Distinguishing Characteristics of Tornadic and Non-Tornadic Supercell Storms from Composite Mean Analyses of Radar Observations

Distinguishing between tornadic and non-tornadic supercell storms has been an important challenge faced by the research and forecasting communities. Observations of storms and their environments have been used in previous attempts to improve discrimination, but distinguishing between these storm types has remained difficult. In some cases, this is driven (in part) by a limited number of adequate observations. Detailed case studies using very high spatial and temporal resolution data have led to significant advancements in understanding, but these data are not routinely available. Here, we attempt to overcome some of these limitations using historical radar observations from the NEXRAD WSR-88D network. In particular, data from 127 severe weather days during a 7-year period (2011-2017) are used to: i) objectively identify and track supercell storms at 5-min intervals, ii) extract storm-centered volumes of these storms for analysis, and iii) produce probability-matched composite means (PMMs) in three dimensions of available physical and kinematic variables, including radar reflectivity at horizontal polarization, differential reflectivity, specific differential phase, copolar correlation coefficient, velocity spectrum width, radial divergence, and azimuthal shear. This approach results in 478 non-tornadic supercell storms and 294 tornadic supercell storms identified for analysis. PMMs centered on echo top maxima and in a horizontal coordinate system rotated such that analyzed storm motion points in the positive x-direction are created in an altitude coordinate relative to ground level for non-tornadic supercells at the time of “peak strength” (maximum echo top altitude) and for tornadic storms both at and prior to tornadogenesis. PMM tornadic storm characteristics are found to be stable during time periods preceding tornadogenesis (up to 30 minutes prior) and robust differences are found between tornadic and non-tornadic supercells. In particular, low- (1-3 km AGL), mid- (4-7 km AGL), and upper-level (8+ km AGL) azimuthal shear is found to be vertically aligned in tornadic supercells at and prior to tornadogensis and vertically misaligned in non-tornadic supercells. Physical metrics also show unqiue differences, including an “open” low-level hook echo in tornadic storms and a “closed” hook echo in non-tornadic storms from radar reflectivity analysis, a mid-level differential reflectivity dipole that is oriented relative to storm motion differently between supercell types, and a larger area of enhanced low-level velocity spectrum width in tornadic storms. An illustration summarizing these differences is provided here.