Publication Date
2024
Document Type
Dissertation/Thesis
First Advisor
Haberlie, Alex M.
Degree Name
M.S. (Master of Science)
Legacy Department
Department of Earth, Atmosphere, and Environment (EAE)
Abstract
This work examines potential changes in convective mode associated with hazardous convective weather (HCW) using a suite of high-resolution regional climate simulations and artificial intelligence/machine learning (AI/ML). An AI/ML model called a convolutional neural network (CNN) is first generated and validated using an existing dataset of composite reflectivity images centered on tornado reports from 1996 – 2017 called SVRIMG (SeVeRe IMaGes). Each image in SVRIMG is annotated with four possible convective modes: 1) cellular; 2) mixed mode; 3) linear; and 4) other. The best-performing model, which achieves an accuracy exceeding 90% on testing images, is then used to diagnose storm mode in select simulated reflectivity images from the regional climate simulations. To align with SVRIMG, these images are spatiotemporally centered on simulated HCW events. Simulated HCW is quantitatively defined in this work as the occurrence of simulated updraft vertical velocity exceeding 25 m s–1 and a simulated reflectivity factor exceeding 40 dBZ. The CNN classifies convective mode associated with HCW from the three separate time periods found in the simulations: 1) historical (HIST; 1990-2005); 2) mid-century (MID; 2040-255); and 3) end-of-century (END; 2085-2100). MID and END are driven by climate change scenarios outlined in representative concentration pathways 4.5 and 8.5. CNN convective mode classification of HCW suggests increases in cellular frequency in areas East of the Mississippi River Valley during the spring and summer. Additionally, there is larger frequency variability noted in the future scenarios when compared with the retrospective.
Recommended Citation
Corner, Jeremy Malcolm, "The Future of Convective Mode in the United States" (2024). Graduate Research Theses & Dissertations. 7954.
https://huskiecommons.lib.niu.edu/allgraduate-thesesdissertations/7954
Extent
64 pages
Language
en
Publisher
Northern Illinois University
Rights Statement
In Copyright
Rights Statement 2
NIU theses are protected by copyright. They may be viewed from Huskie Commons for any purpose, but reproduction or distribution in any format is prohibited without the written permission of the authors.
Media Type
Text
