Applications of Synthetic Tracks for North Atlantic Tropical Cyclone Health Impact Assessment

Abstract

BACKGROUND AND AIM: Tropical Cyclones (TCs) can cause immense harm to people, so estimating their potential health impacts is an essential part of disaster planning. Potential TC-associated exposures and health impacts are mediated by TC frequency, location, and intensity, and thus cumulative landfall hazard. These TC features are influenced by recurring climate patterns like the El Niño Southern Oscillation (ENSO) and Atlantic Multidecadal Oscillation (AMO). However, each combination of climate patterns comprises few years in the reliable historical record, making existing observations poor representations of true risk profiles, and risking emergency preparedness “blind spots.” Here we aim to overcome this challenge in TC health impact assessment (HIA) by generating large synthetic North Atlantic TC datasets using resampled historical data representing multiple ENSO and AMO interactions. We further explore public health applications of this approach. METHODS: We have developed a TC HIA approach which explores different phases of recurring climate patterns while accounting for these challenges. We draw on a climatology- and statistics-based resampling algorithm to generate synthetic TC tracks representative of specific recurring climate patterns. We model these scenarios using combinations of ENSO (positive, neutral, or negative) and AMO (positive or negative) phases in historical data. We use projected maximum county-level experienced wind speeds in a Bayesian exposure-response model to estimate all-cause mortality among U.S. Medicare recipients. RESULTS:We have integrated algorithmically-generated TC tracks with an exposure-response model to create HIAs under varying climate pattern scenarios. Our approach is modular with respect to TC data, demographics, and exposure-response models, allowing adaptation for multiple TC HIA tasks. Our approach accommodates centuries of synthetic tracks without major obstacles to further scaling. CONCLUSIONS:We have demonstrated synthetic TC track applications for exploring TC activity and public health risk under varying climate pattern scenarios. Our approach could, with minor modifications, be extended to other climate and public health outcomes. KEYWORDS: Tropical Cyclones, Risk Assessment, Prediction, Modeling, Climate, Natural Disaster