Paleoclimatic reconstruction may be considered to proceed through a number of stages or levels of analysis. The first stage is that of data collection, generally involving fieldwork, followed by initial laboratory analyses and measurements. This results in primary or level 1 data (Hecht et al., 1979; Peterson et al., 1979). Measurements of tree-ring widths or the isotopic content of marine foraminifera from an ocean core are examples of primary data. At the next stage, the level 1 data are calibrated and converted to estimates of paleoclimate. The calibration may be entirely qualitative, involving a subjective assessment of what the primary data represent (e.g., "warmer," "wetter," "cooler" conditions, etc.) or may involve an explicit, reproducible procedure that provides quantitative estimates of paleoclimate. These derived or level 2 data provide a record of climatic variation through time at a particular location. For example, tree-ring widths from a site near the alpine or arctic treeline may be transformed into a paleotemperature record for that location, using a calibration equation derived from the relationship between modern climatic data and modern tree-ring widths (see Chapter 10).
Different level 2 data may also be mapped to provide a regional synthesis of paleoclimate at a particular time, the synthesis providing greater insight into former circulation patterns than any of the individual level 2 data sets could provide alone (Nicholson and Flohn, 1980). In some cases, three-dimensional (3D) arrays of level 2 data (i.e., spatial patterns of paleoclimatic estimates through time) have been transformed into objectively derived statistical summaries. For example, spatial patterns of drought in the eastern United States over the last 300 yr (based on level 1 tree-ring data) have been converted into a small number of principal components (eigenvectors) that account for most of the variance in the level 2 data set (Cook et al., 1992b). The eigenvectors show that there are a small number of modes, or patterns of drought, which characterize the data. The statistics derived from such analyses constitute a third level of paleoclimatic data (level 3 data).
Most paleoclimatic research involves level 1 and level 2 data at individual sites, though regional syntheses are becoming more common (e.g., see individual chapters in Wright et al., 1993). At the larger, hemispheric or global scale, there are few studies of the spatial dimensions of climate at particular periods in the past. Notable exceptions are the CLIMAP and COHMAP reconstructions of marine and continental conditions at 3000-yr intervals from 18,000 yr B.P. to the present (CLIMAP, 1976; COHMAP, 1988; Webb et al., 1993a). Such syntheses provide rigorous tests of the ability of general circulation models to simulate climate under different boundary conditions and different forcing mechanisms (see Chapter 12).
Was this article helpful?