The 42 papers presented in this volume span the field of natural climate variability on decade-to-century time scales. Together with the essays, commentaries, and discussions, they show that impressive progress has been made toward the goals of describing, understanding, and modeling the spatial and temporal structure, the magnitude, and the patterns of natural variability. Taken as a whole, they have provided the Climate Research Committee with the perspective needed to draw the conclusions that are discussed in this chapter. These conclusions suggest the research directions and priorities most likely to yield useful insights and further progress; they also were valuable as points of departure for the committee discussions that yielded the recommendations in Chapter 7.
The relatively short instrumental record of climate (the last 50 to 100 years), which reflects anthropogenic change as well as natural variations, does not represent a stationary or steady record. Instead, climate fluctuations over the past few millennia or so will need to be analyzed to establish a baseline of natural variability against which future (and present) variations can be gauged. Many of the papers in this volume contribute data or insights toward this end; they show that this natural propensity for change has manifested itself through all the possible modes of change shown in Figure 1 of the Introduction—periodic variations, sudden shifts, gradual changes, and changes in variability.
Periodic variations (Figure 1a) are apparent, for example, in the 2290-year-long Tasmanian tree-ring record presented by Cook et al. in Chapter 5. The rings register temperature swings over periods averaging 31, 56, 79, and 204 years, and since 1700 the 79-year fluctuation has been marching nearly in step with a similar variation related to the 11-year sunspot cycle. Nearly periodic fluctuations in estimated mean annual temperature are apparent in the last few thousand years of the Greenland ice-core records (see Grootes's paper in Chapter 5). Similarly, periodic fluctuations are apparent in the 90-year recorded relationship between North Atlantic surface wind and air temperature (see Deser and Blackmon's paper in Chapter 2) and in 130 years of global surface air temperature data (see Keeling and Whorf's paper, also in Chapter 2, and Figure 1 in Michael Ghil's essay introducing atmospheric modeling in Chapter 3). Quasi-decadal periodicities in North Atlantic ocean properties have been documented at the surface (by Deser and Blackmon, e.g.), where there is a dipole of opposing tendencies with centers east of Newfoundland and off the southeastern United States. At depth in the North Atlantic, significant changes in salinity and temperature with decadal and longer time scales have also been observed (see Lazier's and Levitus's papers in Chapter 3).
Sudden regional shifts or jumps (Figure 1b) of several degrees in mean annual temperature, sometimes in just a few years, can be seen in Grootes's Greenland ice-core data; they may exceed 10° in a century. The Northern Hemisphere land-temperature records used in the analyses of Jones and Briffa (Chapter 5) show regional jumps of autumn temperature of more than 0.5°C during the 1920s (the higher levels persisted for 20 years or so), and the temperature data used in Karl et al. (Chapter 2) shows a jump in the variation of diurnal temperature range of about 0.3° in the 1950s. Decadal-scale variability is by no means limited to temperature fluctuations, as is clearly demonstrated by Figure 1 in Thomas Karl's essay introducing the atmospheric observations section, and in the papers by Nicholson, by Shukla, and by Groisman and Easterling in that section. The precipitation in the United States was 5 to 10 percent higher in the 1970s than in the 1930s or 1950s. In the Sahel. precipitation abruptly decreased by more than 50 percent during the period 1968-93, and has persisted at that reduced level for the past few decades. Precipitation over southern Canada has been shown to have increased substantially (over 10 percent) during the 1970s and 1980s. Jumps in regional ocean temperatures and salinity have also been documented (see, for example, the papers by Mysak, Dickson, and Levitus in Chapter 3). Sudden shifts occurred in the surface properties of, and atmospheric circulation over, the North Pacific in 1976 and 1988 (see Cayan's paper in Chapter 3), and numerous researchers such as Dickson (see Chapter 3) have reported marked changes in ocean properties in various parts of the North Atlantic in response to the passage of a surface salinity anomaly.
Gradual climate changes (Figure 1c) are apparent in a variety of records. For example, the mean hemispheric air temperature records analyzed by Jones and Briffa (Chapter 5) show a gradual warming of approximately