There has been relatively less study of the variability of the Southern Hemisphere (SH) atmospheric circulation than of that of the Northern Hemisphere (NH). There are a number of reasons for this; the lower population and greater ocean areas in the SH mean that observational records are sparser and there have been fewer people to analyze them. In this brief and incomplete review, I will concentrate on studies of the variability of the SH tropospheric circulation, starting at interannual time scales and moving onto longer time scales. This progression is justified by the available observational data and the number of relevant studies, as outlined below.
Detailed and regular observations of the troposphere over much of the SH did not start until the International Geophysical Year in 1957-1958, when the number of SH upper-air sounding stations increased markedly. Operational subjective analyses of the SH circulation have been prepared since that time by the Australian and South African weather services. The introduction of meteorological satellites in the late 1960s led to the development of the first operational objective numerical-analysis scheme for the SH at the Australian Bureau of Meteorology in 1972. Hence, objective numerical analyses for the SH are available for only the last 20 years, and upper-air sounding data are available for only about 30 years.
Hemispheric numerical analyses have been used to investigate the interannual variability of the SH circulation, particularly over the last decade (Trenberth, 1979; Rogers and van Loon, 1982; Mo and van Loon, 1984; Mo and White, 1985; Kidson, 1988a,b; Karoly, 1989a; and others). In addition, radiosonde station data have also been used for similar investigations by Szeredi and Karoly (1987a,b). The results of these studies of the dominant modes of variability of the SH circulation on interannual time scales are summarized in the next section.
Since these upper-air data are available for a short period only, it is not possible to use them to describe variations on decadal time scales or longer. There are longer records of surface meteorological observations for SH stations, some extending back to before the turn of the century. The equivalent-barotropic vertical structure for the low-frequency tropospheric variations (Szeredi and Karoly, 1987a) permits these surface records to be used to provide some information on decadal variability in the SH. However, there have been no comprehensive studies of decadal time-scale variability for the SH. More recently, Jones (1991, hereafter referred to as J91) has reconstructed from station data a gridded mean sea-level pressure data set for the SH that extends back to 1911. J91 and Allan and Haylock (1993) have used this data set to consider decadal variations of the SH circulation. These and a few other studies of long surface records are described briefly below, although the others have generally concentrated on variations of El Niño and the Southern Oscillation.
In an attempt to provide some evidence of the range of SH variability on century time scales, some climate reconstructions based on proxy data are discussed below. Although these proxy data are available for very few sites in the SH, they constitute the only source of data on natural climate variability on century time scales for the SH. More extensive descriptions of proxy data for the SH are provided in Chapter 5 of this volume.
Relatively less attention has been paid to low-frequency variations of the SH circulation than to those of the NH, probably because of the shorter time period for which analysis data sets are available. Trenberth (1979), Kidson (1988a), and Shiotani (1990) have used zonal-mean data to show that there are large low-frequency variations of the zonal-mean circulation in the SH involving variations of the zonal wind and associated dipole variations of geopotential height. Mo and White (1985, hereafter referred to as MW) carried out a teleconnection analysis for the SH (similar to that of Wallace and Gutzler (1981) for the NH) using 8 years of monthly mean SH analyses. From their grid-point correlation maps, they identified three modes of variation: a zonally symmetric dipole mode at middle to high latitudes, a zonal wave-number-three mode in winter, and a continent-ocean contrast in summer. Szeredi and Karoly (1987b, hereafter referred to as SK) used the longer period of upper-air station data available for the SH (up to 30 years) to investigate low-frequency variations. They found two dominant modes of monthly variation in the station data that were primarily zonally symmetric, involving opposing departures of height between middle and high latitudes in one case and between the tropics and middle latitudes in the other. The sparse and irregular station network in the SH meant that only the largest-scale modes could be identified.
More recently, the technique of unrotated and rotated principal-component analysis (PCA) has been applied to grid-point analyses of the SH to clarify the typical modes of low-frequency variations of the SH circulation in a number of different studies. These may be separated into two groups according to the time scale of interest: Periods longer than about 50 days, including interannual variations, have been considered by Rogers and van Loon (1982), Kidson (1988b, hereafter referred to as K88), and Karoly (1989b), while another group concentrated on intraseasonal variability. The results are similar for different studies of the same time-scale fluctuations, and differences can usually be explained in terms of differences in analysis method, such as PCA of the correlation (Karoly, 1989b) or covariance matrix (K88), use of different variables (300 hPa height by