Bulk Microphysical Variability of Thunderstorms in Different Climatic Regions: Comparative Predictive Skills of Melting Level, Cloud Base Temperature and Cloud Base Pressure in a Three-Dimensional Numerical Modeling Study

We present here some further results of a previously reported simulation-based study of case-to-case dynamical and microphysical variability among thunderstorms in diverse climatic zones around the world as well as within a given climate zone, analyzing a total of 105 storm cases. The simulation in question is the Wisconsin Dynamical/Microphysical Model (WISCDYMM), a quasi-compressible nonhydrostatic three-dimensional cloud model equipped with bulk microphysics for cloud water, rain, cloud ice, snow and graupel/hail.

Our previous report, upon which we build here, compared case-to-case the partitioning of total hydrometeor mass among the five individual classes, and between total ice and liquid masses, domain-integrated and time-averaged over a significant portion of the storms' mature stages. Via linear regression, we evaluated to what extent these bulk microphysical storm properties correlate with several pre-storm environmental indices relevant to severe local storm prediction, in particular the ground-relative melting level (ZMLT) both alone and jointly with the convective available potential energy (CAPE). Versus ZMLT alone, both the total and cloud ice fractions showed fair correlations and the rain fraction performed somewhat better; while versus ZMLT and CAPE jointly, these correlations were boosted considerably.

Our further investigation builds on these findings by applying the same linear regression analysis to two other choices of primary predictor, the lifting condensation level temperature (TLCL) and lifting condensation level pressure (PLCL), and comparing our findings for those described above for ZMLT. The two most notable new findings are as follows:

a) Except when predicting hail, TLCL as sole predictor produces considerably stronger correlations than ZMLT, especially for snow. Jointly with CAPE, TLCL produces a dramatic improvement versus TLCL alone for each predictand, and also does considerably better than ZMLT except only slightly so for hail. The bivariate correlation magnitudes are particularly strong for total ice, cloud ice and especially rain;

b) PLCL as sole predictor performs still better than TLCL for the predictands other than rain and snow, albeit only modestly so except more dramatically for cloud water, the weakest of the predictands. However, adding CAPE as a joint predictor with PLCL produces far smaller correlation enhancements than it does for either ZMLT or TLCL, and virtually none when predicting either rain or snow.