Rising Variability in Thunderstorm-Related U.S. Losses as a Reflection of Changes in Large-Scale Thunderstorm Forcing*

Abstract

Abstract Thunderstorm-related normalized economic and insured losses in the United States east of the Rockies from the period 1970–2009 (March–September) exhibit higher peaks and greater variability in the last two decades than in the preceding two decades. To remove the bias from increasingly detected losses over time due to newly built-up locations, only large events that incurred normalized losses of at least $250 million (U.S. dollars) economically ($150 million insured) were selected. These are multistate damage events that are unlikely to have been missed at any time within the analysis period, thus providing for homogeneity of the events covered. Those losses, if aggregated, account for the major proportion (~80%) of all thunderstorm-related losses in the period 1970–2009. This study demonstrates that the pattern of variability in the time series of these losses can be seen as a reflection (“fingerprint”) of the temporal variability in severe thunderstorm forcing. The meteorological information on forcing is inferred from NCEP–NCAR reanalysis data. No final attribution of the climatic variability identified in thunderstorm forcing and losses—either to natural climate variability or to anthropogenic climate change—can be conclusively arrived at in this study because of the chosen methodology. Nevertheless, the expected impacts of anthropogenic climate change on the forcing of convective storms appear consistent with these findings.