The persistent skill of the 500 mb barotropic forecasts in routine numerical weather prediction is emphasized in contrast to the failure of baroclinic prediction models to furnish useful low-level forecasts. In the belief that this failure is first of all a result of lack of control mechanism capable of ensuring well-behaved development, the author suggests the possibility of designing a multi-level prediction model capable of realistic displacements of pressure systems but containing no mechanism for development.
For this purpose, a one-layer model consisting of a stratified fluid is exposed to theoretical analysis by means of the technique of small perturbations. This analysis shows that owing to the stratification of the fluid, the perturbations have a vertical structure resembling that of the atmosphere. The dynamics of the perturbations is studied in some detail. One of the results is an expression for wind divergence applicable to non-linear prediction. It is shown that the stratified model is capable of more realistic displacement of planetary waves than either version of the barotropic models (Rossby, 1939; Cressman, 1958).
More important than this improved control of the planetary waves is the model's apparent capability of realistic displacements of pressure systems at low levels. In application, however, there is the problem of how to best interpret concepts from the perturbation theory such as basic current, thickness of the layer, and the static stability of the basic flow.