The snowfall amounts are considerably higher than predicted, and the timing of this period of snow was not consistent with the traditional rules of thumb used to predict the onset of precipitation in panhandle-hook type storms. The winter storm warning issued at 22Z on March 12 by the NWSFO located in Dousman, Wisconsin, called for a storm total of 6-12 inches of snow for the northern half of the state, to begin in the morning of March 13, and the NWSFO storm report for this event noted that "forecasters seem to have a difficult time with these [convective] situations, as history has shown, and snowfall amounts were well under-forecasted" (Haase, 1997).
Application of the traditional rules for forecasting snowfall assumes that the precipitation will be associated exclusively with the dominant low pressure system. In this case, the initial 12-18 hours of snowfall were largely associated with an upper level feature that was independent of the sea level pressure minimum.
The ingredients maps from the 6-hour forecast of the NCEP-ETA model valid at 6Z on March 13, 1997 indicate that there was ample moisture (RH > 80%) at 700-750 hPa and 600-650 hPa in southern Wisconsin. However, the QG forcing and instability ingredients on these standard isobaric levels at this same time do not indicate a significant potential for heavy snow before 12Z on March 13. The mainly weak to moderate QG forcing over Wisconsin and the absence of instability suggested that any precipitation would probably be of light intensity.
A more complete picture of the ingredient distributions at 6Z on March 13 was obtained through a cross-sectional analysis. The ingredients cross-section, drawn from south of LaCrosse to south of Sheboygan in Wisconsin, reveals features that were not apparent on the isobaric ingredients maps and provides significant clues to the potential for high snowfall rates. This cross-section shows that a layer of conditional instability (PVes < -0.15 PVU) existed in the cross-section between 600-500 hPa at 6Z on March 13. With sufficient moisture at these levels (RH = 90-100% at 600 hPa) and weak to moderate forcing between 500-700 hPa, this instability could be realized and heavy precipitation is possible. Furthermore, a consideration of the efficiency ingredient shows that the temperature in the 600-550 hPa layer (-12 C to -18 C) enabled the maximum depositional growth and condensation rates of ice crystals to occur, providing more support for a possible heavy snowfall event.
For this case, much of the information indicating the potential for heavy snow could have also been obtained by analyzing an isobaric ingredients map in the non-standard layer, 550-600 hPa. Such an analysis reveals a QPV maximum over western Wisconsin at 6Z on March 13, corresponding to a region of weak forcing and negative PVes. The 550-600 hPa QPV maximum coincided with ample moisture in west-central Wisconsin, indicating the potential for a high precipitation rate and convection at this time. Although much of the state was characterized by conditional instability in the 550-600 hPa layer, the heaviest snow occurred in a narrow band in southern Wisconsin where the relative humidity was greater than 80% throughout the 850-550 hPa layer.
The preceding application of the ingredients-based methodology to the diagnosis of the initial snow band in the March 1997 storm revealed clues for the potential for moderate to heavy snow in Wisconsin prior to 12Z on March 13. This analysis indicates that the NCEP-ETA model data available to forecasters prior to the storm contained information that could have alerted them to the possibility of the early snow event. Furthermore, the analysis provided another example of the importance of inspecting ingredients cross-sections in conjunction with the standard pressure level ingredients maps.