Cold pools – a first step in representing convective organization in GCMs

Poster PDF




Anthony D. Del Genio — National Aeronautics and Space Administration Jingbo Wu — Columbia University
Audrey B. Wolf — NASA - Goddard Institute for Space Studies


Most global climate models (GCMs) represent moist convection as an ensemble of convective cells determined by the instantaneous grid-scale environment. Most precipitation and cloud forcing in convective environments, however, comes from organized mesoscale clusters with life cycles that span many hours and a heating profile that shifts from full troposphere heating to a dipole structure of upper/lower level heating/cooling as the stratiform rain region develops. Parameterizing the organization of convection is a multi-step process:
  1. Initiation of convective downdrafts that inject cold air into the boundary layer
  2. Formation of cold pools that spread over time and remain distinct from ambient air
  3. Regeneration of convection by lifting of ambient air at the gust front
  4. Initiation of a stratiform cloud and rain region by detrained convective condensate
  5. Development and evolution of mesoscale updrafts and downdrafts.
We will describe initial efforts to account for the first three steps of convective organization in the GISS GCM. Cold pools are assumed to form when downdraft air with virtual potential temperature colder than that of ambient air enters the boundary layer. The cold pool spreads at a rate determined by the virtual potential temperature contrast, while its depth evolves according to the competition between spreading and further addition of downdraft air. Evolution of cold pool thermodynamic properties is determined by the properties of subsequent downdrafts, offset by relaxation to ambient conditions over a specified time scale. After the cold pool forms, further convection triggering utilizes properties of the undisturbed boundary layer and assumes parcel lifting up to the cold pool top.

Initial GCM tests indicate that parameterized cold pools are ~1–2 km deep and are deeper over land than over ocean; have cold anomalies of ~0.5–2 K or more, with the largest values over drier continental areas; and usually have dry anomalies of ~0–1.5 g/kg, although cold pools at the edges of the continental Intertropical Convergence Zone (ITCZ) tend to be slightly wetter than ambient air. Although these characteristics qualitatively resemble those seen in field experiments and cloud-resolving models (CRMs), the frequency of occurrence (~0.5% globally) and duration (2 hours or less) of cold pools is not sufficient to shift the phase of the diurnal cycle of convection. We are performing SCM tests of the Midlatitude Continental Convective Clouds Experiment (MC3E) field campaign data to try to understand how this limitation of the parameterization might be mitigated.