The preliminary investigation of Edwards and Thompson (2000) will be expanded to include a data set of 458 supercells that were observed across the contiguous United States from March 1999 through June 2001. Supercells were identified in base reflectivity and storm-relative velocity imagery, and an associated proximity sounding for each was interpolated spatially from hourly RUC-2 model analysis grids for each storm at the nearest surface observing site. Details regarding accuracy and representativeness of the RUC-2 analysis soundings will be presented in a companion paper by Thompson et al. (2002).

Our initial analysis of the sounding profiles confirms the findings of several previous studies, in that 0-1 km storm-relative helicity and lifted condensation level (LCL) height are the best single parameters in discriminating between significant tornadic and nontornadic supercells. We will also examine multiple combination parameters, storm-relative winds (both in fixed height and normalized to storm depth) for observed and estimated storm motions, fixed- and variable-depth shear magnitudes, and layer lapse rates and relative humidity, among numerous other parameters. A primary goal of this investigation is to aid in the operational discrimination between environments supportive of significant supercell tornadoes and nontornadic supercells. A sample of 75 proximity soundings for discrete non-supercells will also be used to quantify differences between supercells and non-supercell sounding parameters. Additionally, our findings will aid in the calibration of various supercell-related parameters as part of the hourly objective analysis scheme at the Storm Prediction Center described by Bothwell et al. (2002).