Conclusion of series shows IPCC’s computer models fall way short
First, a bit of arcane background: In a recent Climate Audit blog discussion, U.K. climate modeller Myles Allen commented that the Muir Russell inquiry into Climategate had found no evidence of contamination of the so-called “instrumental record,” or surface-temperature data. He explained the importance of this by pointing out that “we all use the instrumental temperature record all the time …. If there had been anything wrong with the instrumental record, I would have to retract or redo a huge number of papers. It turned out there wasn’t.”
In our subsequent exchange, Allen conceded that a statistical argument relied upon by the Russell inquiry was not one he would recommend his students use, but nevertheless he still gave them the benefit of the doubt on their conclusion that evidence of contamination of the surface-temperature record can be safely ignored.
So what’s all this about? The statistical evidence in question can be thought of as a rival explanation of climate change over land since 1979. It contrasts to the approach Allen and his many colleagues have pursued for several decades. They have built about two dozen large computer systems called General Circulation Models (GCMs), which represent the behaviour of the global climate largely on the assumption that greenhouse gases play the dominant role in climate change. These models underpin the conclusions of the Intergovernmental Panel on Climate Change (IPCC) about the role of greenhouse gases in 20th-century warming, and its forecasts for much more warming in the future.
The rival model explains patterns of warming over land as a result of urbanization and the varying patterns of socioeconomic and industrial development. A series of studies over the past decade have shown this hypothesis to have significant explanatory power, even though it has nothing to do with greenhouse gases.
Of course, both models might be partly right. But the IPCC has taken an extreme position, that the socioeconomic patterns have no effect and any temperature changes must be due to global “forcings” like carbon dioxide emissions. Studies that claim to detect the effects of CO2 emissions on the climate make this assumption, as do those that estimate the rate of greenhouse gas-induced warming. As Allen says, if this assumption isn’t true, there are a lot of papers that would have to be retracted or redone.
In a new paper with my colleague Lise Tole of Strathclyde University, just published in the journal Climate Dynamics, we took the socioeconomic data and evaluated it alongside the 22 GCMs used by the IPCC in its last assessment report whose data were available online. We examined which approach best explains the spatial pattern of temperature trends over the past few decades: climate models or socioeconomic data. The answer is you need both, especially the latter.