The following HTML text is provided to enhance online
readability. Many aspects of typography translate only awkwardly to HTML.
Please use the page image
as the authoritative form to ensure accuracy.
Natural Climate Variability on Decade-to-Century Time Scales
To illustrate the potential value of proxy data further, we concentrate on dendroclimatic reconstructions. Some of the longest represent "warm-season" temperatures in northern Fennoscandia, the northern Urals, Tasmania, and northern Patagonia (Briffa et al., 1990, 1992a, 1995; Graybill and Shiyatov, 1992; Cook et al., 1991, 1992; Villalba, 1990; and Lara and Villalba, 1993, respectively). Details of some of these reconstructions are given in Table 3 and in the caption of Figure 4. As noted in an earlier section, summer (particularly in the Northern Hemisphere) is the season that appears from the instrumental record since 1850 to be most atypical in its long-term changes (see also Briffa and Jones, 1993). Another potential limitation of dendroclimatic reconstructions is that they may not fully represent very-long-time-scale variability because of the need to transform original ring-width measurements into indices in order to minimize any potential bias in the mean chronology resulting from different ages of the constituent trees (Cook et al., 1990, 1995a). The apparent character, in terms of long-time-scale variability, of the different chronologies (and hence their climate reconstructions) depends on the longevity of the trees (and hence the length of the measurement time series), the ecology of the site(s), and the dataprocessing method used to remove the 'age effect'.
Figure 4 shows the reconstructions filtered with 20- and 100-year low-pass filters. Each of the series has been rescaled to represent anomalies from the 1901-1960 period. The importance of the different chronology-production methods is demonstrated in the upper halves of both panels in Figure 4, which show alternate reconstructions for each of two northern high-latitude regions: northern Fennoscandia, produced using the same ring-width and density measurement data (Briffa et al., 1990, 1992a), and the northern Urals region, based on different data sets (ring width only for June-July (Graybill and Shiyatov, 1992) and ring widths and ring density for May-September (Briffa et al., 1995)). Differences in the way the original growth measurements for the many individual trees were amalgamated within these chronologies has markedly affected the appearance of the long-time-scale variability in the resulting climate reconstructions. However, establishing the veracity, or true confidence levels, associated with the long-period fluctuations is extremely problematic.
Notwithstanding the methodological considerations, Figure 4 also shows that, compared to the southern series, the variance of the northern reconstructions is much greater. This may reflect the strong moderating influence of the southern oceans evident in the lower variance of the observational data in these regions. The explained variances in the southern reconstructions are also generally lower than those in the north (see Table 3). On the 20-year time scale, the temperatures for the two northern regions are largely out of phase between A.D. 1000 and 1200. In these comparisons, we use the two thicker curves for Fennoscandia and the Urals, which show the greater variability on decadal and longer time scales when Fennoscandia was warm and the Urals cold. After 1600, the two curves are more similar, both being cold through much of the seventeenth and the second half of the nineteenth centuries. The drop in temperatures in Fennoscandia began around 1580 and lasted to around 1750, whereas in the northern Urals the drop began earlier, the coldest period being from 1530 to 1670. Both series indicate cold summers around the end of the nineteenth century, although conditions were more severe in the Urals. The Fennoscandian series also shows protracted or relatively severe cold during the sixth, seventh, ninth, early twelfth, thirteenth, and early fourteenth centuries. On the 100-year time scale the Fennoscandian series shows generally continuous warmth exceeding that of the 1930s lasting from about 900 to 1100. Shorter warm periods also occurred during the 760s, around 1160, and notably in the early decades of the fifteenth century. The Urals series shows warmth in the thirteenth and fourteenth centuries, but on this time scale it does not exceed that of the twentieth century.
The nature of the Southern Hemisphere reconstructions is very different from that of those in the north. Low-frequency variability is conspicuously less. In Tasmania, the main feature of the entire reconstruction is a rapid increase in temperature since early this century. For northern Patagonia, the two series show less agreement with each
TABLE 3 Variance (r2) Explained by Calibration of Different Summer Temperature Reconstructions Based on Tree-ring Data (see Figure 4)