Buoyancy reversal and cloud‐top entrainment instability

Steven T. Siems, Christopher S. Bretherton, Marcia B. Baker, Shenqyang Shy, Robert E. Breidenthal

Research output: Contribution to journalArticlepeer-review

80 Scopus citations


In 1980, Randall suggested that cloud top entrainment instability, a runaway positive feedback between mixing, evaporative cooling and entrainment, may cause the transition from the subtropical stratocumulus layers, frequently observed over the cold water of the eastern oceans, to tradewind cumulus. Predicting how the mean position of this transition, and the associated large change in albedo, would vary with, and feed back on, a change in climate is important for any credible model of the climate system. We have used numerical and laboratory simulations, as well as field observations, to study entrainment in buoyancy‐reversing systems. We consider a system in which an upper fluid stably overlies a second lower fluid; but some mixtures are less buoyant than either fluid. the laboratory results show that buoyancy reversal does not cause runaway ‘explosive entrainment’ unless D (the ratio of the density difference between the most cooled mixed parcel and the lower fluid to the density difference across the unmixed layers, a measure of CEI potential) exceeds 1.3, which is far greater than the range of D observed in stratocumulus. For D below this threshold, buoyancy reversal leads to weak circulations rather reminiscent of Rayleigh‐Bénard convection in the lower layer, but only slightly distorts the interface. For D > 1.3 an entrainment tongue of upper fluid penetrates a sizable fraction of the depth of the lower layer and entrainment is explosively rapid. Numerical experiments suggest that even when 0 < D < 1.3, entrainment can break up idealized subtropical marine stratocumulus within about an hour if radiation and surface fluxes are removed, but comparison of numerical simulations with the laboratory experiments suggests that the numerically simulated entrainment rate may be substantially too large in this regime owing to the inability to accurately produce and resolve the entraining eddies. Our observations of stratocumulus from FIRE typically had a strong stable layer above the cloud top inversions, leading to some ambiguity in defining D, but clearly show that stable stratocumulus layers often have 0 < D < 0.2. We suggest that sudden breakup of stratocumulus due to runaway entrainment is unlikely, but evaporatively enhanced entrainment may aid the transition from stratocumulus to tradewind cumulus in conjunction with some other mechanism, such as short‐wave absorption by liquid water, that can help dynamically decouple the cloud layer from the surface layer.

Original languageEnglish
Pages (from-to)705-739
Number of pages35
JournalQuarterly Journal of the Royal Meteorological Society
Issue number493
StatePublished - Apr 1990


Dive into the research topics of 'Buoyancy reversal and cloud‐top entrainment instability'. Together they form a unique fingerprint.

Cite this