USGS Grant Number: G14AP00172
Year Established: 2014 Start Date: 2014-09-01 End Date: 2016-08-31
Total Federal Funds: $248,556 Total Non-Federal Funds: $249,182
Principal Investigators: Gabriele Villarini, Jeffery Czajkowski, Erwann MichelKerjan
Abstract: We propose to develop statistical models to describe the relation between inland flooding associated with North Atlantic tropical cyclones (TCs) and impacts (direct economic losses and insurance claims) in the United States. Inland flooding is of high societal and economic relevance, but regretfully has received very little attention as most U.S. TC loss assessment efforts are focused on coastal flooding. Moreover, it is often the case that the most severe impact from heavy rainfall and fresh water flooding is far removed from the center of circulation of these storms, up to hundreds of kilometers away. The main outcomes of the proposed research are: 1) the identification of the areas that are more at risk from inland flooding from North Atlantic TCs; 2) the characterization of the extent and magnitude of these events; 3) the development of statistical models relating flood magnitude to direct economic losses importantly controlling for the associated exposure and vulnerability aspects over the period 2000-2012; 4) the use of the resulting empirical relationships to perform sensitivity analysis examining the potential impacts of pre-2000 TCs under the current level of exposure and vulnerability. The project will be carried out over a two-year time period. During Year 1, we will assemble spatially and non-spatially hazard, vulnerability and exposure data sets, as well as the associated data regarding claims and losses from the time period of 2000-2012, during which 100 TCs passed at least 500 km from the U.S. coast. This data acquisition and proper assembling is a non-trivial effort given the typical geographic impact of North Atlantic TCs. The statistical modeling of the hazards, vulnerability and exposure will begin towards the end of Year 1. During Year 2, we will complete the statistical modeling of the TC flood hazard and economic losses. We will also measure the degree of flood insurance coverage in place, thus being able to measure how many people exposed to these TCs are not properly protected financially from an insurance perspective and consequently might be requesting federal disaster relief. We will also examine the temporal changes in magnitude and frequency of TC floods. Moreover we will be able to use these developed models to examine what economic impacts TC floods occurring during the 20th century would have had if they had happened in the first decade of the 21st century. Analyses will leverage heavily on USGS discharge data from the hazard side, a unique access to the federally-run National Flood Insurance Program (NFIP) data for flood claims (flood in the United States is mainly insured by this public program), damage and some exposure data, and Hazards-United States (HAZUS) for the remaining relevant exposure and vulnerability data. We will focus on the assessment of the areas that are more severely affected by North Atlantic TC floods east of the Rocky Mountains, and examine the relation between impacts, hazards, exposure and vulnerabilities. Because of the scale of the weather and climate systems at play, we adopt a regional approach from the hazard modeling perspective. One of the problems in dealing with streamflow data at the regional scale is that there is an intrinsic dependence of discharge on drainage area that needs to be accounted for. The approach we will use is to normalize the peaks caused by TCs by the at-site 2-year flood peak. We will describe the relation among impacts, hazards, exposure and vulnerabilities by developing risk assessment statistical models for TC flooding and explore variability in the relationships. The basic idea is that we have a predictand (either claims or losses) and a vector of potential predictors for each of the three main components hazard, exposure and vulnerability. Different modeling approaches will be tested, and different methods for the selection of the relevant predictors will be considered. The procedures and models that we will develop are designed for broad use by USGS National and District Offices for flood and water resource assessment studies. Additional users will include federal, state and local groups for emergency and recovery purposes.