There appears to be an upper bound to these data, represented by the straight line on the plot. The only ''outliers" are the Vostok ice-core points and a group of points representing recent ground-surface temperatures inferred from borehole temperatures in Ontario, Canada (Wang and Lewis, 1992).4 The line (fitted by eye) represents a decrease of the time-rate of warming in the form of a power law, ΔT = 0.38 Δy0.27, where 0.38 is the value of the function at Δy = 1 year. I propose that this represents the maximum rate of increase of global mean temperature observed over the past 850,000 years, no matter what the cause, or whether it arises from internal or external forcings. It seems appropriate to use this function as the standard against which we should compare model predictions of rates of global temperature increase.

Also plotted on Figure 1 (as open triangles) are several points representing Northern Hemisphere or northern temperate-zone temperature increases. These include the data of Angell (Boden et al., 1992) for the 30°-60°N latitude zone; Ontario (Canada) borehole temperatures (Wang and Lewis, 1992)5; and information from the Vostok ice cores (Barnola et al., 1987). From the Vostok data, I have chosen the two periods of most sustained and rapid warming and calculated the temperature difference for those periods.

For these data, I have drawn an upper-limit curve parallel to the global curve, a reasonable first guess.6 The equation for this line is ΔT = 1.05 Δy0.27. As expected, mid-latitude temperature increases are greater than global increases. The upper curve indicates rates of increase about 2.75 times the global curve. Also, the rates of increase in Northern Hemisphere temperature are only slightly greater (about 10 percent) than the whole-globe rates.

The first thing to note on this graph is that the "maximum" observed global rate of increase is greater by a factor of 2 or 3 than the temperature increases observed in the past century. Why should this be so? If greenhouse-gas increases over the past century (estimated to be an increase of 50 percent in the equivalent CO2 concentration (IPCC, 1990a)) have resulted in global warming, we might expect the rates of increase in global mean temperature to be at or above the "maximum" rate indicated by the solid line in Figure 1. The fact that they are well below this line implies that there may be some unknown climatic mechanism partially compensating for any warming caused by increases in greenhouse gases, or even that the observed warming is a natural trend reflecting little or none of the increase in greenhouse gases.

Another possibility is that model predictions of the effect of past increases of greenhouses gases are too high. It is thus appropriate to compare model-projected rates of increase for the past century with the observational record.


IPCC-I modeled the past century's observed climate, using observed increases in greenhouse gases, for various "climate sensitivities" (Figure 2). Their "best estimate" is a global mean temperature increase of 0.73°C (plotted on Figure 1 as the lowest open circle). Since the observed increase has been only 0.45°C, or even less if one accepts Balling's (1992) reanalysis, the IPCC-I model predictions are high by at least 60 percent, and perhaps as much as 150 percent.

Figure 2 

Global mean temperature change for the past century  as modeled by IPCC-1, for various values of climate sensitivity  (temperature change for a doubling of greenhouse gases) and the  recommended ocean thermal diffusivity (K) and downwelling  parameter (π). Observed global mean temperature is superimposed  on the modeled curves. (Adapted from IPCC, 1990a, Figure 8.1a;  reprinted with permission of the Intergovernmental Panel on Climate Change.)


Kirk Bryan (personal communication, 1992) has suggested that there are two spectral regimes in the temperature record, one for the ice age with relatively large oscillations, and another, relatively calmer, for the last 8,000 years, in which oscillations have been less rapid. However, it is likely that the physical causes limiting the rate of temperature rise would be the same during both periods; only the magnitudes of the temperature changes would be expected to be different in the two periods.


The average ground-surface temperature was inferred from the temperature profile in the boreholes and the thermal conductivity of the subsurface material. This inferred temperature increase over the past century is much greater than the measured increase in global temperature, perhaps because the rate of increase is larger at high latitudes than at low latitudes. The borehole data could (and should) be checked against standard meteorological observations from nearby weather stations. I scaled the data off rather small, perhaps not perfectly accurate, diagrams.


I have ignored one outlier, the temperature increase marking the end of the Younger Dryas as inferred from Greenland ice cores by Dansgaard, et al. (1989). This point implies a rate of warming twice that of the high-latitude curve of Figure 1.

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