A series of experiments concerned with the variation of parameters, and of the finite difference formulations for the equivalent barotropic model in association with an automatic weather analysis system, have been carried out for the Australia region. These tests precede a more extensive testing of suitable formulations of the mode l with a view to application in routine forecasting. The importance of numerical scale matching of both the automatic analysis system and the prognosis system is discussed, and it is shown that in the system presented here some first order approximations to derivatives in the prognosis system can be mis-matched with the analysis system output. The use of higher order finite difference approximations to the Laplacian onerator, and of an energy conserving Jacobian operator is investigated in association with a "balanced" equivalent barotropic model and a geostropic equivalent barotropic model. The effect of varying amounts of smoothing, neglect of map factor on a Lambert conformal map, variation of Coriolis parameter, varying time step and initial conditions, and the practical stability of the model is discussed.
R. Maine (Fri,) studied this question.