Examining Thousands of Tornadic and Nontornadic MCS Cells Using Gridrad-Severe

Although anticipating tornadogenesis is overall fairly difficult, it can be even more difficult for tornadoes from nonsupercellular storms, which are much more likely than tornadoes from supercellular storms to be warned after tornadogenesis has occurred. In this study, we specifically examine the three-dimensional radar-observed differences between thousands of tornadic and nontornadic non-supercellular mesoscale convective system (MCS) cells using the GridRad-Severe dataset. Their general kinematic characteristics are examined and a null population is selected from nontornadic storms that otherwise have kinematic characteristics approximating those of tornadic storms. Three-dimensional radar data are analyzed and compared between the tornadic and null populations, and analyses show measurable differences between the null storms (at peak intensity) and tornadic storms (both at tornadogenesis and 20 minutes prior), particularly in single-polarization data. Knowing the variation in the mean presentation of these storms has the potential to aid in the warning process, particularly in directing future efforts for tornadogenesis prediction using machine learning.