Scripps Institution of Oceanography
La Jolla, CA 92093-0224
Our poor understanding of cloud processes limits our ability to make realistic climate change predictions. Part of the problem is that we have too many cloud parameterizations and too few observations. Lack of contact between observationalists and modelers exacerbates this problem.
Somerville and Sam Iacobellis have developed a diagnostic model in the form of a single-column version of a general circulation model (GCM), which is used to test the various available parameterizations. The principle behind the single-column model is that the horizontal convergence of heat, momentum, and moisture is specified from observations (typically, analyzed fields, such as numerical weather prediction data), allowing diagnostic prediction of model profiles of temperature and humidity from local sources and sinks of heat and water; plug-compatible parameterizations are used to calculate the sources and sinks.
The single-column model is currently being applied to a 200- kilometer by 200-kilometer region centered on the Oklahoma ARM site. Initial tests have shown that significantly different cloud fractions are predicted using three different cloud parameterizations (a relative humidity model and two prognostic cloud liquid water models) (Figure 19.1). There is a need to compare the parameterizations using variables which are readily measurable observationally. One such variable is the net solar flux at the surface which also shows significant differences when the various cloud parameterizations are used.
The parameterizations are also being tested using the CCM-2 GCM running on a workstation (Figure 19.2). This eliminates the limitation of externally specifying horizontal convergence. When a liquid water budget is included in the model, some high tropical clouds are underpredicted. All versions predict a strong temperature increase in the upper troposphere. This is because a large vertical transport of heat is produced by the CCM2 mass flux convective scheme, regardless of the cloud parameterization used.
The combination of well-thought-out field programs, diagnostic modeling, (as with single-column GCMs), and the increased accessibility of GCMs to the university community (by making them available on workstations), should lead to more rapid progress in validating parameterizations.
Reference
Iacobellis, Sam F. and Richard C.J. Somerville, Diagnostic Modeling of the Indian Monsoon Onset. Part I: Model Description and Validation, J. Atmos. Sci., 48, pp.1948-1959, (1991)
