Abstract
What processes control large-scale variations of deep convection (LSVDC) in the tropics? Here ‘large-scale’ is taken to mean any coherent variations, in either space or time, comprised of statistical populations of separate convective cloud systems. This essay highlights the distinction between processes which supply moisture or available energy over the depth of the convecting layer (equilibrium control), versus inhibition and initiation processes at low levels (activation control), as hypotheses for explaining LSVDC.
Conceptual separations of the LSVDC problem are reviewed. Scale separation, though rigorous, is artificial, since net heating makes deep convective clouds multiscale, or spectrally red. Moist-dry, or diabatic-adiabatic, separation is more useful. An ill-posed hybrid separation — ‘the interaction of moist convection with large scales’ — has spawned confusion. Correlations between deep convection and its own large-scale components (suggestively labeled ‘forcing’) have been misinterpreted as evidence for equilibrium control. This externalization of large-scale vertical velocity also encourages overinterpretation of a fictitious ‘compensating subsidence’ term.
Published evidence for equilibrium control theories is critically re-examined. The deep-cloud quasi-equilibrium observations of Arakawa and Schubert would hold for arbitrarily determined variations of convection, because stratified fluid dynamics efficiently redistributes localized heating, not because convection is controlled by slow deep large-scale ‘forcing.’ A more sensitive test indicates that most tropical LSVDC are not forced by preexisting deep upward motions.
Activation-control processes can operate on the large space scales and long time scales that define LSVDC. Modulation of convection by easterly waves and upper-tropospheric troughs, and the climatological distribution of convective cloudiness, are examined as examples. Lower boundary flux enhancements and deep lifting exert both equilibrium and activation influences on convection. The hypothesis that activation control may prevail on all scales short of globally-averaged climate is difficult to refute.
Systematic study of convective cloud-ensemble sensitivities is badly needed. Unfortunately, current cloud ensemble modeling strategies, based on unnatural equilibrium-control assumptions, render the results merely diagnostic.
To represent activation controls, large-scale models need multiple levels, and some prognostic representation of the subgridscale inhomogeneity of low-level fields.