A0637
Title: Model-based clustering on the spatial-temporal and intensity patterns of tornadoes
Authors: Yana Melnykov - The University of Alabama (United States)
Yingying Zhang - Western Michigan Univesity (United States)
Rong Zheng - Western Illinois University (United States) [presenting]
Abstract: Tornadoes are one of nature's most violent windstorms that can occur all over the world except Antarctica. Previous scientific efforts were spent studying this natural hazard from genesis, dynamics, detection, forecasting, warning, measuring, and assessing. At the same time, the aim was to model the tornado datasets using modern, sophisticated statistical and computational techniques. The goal is to develop novel finite mixture models and perform cluster analysis on tornadoes' spatial-temporal and intensity patterns. First, to analyze the tornado dataset, a Gaussian distribution with the mean vector and variance-covariance matrix represented is used as exponential functions of intensity and time. Then, a Gaussian mixture model is employed, with mean vector and variance-covariance represented as exponential functions of intensity and time. Thirdly, manly transform parameters are added to the Gaussian mixture model to take care of the skewness in the tornado dataset. Computer algorithms obtain results. A summary of insights is provided about tornado forecasting and assessing.
