Evaluation of Severe Weather Outbreak Diagnoses for Multiple Ranking Indices

In recent years, several attempts have been made at ranking severe weather outbreaks of various types or of any type based on preconceived notions of their relative severity. Many of these schemes (indices) use multiple variables with relative weights, in an attempt to limit the biases and nonmeteorological artifacts associated with a single severe weather report variable. However, because no formal definition of severe weather outbreaks has been established, secular trends exist with many severe weather report variables often used to quantify the severity of outbreaks, and reports of severe weather are often estimated, erroneous, and/or incomplete, considerable uncertainty exists regarding the relative severity of severe weather outbreaks. Consequently, there is no obvious single index that ranks severe weather outbreaks most accurately.

Recent evidence suggests that the spatial extent for which magnitudes of severe weather diagnostic variables exceed values favorable for significant severe weather can be used to diagnose the severity of the outbreaks with substantial skill. This study attempts to quantify the uncertainty in skill associated with this so-called areal coverage technique by varying the weights of severe weather report variables used to rank the outbreaks. All severe weather events exceeding threshold report number and density criteria from 1979 to 2013 are evaluated. The weights of each report variable used in the ranking indices are changed objectively to generate over 30,000 ranking indices. Preliminary findings suggest that (1) considerable skill exists in the identification of the most significant events for each index developed, (2) large variability in this skill exists among the indices, and (3) consistently greater skill is noted when tornado and hail report variables are weighted higher than wind report variables.