749 Persistence of Lightning and Radar Parameters in Consecutive Storm Samples

Thursday, 10 January 2013
Exhibit Hall 3 (Austin Convention Center)
Dustin T. Shea, Univ. of Maryland, College Park, MD; and S. D. Rudlosky, H. E. Fuelberg, and S. Goodman

This study is part of a larger project which used the Warning Decision Support System – Integrated Information (WDSS-II) and Geographic Information System (GIS) software to investigate many severe and non-severe storms in the Mid-Atlantic States during 2007–09. Procedures were developed to examine radar-derived parameters as well as intra-cloud (IC) and cloud-to-ground (CG) lightning characteristics in 460 and 1207 severe and non-severe storms, respectively. The fine temporal resolution (2-min) of the lightning and radar dataset introduces serial correlation, inflating the statistical significance of any results. Autocorrelation functions are used to determine decorrelation times and effective sample sizes to increase confidence in the statistical analyses. Since the decorrelation time also can be seen as a persistence time scale, this provides an opportunity to investigate persistence within storms. Although the absolute value of the decorrelation time can be of limited value, for time series of different variables with common time increments, the relative decorrelation times are useful for comparing system memory. Analysis of autocorrelation functions for individual lightning and radar parameters reveals important differences, especially when storms are grouped according to severity or type of severe weather. Despite this storm-type variability, averaging autocorrelation functions for each storm shows that decorrelation times are nearly always between 3-6 lags (6-12 min).
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner