For each severe weather type, we evaluated which parameters differentiate best between the intensity categories using non-parametric statistical tests, namely Mann-Whitney U-test and Kruskal-Wallis test. We evaluated whether and how well the parameters / indices differentiate between the intensity categories of individual severe weather threats.
We found that for hail severity discrimination, buoyancy related parameters attained the best scores, whereas wind profile-related parameters performed better for severe wind events. However, for both of them, the combination of CAPE and Deep Layer Shear (bulk 0-6 km shear) performed even better. This was different for excessive precipitation for which low-level moisture and the depth of the warm cloud layer were the best predictors. In case of tornadoes, the vertical wind profile is more relevant than paramaters related to buoyancy.