The generation of zonal and eddy available potential energy (G_z and G_e) as envisioned by Lorenz (1955) are computed on a global, daily, synoptic-scale basis. Using global, mostly satellite-derived datasets for the diabatic heating components and the temperature enables us to obtain G_z and especially G_e with greater accuracy and at higher temporal and spatial resolution than previously possible. In particular, we are able to consider the contribution of each diabatic heating component separately and in combination. We use this information to determine how various processes contribute to the energy available for the general and eddy circulations.

Contributions to the global mean daily mean G_z and G _e are computed at a horizontal resolution of 2.5° for the lower troposphere (surface to 680mb), middle troposphere (680–440mb), upper troposphere (440mb to 100mb) and stratosphere (100mb to TOA) for 1997 through 2000. Comparisons to earlier estimates of the generation terms by Lorenz (1967) and Peixoto and Oort (1992) are made. The seasonal and spatial variability of the total generation and of the individual contributions of heating by radiative flux convergence, latent heating and sensible heat flux from the surface are discussed.

The generation of zonal and eddy potential energy (G_z and G_e) as envisioned by Lorenz (1955) is computed for seven climate models from the World Climate Research Programme's (WCRP's) Coupled Model Intercomparison Project phase 3 (CMIP3) multimodel dataset (Meehl et al 2007). G_z and G_e are computed directly from the temperature and diabatic heating fields of the current climate and a doubled CO₂ climate. The seasonal and spatial variability of the total generation and of the individual contributions of heating by radiative flux convergence, latent heating and sensible heat flux from the surface of each model are compared to one another, and evaluated with respect to the same quantities computed from observations. In contrast to a recent study of AMIP2 model runs by Boer and Lambert (2008), and an older study by Siegmund (1995), which found that the simulated rate of working of the models' climates were too strong compared to the observations, our results find that the data-derived generation is somewhat larger than that of the models studied here. The reasons for this are discussed in terms of the contributions to the total G from each diabatic heating component.

The response of the models' G_z and G_e to doubled CO₂ is described for the total, and for each heating component. As in an earlier study by Boer (1995), we find that in every case but one, the models exhibit smaller G_z in a warmer climate, however our results differ from Boer's in that we also observe reductions in DJF G _e in response to doubled CO₂ in five out of the seven models studied.

The models' total G_z and G_e, the contributions to each from the individual diabatic heating components, and the change in these quantities in response to increased greenhouse gases is evaluated as a function of climate sensitivity.