pressure and temperature anomalies since 1886, smoothed to highlight decadal-scale fluctuations. Each series shows persistently negative anomalies in the 1895 to 1920 period, indicating a clear link between anomalously low warm-season SLP and below-average temperatures over Tasmania, presumably related to an expansion of the circumpolar vortex at that time and the northward migration of the subtropical high-pressure belt off the eastern coast of Australia. Figure 7 also indicates a link between SLP and the recent warming in Tasmania. Since about 1960, both warm-season SLP and temperature have been anomalously high, which suggests a southward movement of the subtropical high-pressure belt and a greater frequency of warm, northeasterly winds over Tasmania. This inference is supported by Lough (1991), who showed that the high-pressure belt moved southward about 2° over the 1942-to-1981 period. It is also supported by Coughlan (1979), who showed that annual mean maximum temperatures over Tasmania were significantly correlated with the latitude of the subtropical high.
There is also a possible link between ocean circulation features in the Tasman Sea region and changing land and air temperatures over Tasmania. This link is suggested by changes in the correlation fields of warm-season (November to April) land and marine temperatures versus reconstructed Tasmanian temperatures for two 40-year periods (1888 to 1927 and 1950 to 1989) containing the cold and warm periods indicated in Figure 7. To generate the correlation fields, we obtained corrected land and marine monthly temperature data for 5° × 5° grid cells covering the southern Australia-Tasmania-New Zealand sector. These data were kindly provided by P.D. Jones and K.R. Briffa of the Climatic Research Unit, University of East Anglia, Norwich.
Figure 8 shows these correlation fields. The integer in each cell reflects the correlation coefficient rounded to the nearest tenth. For the cool early period (1888 to 1927), the correlation field is weighted toward the grid boxes off the west coast of Tasmania. This result implies that the inferred expansion of the circumpolar vortex coupled with increased southwesterly winds in the 1890 to 1920 period may have caused the cold West Wind Drift to be displaced northward from its normal position, resulting in anomalously cool temperatures over Tasmania (see Figure 7). In contrast, the correlation field of the late period (1950-1989) is strongly weighted toward the Tasman Sea east of Tasmania. Given the anomalously warm temperatures in Tasmania at that time (Figure 7), this result suggests concomitant warming in the Tasman Sea and more southerly penetration of the warm East Australian Current. In fact, the actual SSTs during this time were anomalously warm, in parallel with warm temperatures over Tasmania. This result is consistent with the southward shift of the subtropical high-pressure belt (Lough, 1991) and increased frequency of warm northeasterly winds over Tasmania.
None of the empirical relationships between Tasmanian
temperatures and large-scale atmosphere/ocean processes can explain why oscillations (which have apparently persisted over the past 2290 years) are present in the reconstruction. However, should a physical model be developed, it will probably involve interactions in the atmosphere/ocean/ cryosphere system that influence the movements of the circumpolar vortex and subtropical high-pressure belt. Recently, Stocker and Mysak (1992) proposed that selfsustained oscillations, of similar order to those found here and in many other proxy climatic records. could be induced by internal changes of the ocean thermohaline circulation. Using a simple two-dimensional ocean model, they showed that internal nonlinear dynamics alone could produce self-