Effects of a convective organization scheme in CAM5

We implemented a prognostic scheme for convective organization in CAM5, in the context of a two-plume version of the Park-Bretherton (PB) updraft plume convection scheme. The PB scheme is often called a shallow convection scheme, because it operates alongside a separate deep convection scheme, but we have disabled that deep scheme and instead added a second, less-entraining plume to the PB scheme. "Organization" is defined in this context as a linear propensity for the second plume computation to be given a lower environmental mixing rate and a larger cloud base mass flux. Sources of organization include recent precipitation re-evaporation and coastlines (0.1 < land fraction <0.9 within a grid cell). The effects of the scheme on model climate and variability have some expected aspects, and some surprises. It fulfils its design purpose to evade the "entrainment dilemma," an unwanted tradeoff between thermodynamic biases vs. spatial gradients and time variations. But it fails to significantly delay the diurnal cycle of precipitation over land, as might have been hoped. Most surprisingly, the coastal organization source actually reduces mean precipitation over the Maritime Continent regions where coastlines are ubiquitous, although it increases precipitation over more linear coasts elsewhere. The scheme has also been tested in initialized CAPT forecasts for evaluation, although not with a clinching control case to compare to. Prospects for such schemes, and the broader need for organization effects to be treated somehow, will be reviewed in light of these results.