THE DISPOSITION OF RADIATIVE ENERGY IN THE GLOBAL CLIMATE SYSTEM - GCM-CALCULATED VERSUS OBSERVATIONAL ESTIMATES

A comprehensive dataset of direct observations is used to assess the r epresentation of surface and atmospheric radiation budgets in general circulation models (GCMs). Based on combined measurements of surface a nd collocated top-of-the-atmosphere fluxes at more than 700 sites, a l ack of absorption of solar radiation within the atmosphere is identifi ed in the ECHAM3 GCM, indicating that the shortwave atmospheric absorp tion calculated in the current generation of GCMs, typically between 6 0 and 70 W m(-2), is too low by 10-20 W m(-2) The surface and atmosphe ric radiation budgets of a new version of the Max-Planck Institute GCM , the ECHAM4, differ considerably from other GCMs in both short- and l ongwave ranges. The amount of solar radiation absorbed in the atmosphe re (90 W m(-2)) is substantially larger than typically found in curren t GCMs, resulting in a lower absorption at the surface (147 W m(-2)). It is shown that this revised disposition of solar energy within the c limate system generally reduces the biases compared to the observation al estimates of surface and atmospheric absorption. The enhanced short wave absorption in the ECHAM4 atmosphere is due to an increase in both simulated clear-sky and cloud absorption compared to ECHAM3. The incr eased absorption in the cloud-free atmosphere is related to an enhance d absorption of water vapor, and is supported in stand-alone compariso ns of the radiation scheme with synchronous observations. The increase d cloud absorption, on the other hand, is shown to be predominantly sp urious due to the coarse spectral resolution of the ECHAM4 radiation c ode, thus providing no physical explanation for the ''anomalous cloud absorption'' phenomenon. Quantitatively, however, an additional increa se of atmospheric absorption due to clouds as in ECHAM4 is, at least a t low latitudes, not in conflict with the observational estimates, tho ugh this does not rule out the possibility that other effects, such as highly absorbing aerosols, could equally contribute to close the gap between models and observations. At higher latitudes, however, the inc reased cloud absorption is not supported by the observational dataset. Overall, this study points out that not only the clouds, but also the cloud-free atmosphere might be responsible for the discrepancies betw een observational and simulated estimates of shortwave atmospheric abs orption. The smaller absorption of solar radiation at the surface in E CHAM4 is compensated by an increased downward longwave flux (344 W m(- 2)), which is larger than in other GCMs. The enhanced downward longwav e flux is supported by surface measurements and by a stand-alone valid ation of the radiation scheme for clear-sky conditions. The enhanced f lux also ensures that a sufficient amount of energy is available at th e surface to maintain a realistic intensity of the global hydrological cycle. In contrast, a one-handed revision of only the shortwave radia tion budget to account for the increased shortwave absorption in GCM a tmospheres may induce a global hydrological cycle that is too weak.