How does rising atmospheric CO2 affect marine organisms?

Click to locate material archived on our website by topic


Shortcomings of GCM Cloud and Radiation Parameterizations
Reference
Lane, D.E., Somerville, R.C.J. and Iacobellis, S.F.  2000.  Sensitivity of cloud and radiation parameterizations to changes in vertical resolution.  Journal of Climate 13: 915-922.

Background
A frequent criticism of present-day general circulation models (GCMs) is their lack of resolution on both the horizontal and vertical scales.  Lack of adequate resolution forces modelers to parameterize or simulate the ensemble large-scale effects of processes that occur on smaller scales than the horizontal or vertical extent of a model's grid size.  Such is the case when it comes to representing physical processes such as cloud formation and cloud-radiation interactions.  The question naturally arises, therefore, as to what vertical resolution is adequate for the treatment of such interactions and processes.  All of the atmospheric GCMs used for the transient coupled atmosphere-ocean GCM experiments described in the 1995 Intergovernmental Panel on Climate Change Report (Houghton et al., 1996), for example, contained 19 or fewer layers.  Is that number sufficient or are more needed?

What was done
The authors investigated the performance of a suite of cloud-radiation parameterizations typically found in contemporary GCMs to test the sensitivity of these parameterizations as a function of vertical model resolution, varying the resolution from 16 to 60 layers in increments of four and comparing them to observed values.

What was learned
Simulated values of cloud-radiation variables showed a marked sensitivity to changes in vertical resolution.  Cloud fraction varied by about 10% over the range of resolutions tested, which corresponded to about 20% of the observed fraction of cloud cover.  Outgoing longwave radiation varied by 10 or 20 Wm-2 as the resolution was varied, amounting to around 5 to 10% of the observed value.  Incoming solar radiation also experienced significant variations across the range of resolutions.  Furthermore, the model results did not converge even at a resolution of 60 layers, and there were significant systematic differences between model results and observations.

What it means
It is clear from the results of this study that GCMs still have a long way to go before realistic simulations of climate can be made; for significant errors in cloud-radiation interactions still persist, even when a vertical resolution three times greater than that used in present-day models is employed.  In the words of the authors, these tremendous inadequacies demand future research "to produce parameterizations that are physically more realistic and numerically better behaved than those of present-day climate models."  Until that time, we ought not rely on current models as a basis for Kyoto-inspired fossil fuel restrictions.


Reviewed 1 April 2000