Abstract
An efficient longwave scheme for climate models originally suggested by Sasamori is modified to correctly simulate the water vapor-temperature feedback mechanism. It is found that the modified scheme with a fixed cloud-top altitude (FCA) correctly simulates the longwave sensitivity to surface temperature, ∂F†/∂Ts, over the clear portion of the sky, but that over the cloud portion of the sky ∂F†/∂T<s remains too high. The fixed cloud-top temperature (FCT) method is similarly reviewed and tested. Comparisons with observational Budyko-type correlations are shown to be indecisive over the question of FCA vs FCT.
A direct observational correlation between the effective cloud-top and surface temperatures, based on annually averaged cloud statistics data, suggests a variable cloud-top temperature (VCT) model. In such a model, the temperature of the elective cloud-top layer is varied according to changes in the surface temperature at a rate which is intermediate between that of the FCA and FCT models. This model results in a reasonable ∂F†/∂Ts over the cloud portion of the sky.
The modified longwave scheme is implemented into a zonally averaged dynamic climate model. It is shown that when the VCT mechanism is invoked, climate sensitivity is doubled compared to that simulated with the FCA model. The importance of simulating not only the correct longwave flux, but also the correct longwave sensitivity to temperature changes is therefore stressed for radiation schemes in studies involved with climate change.
