Ensemble Sensitivity Analysis of Controls on Storm-Scale Vertical Vorticity for Two Southeastern U.S. Tornado Events

Sensitivity analysis is a useful technique for diagnosing how small perturbations in initial conditions can influence the forecast of the future state. Recently, ensemble sensitivity analysis (ESA) began to be used for such purpose, where the linear component of the relationship between a forecast metric (e.g., updraft helicity, maximum reflectivity) and the initial conditions of the simulation could be quantified. With such knowledge, identified areas of sensitivity can be adaptively sampled with surface and upper-air observations to reduce the forecast error variance.

Much of the existing work using ESA has involved larger time and space scales, where the assumptions of linearity inherent to ESA are relatively appropriate (e.g., Ancell and Hakim 2007; Torn and Hakim 2008). Ongoing work has successfully applied this technique to the mesoscale to assess how heterogeneities in initial conditions influence the development of deep convection along the Southern Plains dryline (e.g., Hill et al. 2016).

This paper will explore the utility of ESA on the mesoscale and storm scale, to understand the specific environmental and storm-generated controls on low-level vertical vorticity in simulations of storms from two cases over the southeastern United States: the 27 April 2011 tornado outbreak across Mississippi and Alabama, and the 5 April 2017 case from the Verification of the Origins of Rotation in Tornadoes Experiment – Southeast (VORTEX-SE).

Ongoing work of the 27 April 2011 case suggests that 2-5 km updraft helicity (a combination of updraft vertical velocity and vertical vorticity) of the Tuscaloosa/Birmingham storm is sensitive to a number of controls encompassing both the inflow environment (e.g., virtual potential temperature of air parcel in the far-field inflow) and storm-generated baroclinity, particularly on the left flank of the parent supercell thunderstorm where recent tornadogenesis research has had a considerable focus. These results will be shown, in addition to exploration of other response metrics such as low-level circulation. Common sensitivity signals across multiple analysis times, and between the two cases, will be highlighted, permitting guidance on potential strategies for adaptive sampling of severe storms (and environments) to improve prediction of specific hazards.