Session Abstracts:
Postdoctoral Scholar
UC Berkeley
Nonstationary joint probability analysis of coastal wind and flood hazards under climate change impact
Co-Authors: Eyitayo Opabola (University of California, Berkeley)
Abstract: The United States coastal areas are highly susceptible to the impacts of extreme coastal wind and flood hazards. While both hazards can occur independently, climate change-induced fluctuation in hazard intensity and frequency is increasing the joint occurrence of coastal wind and flood hazards. To improve multihazard risk analyses of coastal communities, this study presents an approach for simulating the joint occurrence of intensity measures of coastal wind and flood hazards (i.e., wind speed and water level, respectively) under climate change impact. First, the nonstationary trends in the intensity measures recorded by meteorological stations over the last three decades are modeled using relevant climate variables. Then, the intensity measures are detrended and used to develop a copula-based joint probability model, resulting in joint hazard curves (i.e., wind speed vs. water level at a return period of interest). Finally, incorporating future climate projections from climate models, we develop a series of nonstationary joint hazard curves for coastal wind and flood hazards until 2100. The proposed approach is used to generate a series of nonstationary joint hazard curves for San Francisco, California, and Miami, Florida. These curves can be used to define multihazard scenarios for community resilience assessment.
Professor
University at Buffalo
Knowledge-Enhanced Deep Learning for Hurricane Track Synthesis under Changing Climate
Co-Authors: Baichuan Deng (University at Buffalo)
Abstract: The future hurricane activity plays a pivotal role in coastal risk assessment and resilience planning. To quantify the climate change impacts on hurricane hazards, it is necessary to incorporate the evolving climatic conditions into synthetic hurricane tracks consisting of genesis, translation and intensity modules. Either statistical or physical method can be used in the development of each module. However, the significant computational demands of physical models limit their application in hurricane risk assessment. Hence, the statistical models are usually preferred by engineers and typically formulated as polynomials with parameters identified through multiple linear regression using historical data. These polynomial models often fail to capture the highly nonlinear nature of hurricanes. To address this issue, a deep learning-based hurricane track model using multilayer perceptron is developed with the climate-sensitive environmental conditions as network inputs and the hurricane genesis, translation or intensity as network output. Due to the limited training dataset of historical hurricane records, additional domain knowledge is leveraged to improve the data efficiency. Specifically, the empirical equation-based knowledge of genesis potential index is incorporated into deep learning of hurricane genesis, the semi-empirical equation-based knowledge of beta-advection mechanism is incorporated into deep learning of hurricane translation, and the nonlinear differential equation-based knowledge of air-sea interaction is incorporated into deep learning of hurricane intensity. The integration of domain knowledge significantly enhances the convergence speed during network training compared to pure deep learning. Validation results also suggest that the high synthesis accuracy using the developed knowledge-enhanced deep learning for hurricane tracks is achieved.
Assistant Professor
Rice University
Future shifts in tropical cyclone risks across the US due to storm climatology change and socioeconomic growth
Co-Authors: Ning Lin (Princeton University), Daniel Chavas (Purdue University), and Michael Oppenheimer (Princeton University)
Abstract: Tropical cyclone (TC) hazards coupled with dense urban development along the coastline have resulted in trillions in US damages over the past several decades, with an increasing trend in losses in recent years. So far, this trend has been driven by increasing coastal development. However, as the climate continues to warm, changing TC climatology may also cause large increases in coastal damages in the future. Approaches to quantifying regional TC risk typically focus on total storm damage. However, it is crucial to understand the spatial footprint of TC damage and ultimately the spatial distribution of TC risk. Here, we quantify the magnitude and spatial pattern of TC risk (in expected annual damage) across the US from wind, storm surge, and rainfall using synthetic TCs, physics-based hazard models, and a county-level statistical damage model trained on historical TC data. We then combine TC hazard projections with US population growth and wealth increase projections to estimate end-of-century TC risk across the US Atlantic and Gulf coasts under a moderate (SSP2 4.5) emissions scenario. We find that neglecting rainfall and storm surge results in much lower risk projections. TC climatology change and socioeconomic change drive similar magnitude increases in overall risk (roughly 160%), and that their combined effect (638% increase) is highly nonlinear.
Assistant Professor
New York University
Identifying, measuring, and managing socioeconomic impacts of floods amid climate change
Abstract: Climate change worsens the impacts of hurricane-induced flooding, making it crucial to develop effective methods for identifying, measuring, and managing these escalating risks. Addressing such risks triggered by climate change and natural hazards is challenging due to inconsistent historical data, unpredictable weather patterns, conflicting stakeholder objectives, and a multitude of potential adaptation solutions. My research offers practical tools for assessing the socioeconomic impacts of floods and other climate-related risks. These tools are designed for real-world application, incorporating feedback from a range of stakeholders—including municipal offices, emergency management teams, and water resource experts. This talk will cover analyses at multiple scales, including local (New York City), national, and global levels.
Graduate Student Researcher
University at Buffalo
Multi-level Probabilistic Resilience Assessment for Bridges under Hurricanes
Co-Authors: Teng Wu (The University at Buffalo SUNY)
Abstract: Bridges are crucial parts of the transportation network, and their functionalities are highly vulnerable to hurricane hazards (e.g., wind, rain, surge/wave). Resilience assessment is an effective way to evaluate the ability of a bridge to withstand and recover from hurricane disruptions by characterizing its functionality change during and after a hurricane. Considering uncertainties related to hurricane hazards, bridge capacity and traffic demand, methods for assessing bridge functionality and resilience are necessarily probabilistic. While many studies investigated the uncertainties by assigning prescribed distributions at the structure level, they usually overlooked the accurate consideration of uncertainties resulting from the bridge component level and the transportation system level. To characterize uncertainty contributions from structure components (e.g., fragility and recovery analyses) and transportation system (e.g., traffic analysis) to bridge functionality, this study proposes a framework of multi-level probabilistic resilience assessment for bridges under hurricanes. Monte Carlo simulation is used to quantify the uncertainty of component-level bridge capacity under future hurricane hazards, and dynamic user equilibrium is employed to determine system-level bridge traffic load during hurricane disruptions. The multi-span simply supported (MSSS) concrete girder bridge is utilized as a case study to demonstrate the improved accuracy of the proposed multi-level resilience assessment approach compared to the structure-level resilience assessment.
Graduate Student Researcher
Rice University
Integrated Framework for Coastal Community Resilience to Storm Events: A Case Study of the Galveston Testbed
Co-Authors: Jamie Padgett (Rice University) and Stanley C. Moore
Abstract: The increasing frequency and intensity of hurricanes pose significant challenges to coastal communities, leading to widespread damage to physical infrastructure, social systems, and local economies. These events highlight the need for enhanced resilience strategies to mitigate the cascading impacts of such disasters. This study introduces an integrated framework to assess the resilience of coastal communities facing storm events, with a focus on the Galveston Testbed. The framework incorporates various computational tools and resources, with DesignSafe’s advanced cyberinfrastructure playing an important role in supporting the modeling of complex interdependencies between physical infrastructure and social systems in the aftermath of hurricanes. High-resolution assessments capture damage to buildings, power grids, and transportation systems under multiple hurricane-induced hazards, including wind and storm surge. Beyond quantifying direct physical impacts, the methodology simulates recovery by tracking the functionality of critical infrastructure networks and evaluating population dynamics following disruptions. By incorporating these coupled socio-physical factors, the model offers a comprehensive view of community resilience, allowing for the projection of key recovery metrics, such as access to essential services, housing availability, and population displacement over time. The focus on the Galveston region provides valuable insights into the challenges coastal communities face and how this framework can inform more effective recovery planning. The study emphasizes the importance of understanding infrastructure performance, social vulnerabilities, and recovery trajectories, offering actionable strategies to bolster community resilience against increasingly severe storm events.