Parameterization of convection has focused on the ensemble mean
effects of subgrid scale convective elements as a deterministic
function of the large-scale variables. The effect of the
distribution of convective heating (and related quantities)
about the ensemble mean has traditionally been neglected.
These can be included by representing the convective heating
as a random process whose distribution is parameterized on the
large scale. Some initial explorations are presented that
help to illustrate impacts this may have on the large-scale
flow, parameters that may be important and some of the
challenges that may arise.