Tornado Outbreak Climatology in the United States (1995–2020)

Tornadoes are one of the most dangerous natural hazards on Earth that often occur in clusters commonly known as an outbreak. On average, tornado outbreaks are getting larger and more productive with more tornadoes occurring per cluster. As a result, tornado outbreaks pose an increasing risk to life and property annually. For example, the February 5, 2008 outbreak produced 85 tornadoes over 14 hours and 482,000 sq. km. which resulted in 482 casualties. Tornado outbreaks can be difficult to forecast. Dynamic and statistical models highlight areas of severe weather potential, but these models do not yet specify the risks and characteristics associated with outbreaks. Additionally, they do not account for the projected influence of climate change on tornado outbreaks. To date, limited research has focused on climate change and environmental factors that influence tornado outbreak characteristics (e.g., cluster severity, tornado counts, casualty counts, and spatial domain). This dissertation addresses two main questions: (1) Do environmental factors statistically explain tornado outbreak characteristics such as accumulated tornado power, tornado counts, and casualty counts? (2) Do climate variables 15 days prior to an outbreak statistically explain environmental factors that are known to enhance tornadogenesis? I situate my research in the sub-disciplines of Physical Geography and Hazards Geography. My work falls within the four traditions. These traditions include the spatial tradition (spatial analysis is crucial to understand geometry and movement of phenomena), man-land tradition (relationships between humans and the inhabited land), earth science tradition (study of Earth's processes), and area studies tradition (in-depth study of an area's unique characteristics). My research falls within the spatial tradition and earth science tradition by using statistical methodologies to understand the physical processes that influence natural hazards (specifically tornadoes) in the United States from 1994 to 2020. Keeping that in mind, the goal of this dissertation is to understand the role of climate change on the environmental factors that influence tornado outbreak characteristics, including cluster severity, tornado counts, casualty counts, and spatial extent of the outbreak. This research's intellectual merit is the quantification of the change in tornado outbreak characteristics for changes in the environmental conditions at the synoptic scale, which is commensurate with the scale of the outbreak. The broader impacts of this research are quantitative methods for addressing the link between climate change and hazards, and the potential to use these methods to better anticipate future severe weather outbreaks. The dissertation is outlined as follows. I describe the foundations of tornado dynamics and formation in Chapter 1. I provide a brief historical description detailing the tornado climatology in the United States in Chapter 2. Then, I develop an objective clustering methodology to group tornadoes into their respective outbreaks using distances between tornadoes in both space and time in Chapter 3. I evaluate and assign values of the environmental factors that are present three hours before the initiation of the cluster with at least ten tornadoes in Chapter 4. I create a cluster severity metric called Accumulated Tornado Power (ATP) to understand the relationship between tornado and casualty counts for clusters with at least ten tornadoes in Chapter 5. I fit regression models to quantify the relationship between tornado and casualty counts for clusters with at least ten tornadoes in Chapter 6. Then, I develop a series of regression models to predict convective available potential energy, deep-layer bulk shear, and shallow-layer bulk shear using sea surface temperatures and sea level pressure conditions averaged over the 15 days prior to the occurrence of a cluster with at least ten tornadoes in Chapter 7. Finally, I provide a summary and highlight future work in Chapter 8. The findings in this dissertation quantify the relationships between cluster severity, tornado counts, casualty counts, environmental factors, and climate variables. The regression model in Chapter 5 indicates that cluster severity increases by 33% for a 1000 J/kg increase in convective available potential energy and by 125% for a 10 m/s increase in bulk shear. The regression models in Chapter 6 indicate that tornado counts increase by 4.7% and casualties increase by 28% for a 1000 J/kg increase in convective available potential energy. Additionally, tornado counts increase by 13% and casualties increase by 98% for a 10 m/s increase in bulk shear. The regression models in Chapter 7 indicate that sea surface temperatures and sea level pressure 15 days prior to a tornado cluster can be used to predict the convective available potential energy, deep-layer bulk shear, and shallow-layer bulk shear on a day with at least ten tornadoes.