10B.2
An investigation of alternative verification schema for the National Weather Service: Results of analysis of March 2001 storms in North Florida and Southeast Georgia
Dawn Hayes, Edward Waters College, Jacksonville, FL; and P. T. Welsh
The month of March 2001 brought three severe weather outbreaks to North Florida and South Georgia. These events are reconstructed from the storm database and alternate methods for computing standard severe weather verification statistics are compared. The comparative methods are to compute the same statistical measures counting each county as a separate warning, which is the standard for the National Weather Service, or to compare the statistics using the Advanced Weather Information Processing System (AWIPS) graphical coordinates and locations specifically included within each warning message as a singular event. This latter method has only recently become possible due to the implementation of AWIPS.
Each warning plot was compared to the location(s) of reported damage (or none) during the period of the warning. Plotting error is considered to be on the order of one mile, since that is the minimal recorded location within the storm database, as well as the positional accuracy of the graphical coordinates of the warning polygon vertices. Damage locations within this plotting error of the polygon perimeter were a priori assigned as hits and considered to verify the warning.
Statistics on the probability of detection of the severe weather, false alarm rate, and critical success index are calculated and compared for the two methodologies. The utility of the compared methods is discussed and analyzed for future verification studies of severe weather.
Session 10B, warnings, dissemination and verification
Wednesday, 14 August 2002, 1:30 PM-3:00 PM
Previous paper Next paper