Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Atmospheric and Earth Science

Committee Chair

Phillip Bitzer

Committee Member

Larry Carey

Committee Member

John Mecikalski

Committee Member

William Koshak

Committee Member

Chris Schultz


Lightning--Remote sensing, Atmospheric electricity, Hurricane Harvey 2017


The Geostationary Lightning Mapper (GLM) detects total lightning continuously from space without distinguishing intra-cloud (IC) from cloud-to-ground (CG) lightning. This research focuses on improving our understanding of the differences between CG and IC lightning as detected by GLM with the potential for classification of these flashes. First, we explore optical differences in flash characteristics for CG and IC lightning by implementing a Random Forests (RF) model to predict flash type, with Earth Networks Total Lightning Network (ENTLN) acting as truth. The results show the most important flash characteristic for distinguishing flash type is the maximum-group-area while other important features include time-of-day, elongation, and propagation. Skill scores showcase the model’s ability to distinguish flash type with 81% POD, 71% PC, 36% FARate, 36% FARatio, and 56% CSI. These scores improve further when the study area is limited to CONUS. Separating the model further into different regions of CONUS has mixed results, with improvements in the Southeast and Great Plains, while other regions such as western CONUS worsen. Overall, the model performs best over the Southeast, likely due to better detection efficiency of GLM and ENTLN in this region. Seasonally, we find that the RF model performs best in winter and worst in early fall. As an application of our RF model, the cloud flash fraction (CFF) is analyzed both in-depth for hurricane Harvey, and for the 2020 hurricane season. We hypothesize that eyewall CFF is correlated to periods of intensification due to the nature of convection and charge regions in the eyewall. Overall, CFF shows statistically significant positive correlation to wind-speed, both on the individual level (0.4-0.8) and overall (0.2). Eyewall brightness temperatures also show statistically significant negative correlation to CFF, implying stronger convection corresponds to larger IC lightning proportion. For eyewall GLM flash characteristics, groups-per-flash, energy, and footprint all have statistically significant correlation to wind-speed for the future 6-hour change, implying potential use in prediction. In general, eyewall lightning variables have stronger correlations to wind-speed than the rainband. Ultimately, these results can aid in future climatological analysis of flash type, as well as future studies of hurricane intensification and weakening.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.