Abstract
Although the steady, entraining, updraft plume is widely taken as the foundational concept of cumulus convection, past studies show that convection is typically dominated by thermals that are transient, more isotropic in shape, and possess interior vortical circulations. Here, several thousand such thermals are tracked in cloud-resolving simulations of transient growing convective events. Most tracked thermals are small (with radius R < 300 m), ascend at moderate rates (~ 2–4 m s−1), maintain an approximately constant size as they rise, and have brief (4–5 min) lifetimes, although a few are much larger, faster, and/or longer lived. They show slight vertical elongation, but few, if any, would be described as plumes. As convection deepens, thermals originate higher up, are larger, and rise faster, although radius and ascent rate are only weakly correlated among individual thermals. The main force opposing buoyancy is a nonhydrostatic pressure drag, not mixing of momentum. This drag can be expressed in terms of a drag coefficient cd that decreases as convection intensifies: deep convective thermals are less damped, with cd ~ 0.2, while shallow convective thermals are more damped, with cd ~ 0.6. The expected dependence of cd based on theoretical form and wave drag coefficients for a solid sphere is inconsistent with these results, since it predicts the opposite dependence on the Froude number. Thus, a theory for drag on cumulus thermals is not straightforward. Overall, it is argued that thermals are a more realistic prototype for atmospheric deep convection than plumes, at least for the less organized convection types simulated here.
