July 18, 2004

Another Model Error

Filed under: Climate Models

Climate models are not capable of identifying the daily temperature range—as yet another new study demonstrates. So why are the world’s leaders willing to put so much stock into these models’ predictions of our future climate?

Climate models are unable to replicate the observed behavior in the patterns of changes in the daily temperature range. Instead, the models warm daily high temperatures too fast—probably because they don’t account properly for cloud dynamics.

Predicting what our future climate will be like is one use of climate models. But some scientists prefer to look at past data for clues to future climate change. We look at observations for evidence of a greenhouse-gas signal—something that confirms the effect of increasing greenhouse gas concentrations.

One of the most prominent greenhouse-gas signals is the diurnal temperature range (DTR)—the daily maximum temperature minus the daily minimum temperature. (Basically, it’s how warm it gets in the daytime minus how cool it gets at night.) Over most of the globe’s land regions, that range has been declining over time—and the decline is thought to be a global warming indicator. Both maximum and minimum temperatures are rising, but the rise in daily low temperatures has occurred at a much greater rate, so the temperature range has gotten narrower.

This trend is related to increasing greenhouse gas levels because, everything else being equal, an atmosphere with higher greenhouse gas concentrations will raise nighttime temperatures during the time when the sun is down and the planet is cooling off because of the earth’s ability to radiate away heat.

But climate models cannot replicate this effect. Take, for example, a recent study by Australian scientist Karl Braganza and two coauthors from the United States. The authors gathered data from all of the global land areas with sufficiently long periods of record (forcing them to exclude Greenland, Antarctica, part of India, and most of Africa and South America). They then compared the observed global decadal trends in maximum and minimum temperature and DTR to the output of five climate models in which the observed changes in 20th-century greenhouse gas and aerosols levels were simulated.

The results of the comparisons are summarized in Figure 1. Although the climate models, in aggregate, do a good job of reproducing the observed trend in minimum temperature, they overestimate the trend in maximum temperature. Each model does increase the daily high temperatures at a slower rate than the low temperatures. But actual observations show a much slower increase in the daily highs. The net effect of this discrepancy on DTR is that none of the models can properly simulate the observed trend in DTR, which is declining at a rate greater than what the models indicate it should be.


Figure 1. Observed (filled circles) and modeled (open symbols) trend in daily average temperature (Tmean), daily minimum temperature (Tmin), daily maximum temperature (Tmax), and daily diurnal temperature range (DTR) for the period 1951-2000 (source: Branganza et al., 2004).

No one who has been following the climate change debate could possibly be shocked to discover that climate models have exaggerated the warming rate. But the critical issue here is that, since DTR really is a climate change indicator, the models could be mischaracterizing some very fundamental process that is key to being able to accurately model our climate at all. In that case, the models can hardly distinguish between the rates of day vs. night warming, while high temperatures are not increasing nearly so quickly as the models predict.

This time, the fly in the ointment seems to be cloud cover. Cloud cover over land areas increased over the last half of the 20th century. Cloudy afternoons are generally cooler than clear afternoons, so clouds could account for this large discrepancy between climate models and reality.

Of course, you could argue that you really can’t model earth’s climate without getting cloud cover correct, since clouds have an awful lot to do with both planetary temperature and precipitation. You could even argue that, because of this cloud problem, the model’s might be getting the trends in minimum temperature correct by dumb luck given that the fundamental physics are not correct (see our previous WCR for more evidence that dumb luck may play a large role in observed vs. model comparisons).

The bottom line? Over global land areas, nighttime low temperatures are rising faster than daytime highs, and this trend is consistent with increasing greenhouse gas levels. Climate models are (currently) incapable of correctly reproducing the observed trends, and as a result are increasing daytime high temperatures faster than reality. This error is present in all likelihood because the models have not properly captured some fundamental physical component of earth’s climate.

So can someone explain to us why so many politicians from so many nations have put so much political stock into these models’ predictions of our future climate?


Braganza, K., D.J. Karoly, and J.M. Arblaster, 2004, Diurnal temperature range as an index of global climate change during the twentieth century, Geophysical Research Letters, 31, L13217, doi:10.1029/2004GL019998, 2004.

No Comments

No comments yet.

RSS feed for comments on this post. TrackBack URI

Sorry, the comment form is closed at this time.

Powered by WordPress