Abstract
Heating in the atmosphere can be expressed as diabatic divergence δd, which is nearly equal to the actual horizontal divergence δ in tropical convection. High-quality δ profile measurements from airborne Doppler radar “purls” in ten mesoscale convective systems (MCS) observed during TOGA-COARE are examined, and the mean profile is compared with rawinsonde array measurements. Young convective features have strong near-surface convergence, while older cells with better-developed downdrafts and stratiform precipitation areas have their peak convergence aloft. In the mean, then, surface flow is only weakly convergent or oven divergent, so that the main convergence into MCSs is deep and peaked aloft, with a sharp “melting convergence” at 0°C. Divergence prevails above ∼10 km altitude but was undersampled by the radar.
Unusual but well-sampled observations in the purl dataset include: a persistent, widespread δ profile feature in one well-sampled MCS (a cyclone rainband); oscillatory “reverberations” centered on the melting level, with ∼3–4 km wavelength in the vertical; and a conspicuous absence of any high vertical wavenumber features other than the melting reverberations.
All three observations may be understood as consequences of the heating profile of convection adjusting itself to oppose environmental temperature perturbations. This adjustment is predicted by convective cloud conceptual models with diverse dynamical bases, and consequently is simulated by essentially all convective parameterization schemes. One foreseeable consequence of this mechanism is the downward development of initially elevated (cool core) depressions, a key stage in tropical cyclogenesis.
Simple linear models of Hadley and Walker circulations forced by observed MCS δd profiles illustrate the importance of the elevated convergence peak to large-scale circulations, particularly to low-level wind fields.
