The global conversion rates into available potential and kinetic energy (Lorenz's quantities G and C) have been traditionally evaluated on the gridscale (G(grid), C-grid approximate to +2.5 +/- 0.4 W/m(2)). Convective phenomena acting on the sub-gridscale (like, e.g., thunderstorms) have been treated as molecular. In Part I of this study it has been outlined how Lorenz's energy cycle may be extended to include sub-gridscale processes. For this purpose new fluxes, particularly the global mean conversion rate into kinetic energy on the sub-gridscale (C-sub), have been defined. Evaluating them is the purpose of the present Part II. C-sub is closely related to the buoyancy production term of turbulence kinetic energy which can be expressed through the vertical sub-gridscale fluxes of moisture and heat. A thermodynamic diagnostic model (DIAMOD) that estimates these fluxes indirectly from gridscale analyses is applied. In this way the conversion rate has been calculated for three months using global reanalysis data from ECMWF and from NCEP/NCAR. The errors of our results are caused by the analysis data used, by the specification of the ratio between moisture and heat fluxes (the main closure assumption in DIAMOD) and by uncertainties in the radiative heating field; they are given here at the 95% level. We find C-sub = + 2.2 +/- 1.7 W/m(2). The new complete conversion rate C = C-grid + C-sub is + 4.7 +/- 2.0 W/m(2). This figure is the main result of this study, presented here for the first time: Lorenz's energy cycle, if extended to the sub-gridscale, is about twice as intense as in the traditional approximations. In contrast to C-sub the sub-gridscale generation rate G(sub) and therefore the complete G cannot be evaluated. All one can do is to improve the estimate of G(grid) by improving the estimates of the net heating. For G(grid) We find the new value of + 3.1 +/- 0.5 W/m(2).
