Abstract
The main problem in weather forecasting over New Zealand and the Southern Oceans is the lack of sufficient data for good analyses, especially at upper levels. Numerical weather prediction also suffers from this problem, but it can to some degree be alleviated by producing several prognoses from reasonable alternative analyses to alert the forecaster to the sensitivity of the situation to data-blank areas. This procedure allows forecasters to assess the degree of confidence that can be placed in a forecast since unresolvable problems in the initial analyses cause uncertainty in predicting the exact timing and areas that will be affected by short-wave disturbances. It also gives them greater flexibility when later information becomes available and clearly invalidates one of the forecasts.
The performance of a five-layer quasi-geostrophic model developed for this purpose in New Zealand is described and illustrated with one case study of an unusual summer situation in the Australia–New Zealand region where a decaying tropical cyclone is absorbed into the frontal system of an extratropical cyclone undergoing explosive cyclogenesis.
