To determine whether 20th century warming is unusual, it is essential to place it in the context of longer-term climate variability. Owing to the sparseness of instrumental climate records prior to the 20th century (especially prior to the mid-19th century), estimates of global climate variability during past centuries must often rely upon indirect "proxy'' indicators - natural or human documentary archives that record past climate variations, but must be calibrated against instrumental data for a meaningful climate interpretation (Bradley, 1999, gives a review). Coarsely resolved climate trends over several centuries are evident in many regions e.g., from the recession of glaciers (Grove and Switsur, 1994; and Section 220.127.116.11) or the geothermal information provided by borehole measurements (Pollack et al., 1998). Large-scale estimates of decadal, annual or seasonal climate variations in past centuries, however, must rely upon sources that resolve annual or seasonal climatic variations. Such proxy information includes width and density measurements from tree rings (e.g., Cook, 1995; see Fritts, 1991, for a review), layer thickness from laminated sediment cores (e.g., Hughen et al., 1996; Lamoureux and Bradley, 1996), isotopes, chemistry, and accumulation from annually resolved ice cores (e.g., Claussen et al., 1995; Fisher et al., 1998), isotopes from corals (e.g., Tudhope et al., 1995; Dunbar and Cole, 1999), and the sparse historical documentary evidence available over the globe during the past few centuries (see e.g., Bradley and Jones, 1995; Pfister et al., 1998). Taken as a whole, such proxy climate data can provide global scale sampling of climate variations several centuries into the past, with the potential to resolve large-scale patterns of climate change prior to the instrumental period, albeit with important limitations and uncertainties.
The SAR examined evidence for climate change in the past, on time-scales of centuries to millennia. Based on information from a variety of proxy climate indicators, reconstructions of mountain glacier mass and extent, and geothermal sub-surface information from boreholes, it was concluded that summer temperatures in the Northern Hemisphere during recent decades are the warmest in at least six centuries. While data prior to AD 1400 were considered too sparse for reliable inferences regarding hemispheric or global mean temperatures, regional inferences were nonetheless made about climate changes further back in time.
Since the SAR, a number of studies based on considerably expanded databases of palaeoclimate information have allowed more decisive conclusions about the spatial and temporal patterns of climate change in past centuries. A number of important advances have been in key areas such as ice core palaeoclimatology (e.g., White et al., 1998a), dendroclimatology (e.g., Cook, 1995; Briffa et al., 1998b), and geothermal palaeo-temperature estimation (e.g., Pollack et al., 1998). Moreover, the latest studies based on global networks of "multi-proxy" data have proved particularly useful for describing global or hemispheric patterns of climate variability in past centuries (e.g., Bradley and Jones, 1993; Hughes and Diaz, 1994; Mann et al., 1995; Fisher, 1997; Overpeck et al., 1997; Mann et al., 1998, 1999). Such estimates allow the observed trends of the 20th century to be put in a longer-term perspective. These have also allowed better comparisons with possible physical influences on climate forcings (Lean et al., 1995; Crowley and Kim, 1996, 1999; Overpeck et al., 1997; Mann et al., 1998; Waple et al., 2001), and for new evaluations of the low-frequency climate variability exhibited by numerical climate models (Barnett et al., 1996; Jones et al., 1998; Crowley and Kim, 1999; Delworth and Mann, 2000).
The past 1,000 years are a particularly important time-frame for assessing the background natural variability of the climate for climate change detection. Astronomical boundary conditions have strayed relatively little from their modern-day values over this interval (but see Section 2.3.4 for a possible caveat) and, with the latest evidence, the spatial extent of large-scale climate change during the past millennium can now be meaningfully characterised (Briffa et al., 1998b; Jones et al., 1998; Mann et al., 1998; 1999; 2000a; 2000b). Moreover, estimates of volcanic and solar climate forcings are also possible over this period, allowing model-based estimates of their climate effects (Crowley and Kim, 1999; Free and Robock, 1999).
Other reports in this collection