It is well known that the roots of cumulus convection reside in strong convective plumes that originate from the surface layer. Yet, in large scale weather and climate models it is still common practice to parameterize the dry boundary layer and the cloud layer separately. This results in unwanted overlaps between the schemes and double counting of the transport mechanisms. It also has the additional disadvantage that it requires extra boundary conditions for the convective fluxes at cloud base and a seperate decision making process for the onset of cumulus convection, sometimes referred at as the triggering function.

To overcome these problems a updraft parameterization for strong thermals in the subcloud layer has been formulated. Lower boundary conditions can easily formulated using surface layer similarity theory. By subsequently solving updraft equations for vertical velocity, temperature and moisture, this parameterization decides when cumulus convection is initiated, and provides estimates for cloud base height, cloud base temperature and cloud base vertical velocity. Results of an implementation of this scheme in the ECMWF model will be presented and discussed.