At least three-quarters of the climate warming observed since the mid-twentieth century is extremely likely to be caused by human activities. That’s what Reto Knutti and Markus Huber from the Swiss Federal Institute of Technology, Zurich, said last week, confirming previous similar conclusions made using a different method. They say that it is “extremely unlikely” that the temperature changes we have seen are caused by natural variation, even if that variation is much stronger than current models suggest.
The Swiss researchers are clearly not the first to find humans are very likely to be responsible for global temperature rises. However, as a contributor to previous reports by the UN Intergovernmental Panel on Climate Change (IPCC), Knutti felt that the methods used to establish responsibility could be better. “Attribution of past changes was largely presented as a result of “optimal fingerprinting”, and that is a black box for most people,” he told Simple Climate. “It’s statistically complex, makes a number of assumptions and is not physically intuitive.”
The optimal fingerprinting method that previous studies used looks at different measured “forcings” that alter energy flows in Earth’s climate using complex computer models. First, the models simulate patterns caused by energy trapped in the atmosphere by greenhouse gases, for example. Then they go on to simulate patterns caused by other forcings, like the energy reaching the Earth from the Sun.
When each forcing has been considered, all the simulated patterns are then combined to match patterns seen in real life using a statistical technique called regression analysis. This effectively adds the patterns together, after multiplying each by a number known as a scaling factor. The size of the different scaling factors can be used to work out how much of the changes that are seen comes from each forcing. However, this approach relies on models’ ability to accurately simulate warming patterns through time and space. It also does not automatically provide a balance of energy flows that matches reality.
Knutti and Huber’s method, by contrast, is based on running a much simpler model thousands of times with slightly different settings, or parameters. “Then we compare these with the observed warming and ocean heat uptake, and keep the models that are consistent with what’s actually seen,” Knutti said. “We can then run the model with each of those parameter sets for each forcing individually, for example looking at CO2. That gives us a best estimate and uncertainty for the CO2 contribution, and with the other forcings.”
In the models they kept, CO2 and other greenhouse gases in the atmosphere strongly drove temperature increases. However, there was also added cooling from solid and liquid particles in the atmosphere, known as aerosols, which include clouds and sulphur coming from human pollution and volcanoes. “Greenhouse gases alone would have warmed the Earth’s surface more than was observed, and are partly offset by aerosols,” Knutti said. Greenhouse gases contributed 0.85ºC of warming, and aerosols reduced this to give the 0.56ºC increase in average temperature seen since 1950, the scientists found.
Separately, Knutti and Huber could work out how much of the temperature increase we’ve seen is down to natural variability by using simulations of a set of more complex models with no extra forcing. They found less that a one-in-twenty chance that it contributed more than a quarter of the warming since 1950, or more than a fifth of the warming since 1900. In a paper in research journal Nature Geoscience published last week, they wrote that even if natural variability were three times larger than their models said, it would still be extremely unlikely to cause the temperature rises seen.
Eliminating models that didn’t match measured warming and ocean heat means that their approach represents real energy flows much more accurately. As Knutti explained, that’s exactly what he wanted to do. “The fact that we have observed significant changes in both the forcing and surface and ocean response suggested that the energy balance would tell us a lot about the causes of past warming,” he said. “At least to a physicist, conservation of energy is fundamental.”
And having a way to assess climate change causes that is more firmly rooted in physical processes backs the conclusions of earlier efforts, Knutti added. “The fact that our results are entirely consistent with the optimal fingerprinting, but based on a different method, model and type of observation, is an argument for even higher confidence in human induced climate change,” he said. “If optimal fingerprinting tells us that human influence is strong, that is OK, but if we can understand that in terms of physical processes, the greenhouse effect, and if it is measurably consistent with the external forcing and observed energy balance response, that makes a much stronger argument.”