Exploring Southeast United States Severe Environments Using Atmospheric Soundings

Abstract

Proximity sounding studies related to convective storms primarily focus on the Great Plains region and supercells, despite the highest percentage of deadly tornadoes and non-supercellular tornadoes occurring in the Southeast (SE) United States. Using hundreds of observed soundings across 39 cases from recent SE-focused field projects, we create and analyze a proximity sounding dataset for tornadic and nontornadic environments for various storm modes. Environmental differences between right-moving supercell (RM) and quasi-linear convective system (QLCS) modes are evaluated through composite profiles and statistical testing of common forecast parameters for subsets of tornadic vs. nontornadic environments and subsets of stronger vs. weaker mesocyclone environments. We find tornadic environments for both RM and QLCS modes have significantly higher storm-relative wind (SRW) than nontornadic environments, but the largest differences in SRW are found above 1 km for RMs and below 1 km for QLCSs. However, SRW does not significantly differentiate mesocyclone strengths. Our findings of higher statistical significance in the storm-relative parameters and larger differences in the storm-relative vs. ground-relative hodographs emphasize the importance of storm motion on a storm's tornadic potential. We find tornadic RMs deviate significantly more to the right of the Bunkers RM motion estimate than nontornadic RMs. Values of many parameters that are on the low end of favorability for the Great Plains are the mean of the tornadic samples for the SE, further emphasizing the benefits of a SE convective environment climatology in helping to elucidate regional differences in environments conducive for tornadoes. Soundings derived from mesoscale and convection-allowing numerical weather prediction models are a valuable tool for assessing convective environments. Due to model assumptions, sounding profiles exhibit errors that can affect the computation of forecast parameters. In this study, soundings from the Storm Prediction Center Mesoscale Analysis System (SFCOA) and the High-Resolution Rapid Refresh (HRRR) model are compared to observed proximity soundings collected during SE field campaigns in the vicinity of RM mesocyclones and QLCS mesovortices. Prior Plains-centric work has indicated that SFCOA tends to underestimate low-level kinematic fields, whereas HRRR analyses more closely resemble observed soundings. Whether these findings are transferable to the SE remains uncertain and provides the primary motivation for this study. Relative to observations, the largest model errors are found in the low-level wind field. The HRRR generally overestimates inflow and the v-component component of the winds, while SFCOA underestimates these quantities. Despite these opposite-signed biases, both SFCOA and HRRR underestimate low-level vertical wind shear. The use of observed storm motions rather than estimated storm motions in conjunction with modeled soundings can impact the skill of storm-relative forecast parameters. An evaluation of parameter skill indicates that storm-relative wind fields are most skillful when using observed soundings and observed storm motion. In contrast, SFCOA exhibits the highest skill for ground-relative winds, bulk shear, and storm-relative helicity. These results have research implications for environmental studies that rely on SFCOA soundings and operational implications for severe weather forecasting.