"Deconstruction" of the mass-flux convective parameterization problem

35 years have passed since the work by Ooyama and by Arakawa and Schubert that established the basic formulation for the mass-flux convective parameterization. Since then a growing number of both global and regional models aiming operational forecasts, climate research, as well as basic research adopt a mass-flux parameterization. Surprisingly, the basic formulation originally proposed by Ooyama and by Arakawa and Schubert has hardly changed over years in spite of extensive elaborations of schemes. Worse than that, these physical elaborations introduced tend to ad hoc.

Though the inclusion of downdrafts may nowadays be considered to be standard, they remain at odd with the basic formulation with mass fluxes. In spite of its importance, not many schemes include mesoscale downdrafts. Cloud physics are harder to implement consistently into the mass-flux framework: then the convective vertical velocity must be explicitly evaluated within a scheme but only at odd with an asymptotic limit of vanishing fractional convective area already introduced.

Many of the fundamental issues are still left unresolved. The closure is the most notorious example. In spite of extensive elaborations of entrainment-detrainment processes over years, all these improvements are rather phenomenological, and a physically-consistent theory for determining the entrainment-detrainment parameters are still unknown.

The purpose of the present talk is to critically re-examine the formulational structure of the mass-flux convective parameterization with these open questions in mind. A "deconstruction" of the formulation is performed starting from a full physical model (e.g., cloud-resolving model). A hierarchy of representation schemes for subgrid-scale convective processes is identified as a result of the stepwise "deconstruction".

Results with a model named NAM-SCA, resulting from a first major step of "deconstruction" are also presented.