The equivalent-barotropic forecast equation has been tested on Australian data in a form including the horizontal divergence through an adjusted relative vorticity advection. Suitable choice of this adjustment to match synoptic conditions (including phases of predominant cyclone development or decay, achieved by computations backward and forward in time from the stage of incipient maturity) is shown to bring significant improvements over standard barotropic forecasts. The use of a small grid, necessitated by the data deficiencies of the area, is shown to limit the length of forecasts and the scales of motion which can be studied. Boundary conditions influence the entire grid area to some extent, but realistic boundary height adjustments make it possible to study the forecast equation dynamics independently of these errors. Forecasting of major atmospheric systems at the 500 mb level gave good results for 24 hours but became much less accurate and useful after 48 hours. The average error in predicting the positions of cyclone centres was 1 to 1.5 grid units after 24 hours and this had doubled after 48 hours, effectively ruling out longer forecasts. Verification by a variety of statistical parameters led to equivalent results and conclusions. Fields of vertical velocity and diabatic heating implied in the equivalent-barotropic model have been compared with vertical velocity estimates using point and trajectory observations, and with observed rainfall and layer cloud areas. While the model vertical velocities agree only in order of magnitude with other estimates, the observed cloud areas are shown to be conclusively related to the upward velocities both of the 24-hour and, less closely, of the 48-hour forecasts.
Voice et al. (Fri,) studied this question.