Validation of General Circulation Model Radiative Fluxes Using Surface Observations

The surface radiative fluxes of the ECHAM3 General Circulation Model (GCM) with T2 1, T42, and T 106 resolutions have been validated using observations from the Global Energy Balance Archive (GEBA, World Climate Program-Water Project A7). GEBA contains the most comprehensive dataset now available for worldwide instrumentally measured surface energy fluxes. The GCM incoming shortwave radiation at the surface has been compared with more than 700 long-term monitoring stations. The ECHAM3 models show a clear tendency to overestimate the global annual-mean incoming shortwave radiation at the surface due to an underestimation of atmospheric absorption. The model-calculated global-mean surface shortwave absorption around 165 W m−2 is estimated to be too high by 10–15 W m−2. A similar or higher overestimate is present in several other CYCMS. Deficiencies in the clear-sky absorption of the ECHAM3 radiation scheme are proposed as a contributor to the flux discrepancies. A stand-alone validation of the radiation scheme under clear-sky conditions revealed overestimates of up to 50 W m−2 for daily maximum values of incoming shortwave fuxes. Further, the lack of shortwave absorption by the model clouds is suggested to contribute to the overestimated surface shortwave radiation. There are indications that the incoming longwave radiation at the surface is underestimated in ECHAM3 and other GCMS. This largely offsets the overestimated shortwave flux in the global mean, so that the 102 W m-’ calculated in ECHAM3 for the surface net radiation is considered to be a realistic value. A common feature of several GCMs is, therefore, a superficially correct simulation of global mean net radiation, as the overestimate in the shortwave balance is compensated by an underestimate in the longwave balance. Seasonal and zonal analyses show that the largest overestimate in the incoming shortwave radiation of ECHAM3 is found at low latitudes year round and in midlatitude summer, while at high latitudes and in midlatitude winter the solar input is underestimated. As a result, the meridional gradient of incoming shortwave radiation becomes too large. The zonal discrepancies of the duxes are consistent with differences between the simulated cloud amount and a cloud climatology based on surface observations. The shortwave discrepancies are further visible in the net radiation where the differences show a similar latitudinal dependency including the too strong meridional gradient. On the global and zonal scale, the simulated fuxes are rather insensitive to changes in horizontal resolution. The systematic large-scale model deviations dominate the effects of increased horizontal resolution.