There are at least five potential explanations for global-scale decadal to millennial climate variations. Natural, unforced variability, associated with either atmospheric dynamics or interactions within the climate system (without including ocean circulation changes), is one candidate; this potential explanation obviates the need to search for "causative" factors of the observed climate changes, and simply relies on the mechanisms inherent to the system. Extending this concept to include ocean circulation changes broadens the range of possibilities. Fluctuations in the atmosphere-ocean system are observed to occur today and might be a natural characteristic over longer time scales, although they could equally well come into existence as a response to changes in climate forcing. Solar variability exists as a favorite "wild card" possibility, legitimized to some extent by recent observations of solar irradiance changes during the most recent solar cycle. Volcanic aerosol injections appear to cool the climate, and a clustering of such events, or the lack of such events, would likely lead to a change in mean climate. Finally, variations in greenhouse trace-gases such as CO2 and methane have been observed to occur over the last several hundred years, with cumulative climate impact.

Given the lack of quantitative measurements of some of these forcings over time, one approach to estimating the cause(s) of decade-to-century-scale variability is to model each of these perturbations in a generic sense and compare the results to climate reconstructions. The model used for experiments done specifically for this paper is the GISS general circulation model (GCM) (Hansen et al., 1983b), run at 8° × 10° resolution, with sea surface temperatures either prescribed or calculated, as indicated in Table 1. This model has a sensitivity of 4.2°C warming for 2 × CO2 (or an increase of 2 percent in the solar constant) (Hansen et al., 1984), with a high-latitude/low-latitude sensitivity of about a factor of 2. The quantitative calculations given below are a function of this model sensitivity. Calculations from other GCMs are noted where appropriate.

The model results indicate that each of the potential explanations of climate variability listed above has its own characteristic signature that may provide a basis for discrimination in the climate record. In the subsequent sections we will discuss evidence suggestive of the various forcings, as well as relevant modeling studies depicting the possible nature of the climatic response. We concentrate mostly on temperature, since it is this parameter that is most often recoverable in the paleoclimatic data on these time scales. In the discussion section, we will briefly review what evidence is needed to decide on which causes are most likely.


Natural variability suggests an unforced oscillation in climate parameters. It can arise primarily from atmospheric dynamics or from atmospheric thermodynamics. In either mode it must alter the net heating of the surface, either locally, if that is all that is involved, or globally. For oscillations of multidecadal scale to be set in motion, portions of the system with large heat capacities or time constants (ocean, cryosphere) must be involved, or interactions setting in motion positive feedbacks (water vapor, cloud cover), which ultimately reverse direction, must arise. The ocean mixed layer by itself has an e-folding time constant of a few years, with an impact that can be lengthened via other feedback processes (Hansen et al., 1985). An example of the "reversing-direction" process, somewhat analogous to the mechanism proposed by Ramanathan and Collins (1991) although operating on a longer time scale, could be as follows: Warmer sea surface temperatures lead to an increase in high-level clouds and thus to a further greenhouse amplification of sea surface temperatures; eventually the clouds become sufficiently optically thick to reduce incoming solar energy, which set in motion a cooling of the surface and additional water-vapor feedbacks. Wigley and Raper (1990) suggested that high-frequency climate forcing, perhaps associated with clouds, could extend to a century-scale oscillation of a few tenths of a degree Celsius via the ocean's thermal inertia.

TABLE 1 GCM Experiments Discussed in the Text



Ocean Temperatures

Unforced variability

100 years

Calculated SSTs, with mixed-layer depth (MLD) averaging 125 m

Reduced North Atlantic SSTs

6 years

Prescribed colder North Atlantic SSTs, 1/5 full ice-age cooling

Reduced solar inputs of -0.25% or -2%

55 years each

Calculated SSTs with MLD of 65 m

Added volcanic aerosols t = 0.15

55 years

Calculated SSTs with MLD of 65 m

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement