2A.2 Constructing a Procedure to Identify Variables Associated with Storm Reports in High-Shear Low-CAPE Severe Environments within the Southeastern United States

Friday, 28 July 2017: 10:45 AM
Constellation E (Hyatt Regency Baltimore)
Dianna M. Francisco, North Carolina State Univ., Raleigh, NC; and L. Xie

High-shear, low-CAPE (HSLC) environments are prevalent in the Southeastern United States, especially during the cool season and overnight, and produce a large proportion of the tornadoes and significant wind reports that occur within this region. Most of the HSLC severe research have focused on synoptic and mesoscale patterns using likely variables identified through physical explanations to predict the observed storm reports. This research uses the opposite approach, finding the combination of variables that are associated with Storm Prediction Center (SPC) official storm reports and then seeking the physical explanations to uncover connections which may otherwise not be explored. This approach allows the use of higher resolution data to identify environmental variables that can distinguish between event cases (i.e., environments with storm reports) and null cases (i.e., environments with storm watches and warnings but no storm reports).

A procedure was constructed using a combination of statistical techniques, including a clustering algorithm, to identify statistically significant differences between two types of environments (i.e., phenomenon occurring in one environment versus the phenomenon not occurring in the other environment) using a binomial response variable. This procedure was designed specifically for large datasets of highly correlated meteorological data and can create a probabilistic model within minutes. North American Regional Reanalysis (NARR) data for HSLC event and null cases were used to test this procedure, and train nowcasting and forecasting models for preliminary results. Higher resolution data was then explored to improve the models. These statistical models and ensembles can reduce false alarm rates (i.e., false positives), an operational forecasting problem for HSLC severe environments which tend to have rapid destabilization within three hours prior to convection. Providing improved HSLC severe weather forecasting tools for our National Weather Service collaborators is the primary goal of this research.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner