Brian E. Mapes
Cooperative Institute for Research in the Environmental Sciences
Suppose deep convective clouds represent the free buoyant ascent of low-level air, once it is brought to its level of free convection (LFC) by a low-level "activation" process energetic enough to overcome the convecive inhibition energy (CIN). Large-scale enhancements of convection should therefore be attributed to processes such as low-level adiabatic dynamic lifting; enhanced frequency of occurrence of strong activation mechanisms such as gust fronts; enhancements of the warmest, moistest boundary-layer air; and decreased mixing of dry air into updrafts. Quantifying these bulk sensitivities - which involve (intercorrelated) subgridscale distributions as well as mean values - seems necessary for accurate, physically- based parameterization.
Differences of the right sign are present in tropical aircraft data and enhanced-suppressed sounding composite differences, but actual values of CIN, LFC, etc. are quite delicately sensitive to unresolvable details of entrainment, precipitation, parcel definition, and vertical resolution. Still, a model would presumably develop its own internally-adjusted (and tunable) economy of inhibition & activation, and would perhaps yield better simulations for having appropriate sensitivity to low-level inversions. Its climatology needs only be wrong by the amount of these delicate quantities, making their smallness a blessing in disguise.