A Deep Learning Approach to Examining and Predicting TC Rapid Intensification
Abstract
Funds are provided to conduct research on prediction of tropical cyclone (TC) rapid intensification. The PI~s hypothesize that the limitations of empirical and statistical methods for predicting RI stem from an oversimplification of the TC environment and interactions with the TC convective structures. In addition, they hypothesize that the numerical models are limited by inaccurate initializations of the TC state. This project will attempt to establish that deep learning can overcome both of these limitations by utilizing the rich amount of information in passive microwave satellite imagery, along with ancillary data on the TC environment, in sophisticated deep learning models that attempt to predict the precise timing of RI. The goal is to improve our understanding of the variables and processes that trigger RI by establishing tools that probe the sensitivity of various TC structures to environmental conditions while also maintaining unique fidelity to the observed state of the TCs. The will apply the value of the rich in-situ dataset of the North Atlantic and East Pacific basins to the other basins of the globe with mathematical rigor. The goals of this work are to create more accurate and more precisely timed empirical models of RI development, and also to generate unique insights on TC structural sensitivity to environmental conditions leading to RI events.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 17, 2020
- Source ID
- N000142012149
Entities
People
- Anthony Wimmers
Organizations
- Office of Naval Research
- United States Navy
- University of Wisconsin System