So-called “proximity soundings”, profiles from rawinsonde ascents taken in the vicinity of severe thunderstorms, have been invaluable in identifying environmental conditions associated with tornadoes and severe thunderstorms. Typically, soundings have been labeled as “proximity” if they are taken within 1-3 hours and within 100-200 km of severe thunderstorms, although the criteria used have varied greatly. Due to the lack of temporal and spatial coverage and issues with quality control, most of the datasets associated with proximity soundings have been small.

The NCAR/NCEP reanalysis project provides a possible alternate source of soundings for proximity studies. The temporal and spatial coverage is better than the rawinsonde network and some of the quality control issues (e.g., “short runs”, convectively-contaminated soundings, etc.) don’t come into play. We have begun a study of soundings derived from the reanalysis in order to test its applicability for use in proximity studies. The preliminary project has considered all 0000 UTC soundings in the US, east of the Rockies, from 1997-9, a period studied by Craven and Brooks, using observed data.

In general, the results are encouraging. Correlations are high between parameters important to convective weather derived from reanalysis and observed soundings. In addition, the capability of reanalysis-derived parameters to discriminate between tornadic and non-tornadic “significant” severe thunderstorms is almost as good as the observed parameters, using the same parameters (0-1 km wind shear, height of the lifted condensation level). The distribution of parameters is slightly different, but the sense is the same.

These results offer the possibility that a small number of parameters can be used to identify environments favorable to tornadoes and severe thunderstorms, and then the frequency of occurrence of the combination of those parameters can be described. As an example, in the case of the preliminary study, almost 200,000 soundings were created with approximately 160 (0.08%) associated with tornadoes. A combination of 4 parameters favorable to tornadoes identifies 2,000 soundings as “potentially tornadic”. Of those, 110 (5.5%) were associated with tornadoes. It seems plausible that global climatologies of those potentially tornadic soundings can be derived and changes in space and time identified over the period of the reanalysis. Similar proximity studies could be made for other weather events of interest.