A convective cloud's internal thermodynamic variance plays a critical role in its evolution.  Turbulence generates a spectrum of buoyancies via spatially heterogeneous entrainment.  In turn, this spectrum of buoyancies determines the spectrum of levels of neutral buoyancy (LNBs) for the cloud's constituent parcels.  Those LNBs then determine the profile of the cloud's mass flux: each parcel within the cloud detrains at its respective LNB, which is determined by the history of stochastic entrainment events experienced by that parcel.  The Stochastic Parcel Model (SPM) is a convective parameterization designed to capture this physics.  While originally formulated in a computationally expensive Lagrangian framework, the SPM is being redesigned for computational efficiency and eventual implementation in CAM under the SciDAC Multiscale project.