Examining the Sensitivity of Horizontal and Vertical Grid Spacing on Simulations of Cool Season Severe Thunderstorms in the Southeast United States

Previous studies have focused storm-scale simulations on a variety of thermodynamic and kinematic environments supporting severe convection. Many findings related to intense convection evolving in at least moderately buoyant environments have broadened our understanding of supercell and mesoscale convective system (MCS) dynamics. However, the relative emphasis on severe weather events occurring in environments characterized by substantial deep-layer shear (at least 20 m s^-1 of 0-6-km bulk shear) and marginal convective available potential energy (CAPE; less than 1000 J kg^-1 for a mixed-layer parcel) has been limited. Such environments are common during the cool season in the southeast United States and serve as a challenge to forecasters given the limited predictability of severe storms forming in these environments owing to the marginality of CAPE. This study is intended to expand our understanding of cool season severe weather in the southeast United States by examining the effects of varying horizontal and vertical grid spacing in simulations of storms in high-shear, low CAPE environments.

This study varies horizontal resolution using the Advanced Research Weather Research and Forecasting (ARW) model at four grid spacings: Δx = 10, 4, 2, and 1 km, with all other model configuration parameters held constant. Simulations with varying vertical grid spacing are also examined, with the highest vertical resolution maintained at lower levels. The results reveal sensitivities to variations in horizontal resolution, with smaller grid spacing yielding more accurate simulations. Particular emphasis is placed on simulations of a low-CAPE, high-shear severe weather event that occurred in Mississippi on 25 March 2009. Reducing the horizontal grid spacing results in clear differences amongst the simulated convective mode: Δx = 10 km results in an intense uniform squall line, Δx = 4 km depicts mesoscale bowing segments within a broader squall line, and Δx = 2 km yields an isolated pocket of increased dBZ and updraft helicity characteristic of a supercell ahead of a squall line (i.e., the most accurate simulation) while also highlighting localized 1-6 km absolute helicity maxima ahead of the line. Further computations to examine model sensitivity using even smaller grid spacing will be addressed, along with optimal resolutions required for simulating supercell structures in these environments. These sensitivity tests are intended to provide guidance in maximizing computing efficiency while also maintaining accuracy in simulating storms in volatile environments present during the southeast United States cool season.