Colloquium: Modeling Directional Extreme Wind Speed Using an Extreme Value Regression Approach

Eva Murphy, PhD. Wake Forest University

ZSR Auditorium, Tuesday, September 12th 11:00am
Refreshments provided

Modeling Directional Extreme Wind Speed Using an Extreme Value Regression Approach Assessing the risk posed by high wind speeds is critical in numerous fields, ranging from structural design to energy generation and climate and weather forecasting. While extreme value (EV) models are commonly employed for extreme wind speed analysis, conventional methods do not consider the directional characteristics of wind behavior, which can significantly impact wind speed characteristics. In this talk, I will briefly overview the traditional parametric extreme value modeling approaches, including block maxima, peaks-over-threshold, and point process methods. I will then introduce our proposed methodology that extends these techniques to estimate extreme wind speeds conditioning on wind direction. As we will see, our approach employs a two-stage algorithm involving estimating parameters of the EV model through directional binning, followed by a harmonic regression to capture the parameter dependence on wind direction. A simulation study designed to evaluate the performance of our proposed methodology and an analysis of the directional wind speed distribution across various climate scenarios will also be discussed. I will conclude my talk by briefly discussing my related work on spatio-temporal wind speed modeling.