Extended Abstract (616K)5.5
Improved Methodology for Correlating Mesocyclone Detections with Tornadoes
Thomas A. Jones, University of Oklahoma, Norman, OK; and K. M. McGrath and J. T. Snow
The advent of the WSR-88D Doppler radar and its associated vortex detection algorithms have had a significant positive impact in severe weather forecasting. However, the degree of this impact has proven quite difficult to quantify. This work uses mesocyclone detections made by the Mesocyclone Detection Algorithm (MDA) and attempts to associate those with ground truth tornado detections provided by the Storm Prediction Center (SPC) to quantitatively determine the usefulness of mesocyclone detections and their associated attributes. In contrast with previous, more limited, studies, this work will use in excess of two years of Level II data from multiple radar sites processed on a case by case basis. Mesocyclones detected during Level II data processing is combined into a single data set producing a synthetic mesocyclone climatology. These mesocyclones and their attributes where then compared with the SPC tornado data base to determine which mesocyclones were associated with tornadoes. The emphasis of this analysis is focused on attempting to correlate various mesocyclone attributes with the occurrence vs. non-occurrence of tornadoes. To better account for algorithm deficiencies and ground truthing errors, all mesocyclones within predefined spatial and temporal search radii from the tornado track were deemed tornadic. Previous works attempted to correlate a single mesocyclone detection with a specific tornado. Since a realistic mesocyclone is made up of multiple vortices of various strengths, many of which are detected by the MDA, this assumption seems too simplistic for a large scale analysis. Thus, during a single volume scan, it will be possible to have several mesocyclone detections associated with a single tornadic event. The analysis was undertaken using the GIS software ArcView in order to take advantage of its significant spatial analysis tools. Using ArcView, a predefined radius of several kilometers was used to search for possible tornadic mesocyclone detections around the track of a tornado. Temporal analysis was then accomplished to determine which mesocyclones detected near a tornado track truly are associated with that event. Preliminary statistical results using data collected from multiple Southern Plains radars during 2000 revealed that a significant percentage of “strong” mesocyclone detections are indeed associated with tornadoes. Still, many “weak” detections were also associated with tornadoes, just not as great a percentage. It is hoped that further analysis of the mesocyclone detection attributes can help determine with greater skill whether certain mesocyclones are tornadic or not.
Supplementary URL: http://mesocyclone.ou.edu
Session 5, Radar Applications
Tuesday, 13 August 2002, 10:30 AM-12:00 PM
Previous paper Next paper