Assessing United States County-Level Exposure for Hurricane and Tropical Storm Epidemiological Research

摘要

Hurricanes and other tropical storms bring severe impacts to U.S. communities. These impacts can result from a variety of storm-related hazards, including extreme wind, rain, flooding, and tornadoes. Epidemiological studies vary widely in how they classify exposure to tropical storms, using various hazard-based metrics and, in some cases, using distance from the storm as a surrogate for exposure to storm-related hazards. Here we measure county-level exposure to hurricanes and tropical storms in the United States based on distance from the storm, maximum sustained wind, rainfall, flooding, and tornadoes for all land-falling or near-passing Atlantic basin storms for 1988—2015. We show that the locations identified as storm-exposed varied substantially when switching among these metrics. For example, most wind-based storm exposures were limited to southern counties near the coast, while flood- and rain-based exposures often extended to inland and northern counties. We also show that distance to the storm served as at best a moderate, and often a poor, surrogate in identifying exposure to storm-related wind, rain, floods, or tornadoes. Therefore, when impact studies use distance as a surrogate of exposure to tropical storm exposures or use one hazard-based metric (e.g., wind-based) when the impact is partly or fully caused by a different storm hazard, the analysis will be prone to exposure misclassification, which can mask true associations, even strong associations. To facilitate future research, we make this multi-hazard storm exposure data available through open-source software.