The forecasting process is a complex one that is not often completely understood by the meteorological community. Evident in this process is the increasing importance of rapidly developing numerical weather prediction tools to generate accurate forecasts. An evaluation of the use of these tools by forecasters at the Storm Prediction Center is presented here. The evaluation focused on a period several days in advance of the April 17-18, 2008 severe thunderstorm event. As the event approached, forecasters relied heavily on forecast guidance provided by the National Center for Environmental Prediction, the National Severe Storms Laboratory, and the European Center for Medium-Range Weather Forecasts. Several parameters typically used in severe thunderstorm forecasts (e.g. thermodynamic instability, precipitation forecasts, wind, vertical shear, and lift) were comparatively examined using the guidance provided by these centers to quantify the potential for severe thunderstorm activity during the event.

It was found that forecasters rarely depended on a single deterministic solution to generate a forecast. Instead, they used multiple solutions to develop a conceptual model of how the atmosphere would evolve during the forecast period and the resultant threat for severe convective activity. It was also found that their effective use of forecast guidance was heavily dependent on recognition of systematic model errors of atmospheric features that play key roles in model forecasts of atmospheric convection (e.g., dryline or frontal position, convective feedback, unrealistic moistening, etc.). Effective use of models appeared to be hampered somewhat by the rapid evolution of parameterization schemes and the inability of forecasters to “keep up” with the changes.