Predictability studies of the Antarctic atmosphere

In this study estimates are made of the predictability time-scales for the Weddell Sea section of the Antarctic region. Two distinct methods were applied. One of the methods used was a traditional approach due to Lorenz in which the rate of separation of initially close atmospheric states (analogues) was calculated. It was found here that even the best 500 hPa height field analogues were only mediocre and that root mean square differences between the analogues grew very rapidly, suggesting that the predictability time-scales might be quite short. However, the absence of very good analogues shows that an analogue approach is probably not suitable for the Weddell Sea region, and the predictability estimates obtained should be treated with caution. An alternative approach, in which recently developed chaos theory, also known as non-linear systems analysis, was applied to the Australian Bureau of Meteorology archived analyses of 500 hPa cyclone track positions, was also investigated. The divergence of initially close pieces of cyclone tracks was derived from the algorithm of Grassberger and Procaccia (1984). The results suggest the existence of a low-dimensional attractor, describing the dynamics of the weather systems, of fractal dimension somewhere between 5 and 6. Rates of separation calculated from this algorithm also gave short predictability time-scale estimates of about one day for the error doubling time. These results are consistent with estimates of error doubling times found by studies of other baroclinically active regions of the globe.