The lack of widespread instrumental climate records prior to the mid-19th century requires that we turn to "proxy data" in our attempts to reconstruct how the climate has changed over past centuries. When, as is typically the case in studies of the past millennium, our interest is primarily in annually resolved climate variations, we must turn to "high-resolution" climate proxy data, such as tree rings, corals, ice cores, and historical documentary records. Mann et al. (1998) assembled a network of 415 annually resolved proxy data (predominantly dendroclimatic), for use in reconstructing temperature patterns over the past 1000 years. More recently, Mann et al. (submitted) have assembled a much larger network of 1209 annual and decadally resolved proxy data consisting of tree rings (including 105 gridded maximum latewood density or "MXD" tree-ring series - see Briffa et al. 2001), corals and sclerosponge series, ice cores, lake sediments, speleothems, historical
oo of- «o ooo o o ooooloooooo oooooo oo o ooo ft?
Figure 7.1 Spatial distribution ofhigh-resolution climate proxy data. Nine different proxy types are denoted with different symbols. Starting date of proxy record is indicated by color scale.
documentary series, and a set of 89 gridded European surface temperature reconstructions back to ad 1500 that are based on a composite of proxy, historical, and early instrumental data (Luterbacher et al. 2004). The spatial distribution of the proxy data is shown in Figure 7.1.
Most previous proxy data studies have focused on hemispheric or global mean temperature (an example of one reasonably representative reconstruction is shown later in Figure 7.3), although some studies have also attempted to reconstruct the underlying spatial patterns of past surface temperature changes at global (e.g. Mann et al. 1998) and regional (e.g. Luterbacher et al. 2004) scales, and other fields such as regional sea-level pressure, and North American drought. Other studies have focused on the reconstruction of particular climate indices such as the North Atlantic Oscillation (NAO) and related Arctic Oscillation (AO), the Pacific Decadal Oscillation (PDO), and Atlantic Multi-decadal Oscillation (AMO), and indices such as the Southern Oscillation Index (SOI) and Nino3 index, which attempt to describe the variability in the El Nino-Southern Oscillation (ENSO) phenomenon. An extensive review is provided by Jones and Mann (2004).
Most reconstructions of hemispheric or global mean temperatures or of climate indices have employed the "Composite Plus Scale" (CPS) methodology, where a selection of natural climate "proxies", such as tree rings, ice cores, or corals, are first standardized, then composited to form a regional or hemispheric mean temperature series (cf. Jones and Mann 2004). More recently, methodological adjustments have been applied to the CPS method, including the use of proxies selected specifically for their retention of low-frequency variability (Esper et al. 2002; Mann and Jones 2003) or the inclusion of low-resolution (decadal or centennial-scale) proxies that might be well-suited to reconstructing low-frequency climate variability (Moberg et al. 2005). In the latter case, one must take care during the standardizing and compositing process to ensure that the low-frequency component is not artificially inflated relative to the higher-frequency component (Mann et al. 2005b).
An alternative is the climate field reconstruction (CFR) approach, which combines information from multiple proxy records in reconstructing the underlying spatial patterns of past climate change (e.g. Mann et al. 1998; Luterbacher et al. 2004; Rutherford et al. 2005). In this case, hemispheric or global means, as well as any climate indices of interest, are computed directly from the reconstructions of the underlying spatial field. A key advantage of the CFR approach is that the spatial reconstructed patterns can be directly compared with model-predicted patterns of climate responses to forcing (Shindell et al. 2001, 2004; Waple et al. 2002). Climate field reconstruction methods typically make use of both local and nonlocal information by relating predictors (i.e. the long-term proxy climate data) to the temporal variations in the large-scale patterns of the spatial field of interest. Such an approach takes greater advantage of the potential climate information in the proxy data-set. Two good examples are the close link between drought-sensitive tree rings from the western USA and surface temperature patterns in the Pacific Ocean, and ice-core records from Greenland which are closely tied to the behavior of the NAO, which influences cold-season temperature patterns over North America and Eurasia. These relations can be exploited in climate reconstructions. Climate field reconstruction approaches depend more on assumptions about the stationarity of relationships between proxy indicators and large-scale climate patterns than simpler methods such as CPS. Investigations using synthetic proxy data ("pseudoproxies"), however, suggest that these assumptions are likely to hold well for the range of variability inferred over the past one or two millennia. These investigations, which are described in more detail below, demonstrate that CFR methods are quite skilful in independent tests of their fidelity.
Was this article helpful?