City University of New York (CUNY) CUNY Academic Works Dissertations and Theses City College of New York 2020 Using statistical learning approaches to understand trends and variability of tornadoes across the continental United States Niloufar Nouri CUNY City College How does access to this work benefit ou?y Let us know! More information about this work at: https://academicworks.cuny.edu/cc_etds_theses/940 Discover additional works at: https://academicworks.cuny.edu This work is made publicly available by the City University of New York (CUNY). Contact: [email protected] Using statistical learning approaches to understand trends and variability of tornadoes across the continental United States by Niloufar Nouri A dissertation submitted to the Graduate Faculty in Civil Engineering in partial fulfillment of the requirements for the degree of Doctor of Philosophy The City College of New York The City University of New York 2020 © 2020 Niloufar Nouri All Rights Reserved i This manuscript has been read and accepted for the Graduate Faculty in Engineering in satisfaction of the dissertation requirement for the degree of Doctor of Philosophy Naresh Devineni, Chair of Examining Committee Date Ardie Walser, Associate Dean for Academic Affairs Date Examining Committee: Dr. Naresh Devineni, Department of Civil Engineering, the City College of New York, Dr. Reza Khanbilvardi, Department of Civil Engineering, the City College of New York, Dr. Nir Krakauer, Department of Civil Engineering, the City College of New York, Dr. Ardavan Yazdanbakhsh, Department of Civil Engineering, the City College of New York Dr. Valerie Were, Social Science Lead at the NOAA Center for Earth System Sciences and Remote Sensing Technologies THE CITY COLLEGE OF THE CITY UNIVERSITY OF NEW YORK ii Abstract The annual frequency of tornadoes during 1950-2018 across the major tornado-impacted states were examined and modeled using anthropogenic and large-scale climate covariates in a hierarchical Bayesian inference framework. Anthropogenic factors include increases in population density and better detection systems since the mid-1990s. Large-scale climate variables include El Niño Southern Oscillation (ENSO), Southern Oscillation Index (SOI), North Atlantic Oscillation (NAO), Pacific Decadal Oscillation (PDO), Arctic Oscillation (AO), and Atlantic Multi-decadal Oscillation (AMO). The model provides a robust way of estimating the response coefficients by considering pooling of information across groups of states that belong to Tornado Alley, Dixie Alley, and Other States, thereby reducing their uncertainty. The influence of the anthropogenic factors and the large-scale climate variables are modeled in a nested framework to unravel secular trend from cyclical variability. Population density explains the long-term trend in Dixie Alley. The step-increase induced due to the installation of the Doppler Radar systems explains the long-term trend in Tornado Alley. NAO and the interplay between NAO and ENSO explained the interannual to multi-decadal variability in Tornado Alley. PDO and AMO are also contributing to this multi-time scale variability. SOI and AO explain the cyclical variability in Dixie Alley. This improved understanding of the variability and trends in tornadoes should be of immense value to public planners, businesses, and insurance-based risk management agencies. The next phase of the study is focused on the spatial and temporal characteristics of large tornado outbreaks (LTOs) which are rated F2(EF2) or greater on Fujita (Enhanced Fujita) scale and has struck several counties in one day. A statistical assessment of changes in the LTOs clusters for two consecutive 30-year time periods 1950-1980 and 1988-2015 has been performed and the findings show a geographical shift of the central impact locations towards Southeast of the United States. The spatial shift is also accompanied by a reduction in cluster variance which suggests LTOs has become less dispersed between the two period. We investigate changes in tornado inter-arrival rate over time during the period of study using an exponential probability model. Results showed that the arrival rate has changed from 124 days during 1950-1980 to 164 days during 1977-2007, which means LTOs were less frequent in the recent period. The analyses performed in this study support previously reported findings in addition to providing complementary information on LTO clustering behavior and return period. Key Words: Statistical Learning, Hierarchical Bayesian Models, Tornado Data, ENSO, Machine Learning, Trend Analysis, Tornado Clusters, Spatio-temporal Analysis iii Acknowledgement Foremost, I would like to express my sincere appreciation and gratitude to my advisors professor Naresh Devineni and professor Reza Khanbilvardi for the continuous support of my PhD study, for their motivation, scientific advices and the insightful discussions about the research. Without their guidance and constant feedback this dissertation would not have been achievable. Besides my advisors, I would like to thank the rest of committee members Prof. Nir Krakauer, Dr. Valerie Were, Prof. Yazdanbakhsh, for their valuable comments and encouragement, which helped me to broaden my perspectives on this research. I would also like to say heartfelt thanks to my Mom, Dad and my sister for always believing in me and encouraging me to follow my dreams. This research was supported by department of Civil Engineering at the City College of New York and the U.S. Department of Energy Early CAREER Research Program Award # DE-SC0018124. Partial support is also provided by NOAA-EPP/MSI Grant #NA16SEC4810008. iv ردخت تو گر بار دانش بگیرد هب زری آوری چرخ نیلوفری ر ا خس " انصر رو " To Nasrin, Reza and Mina and to those who inspired me but will not read this… v Contents Chapter 1: Introduction ................................................................................................................... 5 1 Tornadoes in the United States ............................................................................................... 6 2 Research background .............................................................................................................. 7 2.1 Studies on (in)consistency in the U.S. tornado data ......................................................... 7 2.2 Studies on tornado frequency and trends ......................................................................... 9 2.3 Statistical models for tornado variabilities ..................................................................... 10 2.4 Studies on the impact of climate variables on tornado activity ...................................... 11 3 The gap in previous research ................................................................................................ 13 4 Research scope ...................................................................................................................... 13 Chapter 2: Explaining the trends and variability in the United States tornado records using climate teleconnections and shifts in observational practices ....................................................... 15 Abstract ......................................................................................................................................... 16 1. Introduction ............................................................................................................................. 1 2. Methods................................................................................................................................... 8 2.1. Data .................................................................................................................................. 8 2.1.1. Tornados ................................................................................................................... 8 2.1.2. Population Density .................................................................................................. 12 2.1.3. Large-scale Climate ................................................................................................ 12 2.2. Analysis and Modeling................................................................................................... 12 vi 2.2.1. Principal Component Analysis and Wavelet Decomposition ................................. 13 2.2.2. Hierarchical Bayesian Models ................................................................................ 14 3. Results ................................................................................................................................... 18 3.1. Explaining the Variance ................................................................................................. 18 3.2. Inference of the Significant Predictors ........................................................................... 26 4. Summary ............................................................................................................................... 35 Chapter 3: Investigating the spatial manifestation and rate of arrival of large tornado outbreaks 38 Abstract ......................................................................................................................................... 39 1 Introduction ........................................................................................................................... 40 2 Method .................................................................................................................................. 42 2.1