Unveiling the Risks: Remote Sensing and Machine Learning for Data-Driven Aboveground Storage Tank Vulnerability Assessment

Abstract

Energy infrastructure critically relies on above-ground storage tanks (ASTs) for petroleum extraction, refining, processing, manufacturing, transportation, and distribution. However, ASTs are particularly susceptible to failure from attacks and natural hazards, particularly flood-related events. When ASTs fail due to flood hazards, hazardous materials can migrate offsite, posing severe environmental and public health risks to surrounding communities. To enhance data and tools for affected community groups and regulatory capacity, comprehensive data on AST characteristics and generalizable models assessing vulnerability and failure risk are needed. This dissertation develops open-source data products representing AST utilization and quantifies failure potential using remote sensing, publicly available hazard data, Bayesian statistics, and deep learning methods.This research produced: 1) a dataset of the location and diameter for over 130,000 ASTs from five labeled classes (external floating roof tanks, closed roof tanks, spherical pressure tanks, sedimentation tanks, and water towers) across the contiguous United States; 2) a quantification of the influence variability in tank and hazard data, particularly in tank height, floodwater velocity, and inundation, plays in the uncertainty in tank failure predictions, 3) the identification of understudied sites, including Long Beach, California, and the New York City metropolitan area, that contain hazard-exposed, failure-prone storage tanks.