Skip to content

As the observed rate of rise in the global average temperature continues to be much less than climate models project, there are a growing number of knights in shining armor, riding to the rescue of the damsel in distress (the damsel, of course, being the climate models).

The rescue attempt generally employs two strategies, namely that 1) there is a bunch of stuff that has going on that the models couldn’t possibly have known about (so it is unfair to hold this against them), and 2) the climate models aren’t really doing that badly anyway.

The list of things proffered that the models couldn’t have known about that have led to slower-than-expected warming over the past 10-15 years includes declines in solar radiation, declines in stratospheric water vapor, increases in stratospheric aerosols, increases in tropospheric aerosols, the timing of El Nino/La Nina cycles, the timing of multi-decadal ocean circulation oscillations, and probably ultimately, the kitchen sink followed by the commode. What’s interesting is that the white knights never really mention these very same influences when they are acting in the opposite sense—that is, when they are acting to speed up the warming (which many were during the 1990s). But now that warming has considerably slowed, these mechanisms seem to have taken on cosmic significance.

What also seems to be conveniently overlooked in the list are changes that the models couldn’t have known about that are currently acting to enhance the observed warming in recent years—these include the recovery from the effects of Pinatubo and the reduction in summer Arctic sea ice (both of these mechanisms are explained in detail in our World Climate Report posts here and here).

But even while raising these excuses, the models’ champions are claiming success nonetheless. Nowhere better is this epitomized than in a forthcoming paper by Ben Santer and a long list of colleagues. Santer et al. have used a familiar procedure (familiar to us at least since we have oft-presented similar work, see here and here for example) to try to demonstrate that the observed trends over different timescales fall comfortably within the range of model expectations. Setting aside some methodological differences of opinion that we have with the analysis, there are still some interesting results to be found.

For instance, Figure 1 (below) taken from the new Santer et al. paper, shows the average of the observed trend set against the distribution of model trends (by the way, this is not really an apples-to-apples comparison; more on this at a later date) for periods of time from 10 to 32 years. The model average projected trend for the lower atmosphere is about 0.25°C/decade over all time scales (from 1979 through 2010) (green line in Figure 1). The 5%-95% spread of model projections is in yellow. The various averages of the observed trends over the different time scale (from several different observational datasets) are in red and blue and range from about 0.14 to 0.21°C/decade. It is obvious that for the longest trends—which is what people should really care about—that observed temperatures are perilously close to falling beneath the 95% confidence limits of the models (right side of the illustration).

Figure 1. A comparison between modeled and observed trends in the average temperature of the lower atmosphere, for periods ranging from 10 to 32 years (during the period 1979 through 2010). The yellow is the 5-95 percentile range of individual model projections, the green is the model average, the red and blue are the average of the observations, as compiled by RSS and UAH respectively (adapted from Santer et al., 2011).

Santer et al. take comfort in this Figure that the average of the observed trends falls within the spread of individual model projected trends of similar length—and are further comforted when considering the myriad influences listed above.

We, however, interpret it to show that over all time-scales from 10 to 32 years, the observed trends in the lower atmosphere consistently fall beneath the model projected trends. And that as the length of the observed trend increases, the consistency with the climate model projections decreases.

Just how much more evidence do you need that climate models are projecting too much warming? Give us all the excuses that you want, but if the excuses are real, then they are important drivers of the climate and need to be considered when offering up future climate projections (and quite possibly have an important impact in climate sensitivity determinations).

The fact of the matter is, that the climate projections offered up thus far, have been, and continue to be, sizeable overestimates of reality.

Consequently, we see no compelling reason why we should bank on scenarios for the future that have been produced from the same set of climate models.

At some point, chivalry becomes chicanery.


Santer, B.D., et al., 2011. Separating Signal and Noise in Atmospheric Temperature Changes: The Importance of Timescale. Journal of Geophysical Research, in press.

World Climate Report, 16 September 2011