfall for the upper midwest during which sites not affected by the lakes received 3" - 7" of snow. The discussion in this paper is focused on Sault Saint Marie, which received a total of 63" of snow over the 3-day period. For much of the event, visible and 10.7 um infrared satellite imagery proved helpful in identifying locations and intensity of lake effect snow bands. 3.9 um data, combined with in-situ soundings and simple model results, provided new insight into the nature of LES convection.