10.1016/0967-0653(93)94173-v

Radiation fields from a perpetual July integration of a T106 version of the ECMWF operational model are used to identify the most appropriate way to diagnose cloud radiative forcing in a general circulation model, for the purposes of intercomparison between models. Differences between the methods 1 and 2 of Cess and Potter (1987) and a variant method are addressed. Method 1 is shown to be the least robust of all methods, due to the potential uncertainties related to persistent cloudiness, length of the sampling period, and biases in retrieved clear sky quantities due to insufficient sampling of the diurnal cycle. Method 2 is proposed as an unambiguous way to produce consistent radiative diagnostics for intercomparing model results. The impact of the three methods on the derived sensitivities and cloud feedbacks following an imposed change in sea surface temperature is discussed. The sensitivity of the results to horizontal resolution is considered by using the diagnostics from parallel integrations with T21 version of the model. The concept of cloud radiative forcing was first discussed in the open literature by Coakley and Baldwin [1984] and was first used by Ramanathan [1987] to identify the impact of clouds on the radiation budget at the top of the atmosphere. It may be defined as the difference between the radiative flux which actually occurs with cloudiness and that which occurs for clear skies. The change in cloud radiative forcing which accompanies a change in climate is known as cloud feedback. In a recent study by Cess et al. [1989, 1990] of the response of 19 atmospheric general circulation models (GCMs) to an imposed change in sea surface temperature (used as a surrogate climate change) an almost threefold variation in the cloud feedback from weakly negative to strongly positive was obtained. This led Cess et al. to conclude that cloud-climate feedback could be a significant cause of intermodel differences in climate change projections. In subsequent analysis of the individual modeling results it has become apparent that there are a number of different approaches used in the computation of cloud radiative forcing. In terms of the energy lost or gained by the Earthatmosphere system, cloud forcing (CRF) can be defined as CRF = Fclea r - Ftota 1 q- Qtotal- Qclear (1) where F and Q are, respectively, the emitted infrared and net downward solar fluxes at the top of the atmosphere (TEA). The concept of cloud radiative forcing was originally introduced with satellite data because it allowed the impact of clouds on the TeA radiation budget and therefore on the