4.4
The performance analysis of total lightning in NCAR's Auto-Nowcaster

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 31 January 2006: 9:30 AM
The performance analysis of total lightning in NCAR's Auto-Nowcaster
A307 (Georgia World Congress Center)
Nicholas L. Wilson, Univ. of Oklahoma, Norman, OK; and D. W. Breed, C. K. Mueller, T. R. Saxen, and N. W. S. Demetriades

Presentation PDF (2.3 MB)

The NCAR Auto-Nowcaster (ANC) is an interactive software package that integrates multiple meteorological datasets in real-time and uses fuzzy logic to combine them and create 60-minute forecasts of thunderstorm initiation, growth and decay every 5 minutes. The ANC has run operationally within the Ft. Worth National Weather Service forecast office since March 2005. Its thunderstorm growth/decay component uses Thunderstorm, Identification, Tracking, Analysis and Nowcasting (TITAN) to identify, track and provide 60-minute forecasts of ongoing convection. Since its development, the ANC has only included TITAN-derived radar reflectivity cell attributes for the growth/decay forecasts. We now introduce total (cloud plus cloud-to-ground) lightning into the Ft. Worth ANC to complement the growth/decay fuzzy logic scheme.

Flash Extent Density (FED) is used to represent the total lightning from Vaisala's Lightning Detection and Ranging (LDAR II) VHF time-of-arrival network centered at Dallas/Ft. Worth International Airport. The FED employs a “branched segment” reconstruction of the cloud flashes using temporal and spatial constraints. This method helps to normalize the effect of decreasing VHF source detection efficiency with range because flash detection efficiency decreases at a much slower rate with increasing distance from the center of the LDAR II network. FED area and intensity thresholds are applied within TITAN to identify “total lightning” cells. Total lightning cell attributes, such as maximum FED value and normalized area growth rate, are used to create fuzzy logic membership functions for automated forecasts produced by the ANC. The quantitative performance of ANC 60-minute growth/decay forecasts with and without TITAN-derived total lightning attributes will be discussed using validation statistics collected on six different convective weather events during spring-summer 2005 in the Dallas-Ft. Worth area.