Convection in Lorenz's global energy cycle with the ECMWF model

Lorenz's global energy cycle includes the conversion rate C between available potential and kinetic energy. In traditional estimates of C only gridscale processes were evaluated; subgridscale processes were lumped into dissipation. It is argued that this is inadequate; organized subgridscale heat fluxes like deep convection cannot be treated as molecular. Here both C-grid and C-sub are evaluated from the ECMWF Integrated Forecast System, for a 1-yr forecast in climate mode. The subgridscale fluxes are obtained from the model parametrization and the results tested for consistency; the largest contribution comes from the convection scheme. The integrand of C-sub, the familiar 'buoyancy flux' -(alpha'omega') over bar, is locally much smaller than its gridscale counterpart -(alpha) over bar(omega) over bar .However, the buoyancy flux is upward throughout, and thus representative for, the global atmosphere. The global annual means are C-grid = ( 3.4 +/- 0.1) W m(-2) and C-sub = ( 1.7 +/- 0.1) W m(-2). Further, the gridscale generation rate of available potential energy is evaluated independently and found to be G(grid) = ( 3.0 +/- 0.2) W m(-2). These results suggest that (i) the subgridscale processes contribute significantly to the Lorenz energy cycle and (ii) the cycle, represented by the total dissipation of D = ( 5.1 +/- 0.2) W m(-2), is more intense than all earlier gridscale estimates have indicated.