Is the frequency of severe storms changing with time? This is a question whose answer is complicated by several factors. Storms are more likely to be observed now than in the past, and in the vicinity of highly populated over less populated areas. Severe storms cannot be resolved well by climate models because they occur at too fine of a scale and because of limitations in climate models. Because of these limitations, we examine the changes in frequency of conditions which favor severe storms with respect to time. First, using previously published results from Craven and Brooks (2004), the relationships between moderately high levels of convective potential energy (CAPE J/kg) and vertical shear (m/s) and the occurrence of observed convective storms are reviewed. This review supports the use of high values of the product of CAPE and shear as a proxy for severe convective events.

Second, to gather information over a greater domain of space and time, CAPE and shear are calculated from the NCEP/ NCAR reanalysis data set using the Skew-t/Hodograph Analysis and Research Program (SHARP) program. This provided data on an approximately 2 degree square grid, 4 times daily from 1958 through 1999. The CAPE, shear and CAPE*shear values are summarized statistically and graphically. Statistically significant trends are identified using general linear models fit to this global data set. Estimation of so many models leads to issues associated with multiple comparisons, which are addressed using a method that limits the false discovery rate. Results show that significant changes in the frequencies of high CAPE*shear events have occurred over the last 40 years. These results vary spatially. An extension of these analyses to the output of global and regional climate models may allow inferences to be made regarding the frequency of convective storms under future climate change scenarios.