Jamie Ellen Padgett

Structures and infrastructure are often densely distributed in coastal regions and highly susceptible to hurricane-induced stressors. Evidence from past hurricane events and experimental studies demonstrates that neighboring structures significantly alter the structure-to-structure demand induced by hurricane-related stressors, such as storm surge and wave loads, thereby influencing the damage pattern observed across a region. Capturing this modified load requires detailed computational fluid dynamics (CFD) analysis, making it difficult to incorporate such effects into current regional risk assessments. This paper proposes a framework for incorporating the modified loads directly into the formulation of parameterized fragility functions in a practical way. Our methodology uses a set of parameters to describe the portfolio layout, and its associated effects on the load modifications (i.e., shielding or channeling) are captured from detailed CFD analyses. The influence of the modified loads on the structural failure probabilities is embedded within the fragility model by considering the portfolio layout parameters during the model fitting procedure. To reduce the high computational expense of CFD analysis, active learning algorithms are leveraged to selectively choose only the most critical parameter combinations for numerical analysis. The proposed framework is demonstrated for aboveground storage tanks (ASTs) located within a tank farm, as a representative structural portfolio. The derived parameterized fragility models reveal that failure probability estimates vary significantly depending on the spatial arrangement of structures within the portfolio. Subsequent scenario-based risk estimates leveraging these fragilities show differences of 21.5% to 54.5% in failure probability and 31.4% in expected volume of hazardous material spills, compared to estimates obtained without considering the influence of neighboring structures. The proposed framework offers a path to improve risk assessment of densely arranged portfolios. We expect our contribution to render tailored structural interventions by revealing heterogeneous impacts, thus mitigating (or reducing) investments when dealing with coastal structural portfolios.

Projections of future changes in tropical cyclone (TC) rainfall are critical for understanding evolving flood risks and infrastructure impacts under climate change. This study couples a physics-based Tropical Cyclone Rainfall (TCR) model with the Columbia HAZard (CHAZ) model, a statistical–dynamical TC downscaling framework, which generates synthetic storm tracks and intensity, to produce large synthetic ensembles of TC generated rainfall downscaled from 12 CMIP6 models. TC rainfall is simulated in the North Atlantic basin based on the CMIP6 historical and future simulations under three Shared Socioeconomic Pathways (SSP2-4.5, SSP3-7.0, SSP5-8.5). Projections from this integrated framework are evaluated against the Geophysical Fluid Dynamics Laboratory (GFDL) Rainfall Climatology and Persistence model (R-CLIPER), a statistical TC rainfall model that provides an alternative to the dynamical simulations. TCR projects widespread increases in TC rainfall, with the strongest changes along the US East and Gulf coasts and for major hurricanes (Categories 3–5). By late century, average rainfall increases reach 40–100% in TCR compared to 30–60% in R-CLIPER, with the largest increases in the Northeast regions. Extreme 24-hour rainfall increases by up to 55% in TCR, roughly twice the magnitude simulated by R-CLIPER. Key drivers include higher TC intensities, increased atmospheric moisture, and projected slower translation speeds, which together contribute to increasing flood risks, especially when combined with more frequent sequential TC events.

Coastal hazards and climate change significantly threaten the resilience of railway systems, increasing stresses on global freight transportation, supply chains and economic stability. When it comes to system resilience, resource availability and allocation have been proven to be leading contributors to downtime and losses, alongside the physical vulnerability to extreme loads. To support the quantification and pursuit of system resilience, here we present a probabilistic framework that addresses gaps in resilience modeling of railway systems. Specifically, it systematically integrates tailored structural damage and restoration models across an infrastructure portfolio, while comparatively assessing network-level functionality over time with alternative approaches to recovery resource allocation. Applied to the railway network in Mobile and Baldwin Counties, Alabama, the framework estimates damage states, restoration costs and times, modeling drop and recovery of network functionality. Findings indicate that sea-level rise considerably affects service reinstatement, reducing resilience index up to 80% when combined with hurricanes. Resource allocation strategies also impact resilience, with variations resulting in up to 75% differences in resilience estimates. These results underscore the need to consider resource constraints and sea-level rise in resilience planning, offering nuanced resilience quantification to support decision-making for mitigation and response strategies, benefiting policymakers, infrastructure managers, insurers, and agencies.

California is a seismically active region that contains approximately 26,000 bridges. Historical earthquakes have caused severe damage and collapse of bridges in California, resulting in casualties, economic losses, and disruptions to transportation networks. Seismic fragility models of bridges estimate the probability of exceeding damage states at varying ground motion intensity levels. These models can be utilized for bridge vulnerability assessment, risk and resilience quantification, and to support earthquake preparedness and response planning. Recent studies have developed a new generation of seismic fragility models for bridges, demonstrating several advancements when compared with the widely used first-generation HAZUS models. These models, termed second-generation fragility functions, are derived through detailed dynamic response analyses and are differentiated at the component level, with more rational, performance-based criteria for bridge grouping and archetype sampling. This study compiles and adapts a comprehensive set of second-generation fragility models from the literature. As part of this compilation, we filtered out fragility functions with outdated capacity models and unrealistic bridge configurations, establishing uniform bridge grouping criteria, and harmonized different cross-model ground motion intensity measures and bridge component definitions. The database comprises approximately 2300 component- and 500 system-level fragility models categorized into 26 bridge groups. It can serve as a valuable resource for researchers and practitioners conducting various related analyses to enhance the seismic resilience of bridge infrastructure in California.

Rapid restoration of transportation systems following earthquake disruptions is essential due to its significant impact on commutes, freight transport, and emergency medical response. This study develops a framework to model bridge post‐earthquake functional recovery, which is a critical step in assessing the seismic resilience of transportation networks. A simulation‐based strategy is used, which is initiated by component‐level damage assessment. This is followed by three major modules that include functional state evaluation, impeding factor definition and duration, and repair or replacement duration. The functional state module determines the bridge closure decisions (e.g. partial or complete lane closure, weight restriction) immediately after the earthquake and during the reopening phase. The duration estimates include the time delays before the initiation of repairs (in the impeding factor module) and the time needed to perform any necessary repairs or to replace the bridge (in the repair or replacement duration module). Worker allocation schemes and repair sequencing are explicitly considered to realistically reflect local construction practices. The framework architecture and duration‐based input parameters were informed by a series of interviews with California bridge engineers and builders. Nonetheless, the overall methodology is flexible and can be easily adapted to other jurisdictions.

Timely and accurate detection of hurricane debris is critical for effective disaster response and community resilience. While post-disaster aerial imagery is readily available, robust debris segmentation solutions applicable across multiple disaster regions remain limited. Developing a generalized solution is challenging due to varying environmental and imaging conditions that alter debris' visual signatures across different regions, further compounded by the scarcity of training data. This study addresses these challenges by fine-tuning pre-trained foundational vision models, achieving robust performance with a relatively small, high-quality dataset. Specifically, this work introduces an open-source dataset comprising approximately 1,200 manually annotated aerial RGB images from Hurricanes Ian, Ida, and Ike. To mitigate human biases and enhance data quality, labels from multiple annotators are strategically aggregated and visual prompt engineering is employed. The resulting fine-tuned model, named fCLIPSeg, achieves a Dice score of 0.70 on data from Hurricane Ida -- a disaster event entirely excluded during training -- with virtually no false positives in debris-free areas. This work presents the first event-agnostic debris segmentation model requiring only standard RGB imagery during deployment, making it well-suited for rapid, large-scale post-disaster impact assessments and recovery planning.