A refinement of cyclone tracking methods with data from FROST

The Antarctic First Regional Observing Study of the Troposphere (FROST proj ect had three one-month Special Observing Periods (SOPs) during which the c ommitment was made to ensure that all additional data collected were passed on via the Global Telecommunication System (GTS) to operational centres fo r use in the construction of the analyses. These analyses can be regarded a s the best available for these times of year, given the special effort to i nclude additional data south of 50 degrees S during these periods. The availability of these high-quality analyses has stimulated us to refine the Melbourne University numerical cyclone tracking algorithm, with additi onal synoptic guidance gained from a manual analysis of southern hemisphere cyclones in the winter SOP (July 1994). Using the refined scheme me have c ompiled and compared statistics of cyclone tracks obtained objectively from the Australian GASP (Global Assimilation and Prediction) system analyses a nd manually from semi-independent analyses. Our results show that the cyclo nes found by the numerical and manual approaches bear considerable similari ty to each other, even for complex systems for which such unanimity might n ot have been expected. In general, the automatic algorithm tended to 'find' more systems than did the manual analyst, with these extra systems being p redominantly those identified as weak and/or open. The results emphasise th e difference in perception of what constitutes a low. The overall behaviour of cyclones revealed by the objective scheme in July 1994 was consistent with that identified in various climatologies in that m any systems were generated in the western part of the ocean basins and move d to the east and, to a lesser extent, to the south. A concentration of tra cks was found just to the north of the Antarctic continent. On the other ha nd, this specific month was anomalous in a number of respects; this was ref lected in the nature and distribution of cyclone activity. The consistency of the findings with those of an experienced, practising synoptician means that the state-of-the-art numerical algorithm can be applied to numerical a nalyses and model output with confidence.