Using Characteristics of Tornadic Debris Signatures to Estimate Tornado Intensity

Tornado debris signatures have been among the most notable uses of the dual-polarization upgrade to the WSR-88D network, as they provide confirmation of an ongoing tornado. Past studies have established correlations between TDS characteristics (height, width, etc.) and metrics of tornado intensity, but the relative novelty of dual-polarization meant that the number of available cases was small. In this study, all significant (EF2+) tornadoes observed within 150 km of a WSR-88D radar with dual-polarimetric capability across the continental U.S. were studied, along with a subset of weak (EF0-1) tornadoes from May 2010 to present (N=779). Given the presence of a TDS, several TDS variables were collected, including maxima in rotational velocity, azimuthal shear, TDS height and width, and vortex diameter. These variables were then compared to observed maximum wind speed estimates obtained from National Weather Service damage surveys via the Damage Assessment Toolkit. By using several statistical procedures, including principal component analysis, multiple linear regression, and multivariate kernel density estimation, a pair of models (linear and probabilistic) was created to predict the maximum wind speed based on maximum TDS height, rotational velocity, and azimuthal shear. Both models displayed good performance, with the linear model having a r² value of 0.65 and a MAE of 12.7 mph, and the probabilistic model having a r² value of 0.75 and a MAE of 14.3 mph. These models could be used in an operational setting to improve the performance of tornado warnings in the Impact-Based Warning (IBW) framework.