Variability of Southern Hemisphere Extratropical Cyclone Behavior, 1958–97

Abstract

An analysis of the variability and trends exhibited by many aspects of Southern Hemisphere (SH) mean sea level extratropical cyclones during the period 1958?97 is presented. The investigation is undertaken by applying a state-of-the-art cyclone finding and tracking scheme to the 6-hourly reanalyses produced by the National Centers for Environmental Prediction. The outcome of this is arguably the most reliable analysis of SH cyclone variability undertaken to date. Across the 40-yr period the annual and seasonal mean cyclone densities have undergone reductions at most locations south of about 40°S (with the greatest reductions near 60°S), and increases to the north. This pattern of change resembles the ?high-latitude mode? identified in many studies of SH circulation features. It is shown that the mean radius of SH extratropical cyclones displays almost everywhere a significant positive trend, and there are also increases in annual mean cyclone ?depth? (i.e., the pressure difference between the center and the ?edge? of a cyclone). The annual average number of cyclones per SH analysis rose from the start of the period to a maximum of about 39 in 1972. Since then, the numbers have shown an overall decline, the counts in the 1990s being particularly low. Similar behavior was evident when the count was confined to the 30°?50°S and 50°?70°S latitude bands. Least squares best fit to the three time series exhibit significant slopes of ?0.58, ?0.26, and ?0.58 cyclones per analysis per decade, respectively. Between 30° and 70°S the annual mean number of cyclones found per analysis assumed a maximum about 1970, but that number has dramatically decreased by about 10% since then. (This analysis suggests that the downward trends in cyclone numbers are associated with a warming Southern Hemisphere.) The overall structure of the time series of annual cyclone per analysis over 30°?50°S and 50°?70°S are similar, but their year-to-year changes are shown to be negatively correlated; hence, there tends to be an interannual compensation of cyclone density between the middle and higher latitudes. The extent to which changes in the semiannual oscillation over the last few decades could be said to have influenced how cyclones are distributed across seasons is briefly examined. The results show, in particular, that the interannual relationship between spring and winter cyclone density cannot be explained in terms of a response to a change in the amplitude of the semiannual oscillation.