Quarterly Journal of the Royal Meteorological Society, doi:10.1002/qj.967.
Centre for Australian Weather and Climate Research, Bureau of Meteorology, Melbourne, Australia.
Adam H. Sobel
Department of Applied Physics and Applied Mathematics, Department of Earth and Environmental Sciences, and Lamont-Doherty Earth Observatory Columbia University, New York, NY.
A single column model (SCM) version of the HadGEM1 is run in weak temperature
gradient (WTG) mode, assuming a free-tropospheric temperature profile obtained from the
same single column model in radiative-convective equilibrium (RCE) over a sea surface
temperature (SST) of 301K. The resulting quasi-steady solutions are compared with climate
statistics from time-dependent solutions of the full 3D atmospheric General Circulation
Model (GCM) sharing the same physics, with the aim of evaluating the strengths and
weaknesses of the WTG parameterization of large-scale dynamics.
Making some allowance for unavoidable differences between quasi-steady 1D solutions and more fully transient 3D solutions, the two models produce grossly similar sensitivities of precipitation and relative humidity to local sea surface temperature. The greatest differences arise from the relatively sharp transition in the SCM between very dry and rainy states as SST is varied, while the GCM statistics vary more smoothly with SST. When a relaxation on the moisture field towards a target profile - a crude parameterization of horizontal moisture advection - is included in the SCM, this difference is reduced. The SCM is then able to produce some convection at low SST, and the increases in humidity and precipitation with SST become more gradual, as in the GCM.
The RCE temperature profile used to obtain these results is colder in the upper troposphere and thus more unstable to deep convection than is the climatological tropical profile from the GCM. When the latter is used in the SCM, the precipitation as a function of SST does not change greatly, but the convection becomes considerably shallower than that in either the GCM, or the SCM with the RCE temperature profile. We speculate that some of these differences may be due to the much greater transience in the GCM solutions compared to the SCM's quasi-steady states.