Ross McKitrick: The Flaw In Relying On Worst-Case Climate Modelling
The purpose of global climate policy is to get us from the dangerous upper end of the forecast range down to the safe bottom end. In fact, we are already there.
Whenever you read a media story about how we’re heading toward catastrophe if we continue operating “business as usual” — i.e., if we don’t slash carbon emissions — the reports are almost always referring to a model simulation using RCP8.5. And you can bet that nowhere in the story will they explain that RCP8.5 is an implausible worst-case scenario that was never meant to represent a likely base case outcome, or that scientists have begun castigating its usage as a prediction of a doomed business-as-usual future.
The term RCP8.5 refers to a greenhouse gas emissions scenario often used by scientists for climate model projections. You might never have heard of RCP8.5 but you have definitely heard of forecasts based on it. Listening to the politicians who make the strongest pleas for radical climate action, it is clear that their fears for the future are driven by RCP8.5 scenarios, yet it is also clear that they have no idea what it is or what is wrong with it.
RCP stands for “Representative Concentration Pathways,” or projections of how much carbon dioxide (CO2) will accumulate in the atmosphere due to fossil fuel use over the coming century. The United Nations’ Intergovernmental Panel on Climate Change (IPCC) generated a set of four RCP scenarios a decade ago, attaching to each a number indicating how much “radiative forcing” (a measure of global warming potential) each one generates. RCP2.6 refers to a benign, low-end emission scenario with correspondingly minimal radiative forcing. In the middle are RCP4.5 and RCP6.0, and at the top end is RCP8.5, a scorcher that predicts historically unprecedented increases in global CO2 emissions.
To appreciate how implausible RCP8.5 is, consider its coal use trajectory. From the 1920s to the year 2000, global coal consumption stayed between 15 and 20 gigajoules per capita, peaking at 20 in 1960, falling back to 15 by 2000, then rising to about 23 earlier this decade with the sudden industrialization of China and India. Groups like the International Energy Agency expect it will gradually return to the 15-20 gigajoule per capita range by 2040.
The RCP8.5 scenario offers a different outcome. Instead of a return to normal, it projects coal use will rise to about 30 gigajoules by 2040, 45 gigajoules by 2060 and 70 gigajoules by 2100. No one seriously believes this is even possible, including people who use RCP8.5 in their climate simulations.
It gets worse. A recent study by Matthew Burgess of the University of Colorado and coauthors, which is currently available in preprint form, pointed out that RCP8.5 doesn’t even make sense in its own modelled reality. It projects so much economic growth that today’s poor countries will be richer in 2100 than the wealthiest countries are today (which would be nice if it happened), but they will also experience so much warming that they become uninhabitable wastelands. How can both be true?
RCP8.5 was created as an outlier; an improbable worst-case scenario, not a likely business-as-usual forecast. Yet countless scientists and economists have been using it as one. You know how the game works: feed RCP8.5 into a climate model, observe the catastrophe, then call it the “likely” scenario if we don’t cut emissions.
A more realistic business-as-usual scenario would look much more like the low end of the RCP range. If you run a model with one of those, the future looks far less worrisome, the policy agenda is far less urgent and your study is far less likely to get any media attention. Which may be why so many modellers prefer using RCP8.5. But last fall, in a commentary in Nature magazine, climate experts Zeke Hausfather and Glen Peters scolded their colleagues for misleading the public this way, and distorting the policy debate in the process.
Exaggerated emission forecasts are nothing new. Another analysis last fall by Zeke Hausfather et. al. compiled CO2 concentration forecasts from the ’70s onward. The figure shown on this page is based on their data. In the ’70s, scientists made CO2 projections through 2000. Reality came in near the bottom end. And from the ’80s on, reality came in right at the bottom end.
The IPCC likewise has a history of playing up worst-case scenarios. Prior to commissioning the RCP group, in 2000 the IPCC issued the Special Report on Emission Scenarios (SRES), a set of emission paths ranging from low and slow CO2 growth to the infamous A1FI top-end scorcher, which climate modellers considered unrealistic at the time but which allowed the IPCC to issue scary-sounding “up to six degrees” warming forecasts. In a column I published on this page on April 4, 2002, I noted that by the time it was shown to experts in early 2000, A1FI had already overestimated global coal consumption growth in the ’90s by 40 per cent.
The SRES scenarios came in for more controversy in 2002 when Ian Castles, the former chief statistician of Australia, and former OECD chief economist David Henderson wrote to the IPCC to inform it that the SRES team had used an incorrect formula for computing foreign exchange rates, which systematically exaggerated the growth and emissions forecasts in low-income countries.
In my own academic research I have also found IPCC emissions scenarios to be exaggerated. In two papers published in 2012 and 2013, I and my coauthors compared the range of IPCC emission scenarios to historical trends and found the upper half of the range to be improbable compared to the lower half.