Modeling Tropical Cyclone and Weather Risk in a Changing Climate: Machine Learning, Hazards, and Socio-Economic Inequalities

Abstract

The changing nature of coastal risk, driven in part by tropical cyclone (TC) variability and sea-level rise(SLR), presents a critical challenge for coastal populations. Recent TC events such as Harvey (2017) and Ida (2021) highlight the urgent need to study TC hazards and their profound impact on human life and infrastructure in the context of a changing climate. Modeling coastal risk necessitates a comprehensive understanding of the dynamics of TC activity, SLR, their associated uncertainties, and the complex interactions that may lead to compounding or cascading impacts. To address these issues there is a need for effective and equitable climate adaptation policies, particularly in light of the unequal distribution of flood risks among different socioeconomic and demographic groups. This dissertation conducts a comprehensive analysis of coastal risk challenges, blending climate science, machine learning (ML) and coastal adaptation strategies, with the purpose of deepening our understanding and management of TC and coastal risk. The key focus areas include: exploring correlations between SLR and TC activity; developing ML-based frameworks for modeling TC-induced storm surge and precipitation; assessing the impacts of Rapid Intensification (RI) on storm surge and precipitation levels; and examining socio-economic disparities in adaptation strategies. The objective of this dissertation to advance the physical modeling of coastal risk in a changing climate, with a particular emphasis on coastal flooding from TCs, leveraging the latest advancements in ML and climate science. This dissertation also provides a framework for implementing equitable climate adaptation policies to protect vulnerable communities from disproportionate flood hazards.