Evaluation of weather catastrophe data for use in climate change investigations

A 1950-1994 data set of major weather losses developed by the property insurance industry was examined to assess its potential utility in climate change research and use in assessing the relevance of recent extreme losses in the United States. A process for adjusting these historical storm losses to ever-changing factors including dollar values, amount of insurance coverage per area, and the sensitivity of society to damaging storms was developed by the industry. Analysis of the temporal frequency and losses of these adjusted weather catastrophes revealed differences according to the amount of loss. Temporal changes since 1975 in the catastrophes causing $35 to $100 million in loss were strongly related to changes in U.S. population, whereas catastrophes that created insured losses greater than $100 million appear related to both shifting weather conditions and to regional population changes. This evaluation revealed that the industry's catastrophe adjustment technique did not adequately allow for changes in various demographic and social factors affecting damage; however, results suggest use of population values for normalizing the adjusted catastrophe database to allow meaningful studies of their temporal variability.