Improving Coastal Flood Risk Assessments for the Northeastern United States: New York City to Boston

Abstract:

Interest in extreme flood vulnerability in the Northeastern U.S. has increased significantly since Hurricane Sandy caused more than $50 billion dollars in damages. Despite increased focus there is still no overall consensus regarding the true return period in the region for flood events of Sandy’s magnitude. The application of Generalized Extreme Value (GEV) theory to water level data is one of the most common techniques for estimating the return period for these rare events. Here we assess the skill of this popular technique by combining modeled, instrumental and sedimentologically derived records of flooding for the region. We show that GEV derived return periods greatly and consistently underappreciate risk for sites from New York City east to southern Cape Cod. This is in part because at these locations maximum annual flood data represents a mixture of two very different populations of storms, i.e. tropically derived disturbances and extratropical Nor’easters. Nor’easters comprise a majority of floods with 10-yr return periods and shorter, hurricanes for 100-yr floods or longer, and a combination in between. In contrast, the GEV technique functions better in estimating the 100-yr flood for points north of Cape Cod including Boston. At these locations flooding occurs more often from just one type of disturbance, i.e. Nor’easters. However, modeled and sedimentary reconstructions of storms indicate hurricanes likely still dominate flood distribution at northern location like Boston for 500 yr or greater events. Results stress the need for separating storm populations before applying the GEV technique, especially where flood behavior can vary depending on the type of disturbance.