When the Coast Floods: Bridging Physics-Based Models, Machine Learning, and Coastal Decisions

Join this presentation In-Person: BEG Building 130, Main Conference Room 1.202 Or Via Zoom

Join Zoom Meeting
Meeting ID: 932 0137 7630
Passcode: 013011

Presenter

Jun-Whan Lee
Assistant Professor, Maseeh Department
Civil, Architectural, and Environmental Engineering
The University of Texas at Austin

Description

Coastal flooding caused by hurricanes and tsunamis is among the most destructive natural hazards, placing growing coastal populations at increasing risk. Digital twins have emerged as a promising tool for addressing this challenge by integrating real-time observations and numerical simulations to support monitoring, prediction, and decision-making. However, effective digital twins require flood models that are accurate, computationally efficient, and reliable. Despite advances in computational capabilities, no single modeling approach has fully achieved this balance. This seminar presents four research directions that advance coastal flood modeling toward this goal. The first develops a physics-informed static flood model that can generate high-resolution coastal compound flood maps at 5 m resolution, covering more than 500 million cells, in under ten minutes. The model also reduces the systematic overestimation common in traditional static approaches through improved inundation algorithms and calibration methods. The second investigates physics-informed machine learning for compound flooding, with a focus on identifying when, where, and how incorporating physical knowledge improves predictive accuracy. The third introduces a game theory-based framework for quantifying the spatial and temporal contributions of multiple flood drivers to hurricane-induced compound flooding. The final research effort uses the Material Point Method to simulate interactions among coastal forests, waves, and debris, demonstrating how natural features that reduce flooding may also amplify structural damage through debris damming. Together, these efforts demonstrate how physics-based simulations and data-driven methods can be integrated into coastal digital twins, supporting the development of more resilient communities capable of withstanding the growing risks of coastal flooding.