Determing key predictors for NCAR's Convective Auto-Nowcast system: Adjusting for different types of convective senarios

The NCAR Auto-Nowcast system is an expert system for forecasting thunderstorm initiation, growth, and decay. This system makes use of numerous data fields which are transformed into predictor fields through the use of membership functions. These predictor fields are then combined using prescribed weights to produce a final forecast. The focus of this paper will be to discuss how recent analyses which suggest that the ANC forecast logic should be modified to account for different types of convective senarios can be used to improve forecast performance.

Data from the 2004 season will be used to show the different attributes of the predictor fields for different types of convective organization. These results will be used to develop updated forecast logic based on the convective senario that is expected during that time period. This updated logic will be applied to several independent cases from the 2005 season and the resulting forecasts will be compared to the results using the original forecast logic.