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.
Surface Temperature Reconstructions for the last 2,000 Years
and Swetnam 1989). Although all crossdated samples are entered into the final chronology, standardization removes any difference in mean growth rate between specimens, so that faster-growing trees do not dominate the record. Any criteria used to form a chronology out of a subset of the crossdated specimens need to be clearly reported and justified.
The identification of year-to-year (high-frequency) climate signals in tree ring records is relatively straightforward since it is based on the elimination of time series autocorrelation using autoregressive models (Biondi and Swetnam 1987, Cook and Kairiukstis 1990). If adequately long instrumental records are available, it is even possible to explore the stationarity of statistical relationships between climatic variables and tree ring parameters by considering multiple time intervals (Biondi and Waikul 2004).
With regard to low-frequency temperature patterns, the length of the individual tree ring records used to produce a master chronology (rather than the length of the chronology itself) can influence the reconstruction (Cook et al. 1995). It is also difficult to distinguish the amount of temporal autocorrelation in tree ring records that is linked to biological processes instead of climatic ones (Fritts 1976). One way to resolve these issues is to compute the expected value of the tree ring parameter (width, density, etc.) as a function of biological age (i.e., time since ring formation), and use the resulting growth curve to standardize the individual tree ring series. This method, which is now called Regional Curve Standardization (RCS), was first proposed in the 1930s (Grudd et al. 2002), later described by Fritts (1976), and made popular by Briffa et al. (1992). In addition to its theoretical appeal, the RCS method is suitable for retrieving low-frequency signals in tree ring records (Esper et al. 2003, Bunn et al. 2004) and is widely employed in dendroclimatic reconstructions of surface temperature (Esper et al. 2002a, Gunnarson and Linderholm 2002, Naurzbaev et al. 2002).
To prevent the risk that a single tree ring chronology could reflect the influence of localized nonclimatic influences (Fritts 1976, Trotter et al. 2002), dendroclimatic reconstructions often rely on networks of site chronologies. Regional tree ring networks typically have strong intersite correlations (e.g., Hughes et al. 1984, Figure 2), and continental-to-hemispheric-scale networks are able to reproduce synoptic-scale climatological patterns (Fritts 1991, Briffa et al. 2002). When based on a number of sites in the Northern Hemisphere, dendroclimatic reconstructions of surface temperatures show that the 20th century warming was unusual since at least 1500 (D’Arrigo et al. 2006; Figures 4-1 and 4-2), in agreement with independent reconstructions derived from written documents (Xoplaki et al. 2005), borehole temperatures (Pollack and Smerdon 2004), and glacier lengths (Oerlemans 2005a). When records are sought for the last two millennia, the number of available tree ring chronologies declines markedly (Hughes 2002), so confidence in reconstructed patterns is reduced.
All paleoclimatic reconstructions rely on the “uniformity principle” (Camardi 1999), which assumes that modern natural processes have acted similarly in the past, and is also discussed as the “stationarity” assumption in Chapter 9. Although limiting factors controlled tree ring parameters in the past just as they do today, it is possible that the role of different factors at a single location or over an entire region could change over time. This possibility has been raised to explain the “divergence” (i.e.,