Gust Front vs. Non-Gust Front Thunderstorms: Application of the Multi-Response Permutation Procedure (MRPP) to Assess the Importance of Storm Characteristics and Environmental Conditions

The implementation of a rigorous statistical technique to differentiate gust front (GF) and non-gust front (NGF) producing thunderstorms, based on storm characteristics and environmental conditions, is discussed. Fourteen single/multicell thunderstorms – nine GF and five NGF storms – from the 2002 International H2O Project (IHOP_2002), conducted over the Southern Great Plains, were examined. For storm characteristics, polarimetric Doppler radar provided radar fields from which mean time-height profiles were created including storm area. Radar measurements were then extracted from an area focused around the strongest reflectivity core; in the GF cases, this area covered some 40 min. leading up to the estimated divergence time defining the GF emergence. For environmental conditions, basic variables were derived from soundings taken at three sites within the radar's domain prior to GF emergence. For simplicity, storm characteristics and environmental conditions were analyzed separately using the Multi-Response Permutation Procedure (MRPP), which tested the null hypothesis that GF and NGF storms exhibited no differences. MRPP presented a quantitative means to determine which variables were significant in differentiating GF and NGF storms. Even more revealing was when and where these variables were significant, particularly for storm characteristics, which was based on time-height-profiles.

Results for storm characteristics revealed that the mean linear depolarization ratio (LDR) field was a strong separator between GF and NGF storms up to 40 min. prior to the estimated surface divergence time. More specifically at ~1 to 3.5 km above ground level (AGL), a mean LDR threshold of approximately -25 dB appeared to be the divider with GF storms having higher negative values relative to NGF storms. This indicated that mixed-phase precipitation at this height range, below the melting level, was significant to create GF storms. In addition, the particle identification field revealed that the majority of GF storms transitioned from light rain to graupel-rain/graupel-small hail throughout the reflectivity core while the majority of NGF storms transitioned from light rain to moderate rain for the same region. The hydrometeor, graupel, was observed more so in GF than NGF storms. It also appeared that hydrometeor type, in conjunction with mean LDR, had a tendency to separate GF and NGF storms (i.e., light rain and high mean LDR produced GF storms, light rain and low mean LDR produced NGF storms).

For environmental conditions the following parameters (ranked by importance) contributed to GF and NGF separation: lifted condensation level, Bulk Richardson number, 700 mb mixing ratio, modified lifted index, surface virtual potential temperature, convective inhibition, and convective available potential energy. For example, compared to NGF storms, the majority of GF storms had higher cloud base (2.5 km versus 2.0 km AGL), drier air aloft (3.9 g/kg versus 5.0 g/kg), and higher convective available potential energy (1022.7 J/kg versus 554.8 J/kg). These conditions are conducive to stronger, more organized thunderstorms with evaporation, due to entrainment, likely assisting in cooling the storm air appreciably and strengthening the downdraft and associated gust front. However, more thunderstorm events need to be investigated and with the use of MRPP these preliminary findings can be substantiated, which can lead to improved forecasting of GF and NGF producing thunderstorms.