Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Leveraging “Big Data” to Better Understand and Predict Deep Convection Initiation
This research explores deep convection initiation (DCI) using large datasets and a variety of statistical and machine learning methods over the central United States. There are three main foci: data development and quality assessment, climatological characteristics of DCI, and DCI prediction. The first focus, data development and quality assessment, compares three thunderstorm identification methods and their accuracy compared to a combination of radar and lightning mapping array data. These results show that a method incorporating lightning data in addition to radar data is best. Thunderstorm tracks using this method are used to generate a large dataset of DCI points used to analyze climatological characteristics of DCI, and train and evaluate machine learning models to predict DCI occurrence. The second focus, analysis of climatological characteristics of DCI, shows that that the overall favorability, interannual variability, and diurnal time series of DCI all vary spatially, with difference regions having unique characteristics. The third focus, DCI prediction, focuses on the value of large datasets for the development of machine learning-based DCI prediction algorithms. Results show that models developed over smaller areas generally do not perform as well as models developed over larger areas, and analysis of models suggests that factors regulating DCI may vary spatially, although large domain models struggle to represent this even when spatial information is incorporated into the machine learning model as predictors.
Atmospheric sciences|Artificial intelligence|Geophysics
Shield, Stephen A, "Leveraging “Big Data” to Better Understand and Predict Deep Convection Initiation" (2023). ETD collection for University of Nebraska - Lincoln. AAI30488188.