SOI fluctuations over several centuries. The temperature reconstructions show marked variability on decadal time scales. The tree-ring temperature reconstruction for Tasmania by Cook et al. (1991) shows a pronounced warming trend over the last few decades, which is consistent with sea surface temperature variations in the SH and in this region over the same period (Folland et al., 1990). The tree-ring series from Tasmania is discussed further in this volume by Cook et al. (1995).
Other proxy evidence of climate variability in the SH has been obtained from coral growth rings (Isdale, 1984), geological lake-level records (Kotwicki and Isdale, 1991), ice cores (Thompson et al., 1984), river-flow records (Wells, 1987; Whetton et al., 1990) and from a variety of sources (Markgraf et al., 1992). In general, however, it is difficult to directly relate these regional proxy data directly to large-scale climate variability in the SH. They provide an indication of the range of climate variability in the source regions for the data. Further work is required to determine the representativeness of some of these regions for the climate of the whole SH.
The dominant modes of intermonthly and interannual variation of the SH circulation have been described using results from several studies of grid-point analyses of the SH tropospheric circulation. The two leading modes are primarily zonally symmetric, representing out-of-phase variations of height between middle and high latitudes in one case (called the high-latitude mode) and between the tropics and middle latitudes in the other (called the low-latitude mode). These two modes, which explain almost a quarter of the variance of monthly mean 300 hPa height in the SH in both summer and winter, have large interannual variations. The dominance of these zonally symmetric modes by comparison with wavelike or regional patterns in the SH is in contrast to what is typical in the NH. Wave-train patterns of anomalies are also found in the SH circulation. These include a wave-number-three mode in winter with large amplitude at high latitudes, a wave-number-four mode in middle latitudes in summer, a meridional wave-train structure originating over Australia, and a monopole or meridional wave train over the Pacific Ocean in winter. The two most important processes leading to interannual variations of the SH circulation appear to be the Southern Oscillation and the wave, mean-flow interaction in the SH storm track, which leads to the high-latitude mode.
Since upper-air data and grid-point analyses are available for only a very short period in the SH, surface data must be used to describe decadal variability of the SH circulation. Because of the barotropic vertical structure of low-frequency variations in the SH, the surface-pressure variations are representative of variations throughout the troposphere. Relatively little attention has been paid to decadal variability in the SH. It appears from indices of the two dominant modes of interannual variations, determined using surface data, that the modes show substantial decadal variability over the last 90 years.
There are few observational data that can be used directly to indicate climate variability on century time scales in the SH. The only possible observational evidence for this is proxy data, such as tree rings.
This brief review represents my own perspective on studies of observed variability of the SH circulation. It is incomplete, and I apologize for any omissions. I wish to thank Phil Jones for providing Figure 4 and the station data for generating the time series of the TPI in Figure 5, Rob Allan for allowing access to his detailed bibliography on ENSO, and my many colleagues around the world who have provided stimulating discussions on the SH circulation over a number of years. This review was supported in part by a Dedicated Greenhouse Research Grant from the Australian National Greenhouse Advisory Committee. I am grateful to the U.S. National Research Council for the invitation to prepare this brief review.