Abstract
The General Circulation Model (GCM) of the Goddard Laboratory for Atmospheric Sciences (GLAS) was integrated for 107 days starting from the initial conditions of 15 May. In this experiment the clouds dynamically generated by the model affect the radiative heating fields continuously. Starting from the initial conditions valid for day 76 of this run, another integration was made for 31 days in which the clouds were specified on certain grid points, remaining fixed during the period of integration. The spatial distribution of the fixed clouds was such that the aggregate cloud frequency for a vertical level and each latitude circle remained the same as in each control run, and the highest cloud frequency grid points were assigned the cloudiness of 100%. The 31-day mean simulation of the second run (fixed clouds) is compared with the last 31-day mean simulation of the first run to study the effects of cloud-radiation feed- back on the mean monthly circulation, atmospheric energy cycle and the hydrological cycle, evaporation and precipitation and the local climate.
Results from these experiments show significant changes in the simulated large-scale dynamical circulation of the global model. Fixed clouds acting as zonally asymmetric radiative heat sources increase the generation of eddy available potential energy (EAPE) and its conversion to eddy kinetic energy. Generation of EAPE by net radiative heating increased by 50%(0.11 W m−2) for the fixed cloud experiment. The increase due to the stationary component was much larger (∼100%) but it was partially compensated by decrease due to the transient component. A substantial increase was found in the variances of the planetary-scale stationary waves and the medium-scale waves (wavenumber 6–10) of 2–7 day period. Although the sea surface temperatures were prescribed identically in both integrations, the changes in evaporation and precipitation were found to be much larger over the oceans compared to those over the land. We suspect that this happens because the ground temperature is determined by the model's beat balance at the earth's surface and therefore internal model feedbacks do not allow the hydrologic cycle over land to be very different between the fixed cloud run and the control run. Based on these calculations, we infer that cloud-radiation feedback is an important mechanism in the general circulation of the model atmosphere. It must be adequately parameterized in numerical experiments designed to simulate the mean climate and/or to examine the sensitivity of GCM's to changes in external boundary conditions or internal atmospheric constituents (such as aerosols and CO2) and their feedback effects.
