Abstract
A nine-level hemispheric primitive-equation model is applied to make a 14-day experimental prediction for the Southern Hemisphere; this is the first numerical prediction on the hemispheric scale for the Southern Hemisphere and the first attempt to use the model for summertime forecasting.
The predictions are verified against hemispheric data, the inadequacy of which is clear. They are also verified in the Australasian sector where the data are more adequate.
For the first 2 days, especially over areas with good data, the forecasts were generally of high quality, decreasing in accuracy with height above the surface. After a serious decline in skill on the fourth day, a recovery is noted. Australian rainfall during the first 2 days was also predicted with useful skill.
Some properties of the model atmosphere are compared with those of the observed Southern Hemisphere atmossphere, and deficiencies in the performance of the model, as well as in the analysis and initialization, are brought out by the comparison. However, on the whole, good agreement is found.
