Nowcasting severe weather using geostationary satellite data and GLM-based lightning trends of thunderstorm cells

Abstract

<p>In December 2022, the third generation of Meteosat (MTG) is launched to provide geostationary imagery over Europe and other parts of the world. MTG also carries a Lightning Imager (LI) to locate lightning discharges from space. A similar instrument, the Geostationary Lightning Mapper (GLM), operationally monitors lightning on the GOES-R series satellites.</p> <p>This work uses EUMETSAT’s NWCSAF nowcasting software to identify and track thunderstorms from space. GLM lightning history for each storm track can be analyzed in order to identify abrupt changes in the storm’s flash rate, e.g., lightning jumps (LJs) and lightning dives (LDs). Severe weather reports from the NOAA Storm Prediction Center (SPC) archive are used as ground truth to verify the retrieved LJs, to determine leadtimes, and to investigate the potential of those lightning trends to nowcast severe weather. As a first step, three different LJ-algorithms are tested and their parameters are optimized for the GLM records. Under the hypothesis that lightning trends are correlated to severe weather occurrences, contingency tables are created and evaluated with well-known scores, e.g., the Critical Success Index (CSI). Overall, more than 45,000 thunderstorms are analyzed in this work. About 5% of the thunderstorms exhibit a LJ and/or SPC severe weather report. It is found that a modification of the so-called sigma-LJ algorithm and a novel Relative Increase Level (RIL) LJ algorithm show the most promising results when using moderately strict parameters. Leadtimes are highly variable with LJs occurring on average 36 minutes before the SPC report. Further research filters the LJs based on thresholds of the storm’s convective rain rates and overshooting top detection to improve the CSI when correlating LJs and SPC reports.</p>