How the “greater 50% of warming since 1950 is human caused” claim is deeply flawed

Over at Cliscep, Jaime Jessop has rather jokingly raised a central claim of the IPCC Fifth Assessment Report, after someone on Twitter had accused her of not being a real person.

So here’s the deal: Michael Tobis convinces me, on here, that the IPCC attribution statement is scientifically sound and it is beyond reasonable doubt that more than half of the warming post 1950 is indeed caused by emissions, and I will post a photo verifying my actual existence as a real person.

The Report states (AR5 WG1 Ch10 Page 869)

It is extremely likely that human activities caused more than half of the observed increase in GMST from 1951 to 2010.

This extremely likely is at the 95% confidence interval and includes all human causes. The more specific quote on human greenhouse gas emissions is from page 878, section “10.2.4 Single-Step and Multi-Step Attribution and the Role of the Null Hypothesis

Attribution results are typically expressed in terms of conventional ‘frequentist’ confidence intervals or results of hypothesis tests: when it is reported that the response to anthropogenic GHG increase is very likely greater than half the total observed warming, it means that the null hypothesis that the GHG-induced warming is less than half the total can be rejected with the data available at the 10% significance level.

It is a much more circumspect message than the “<a href=”http://stocker IPCC 2013″ target=”_blank”>human influence on the climate system is clear</a>” announcements of WG1 four years ago.  In describing attribution studies, the section states

Overall conclusions can only be as robust as the least certain link in the multi-step procedure.

There are a number of candidates for “least certain link” in terms of empirical estimates. In general, if the estimates are made with reference to the other estimates, or biased by theory/beliefs, then the statistical test is invalidated. This includes the surface temperature data.

Further, if the models have been optimised to fit the surface temperature data, then the >50% is an absolute maximum, whilst the real figure, based on perfect information, is likely to be less than that.

Most of all are the possibilities of unknown unknowns. For, instance, the suggestion that non-human causes could explain pretty much all the post-1950 warming can be inferred from some paleoclimate studies. This reconstruction Greenland ice core (graphic climate4you) shows warming around as great, or greater, than the current warming in the distant past. The timing of a warm cycle is not too far out either.

In the context of Jaime’s challenge, there is more than reasonable doubt in the IPCC attribution statement, even if a statistical confidence of 90% (GHG emissions) or 95% (all human causes) were acceptable as persuasive evidence.

There is a further problem with the statement. Human greenhouse gas emissions are meant to account for all the current warming, not just over 50%. If the full impact of a doubling is CO2 is eventually 3C of warming, then from that the 1960-2010 CO2 rise from 317ppm to 390ppm alone will eventually be 0.9C of warming. Possibly 1.2C of warming from all sources. This graphic from AR5 WG1 Ch10 shows the issues.

The orange line of anthropogenic forcing accounts for nearly 100% of all the measured warming post-1960 of around 0.8C – shown by the large dots. Yet this is about 60% of the warming in from GHG rises if a doubling of CO2 will produce 3C of warming. The issue is with the cluster of dots at the right of the graph, representing the pause, or slow down in warming around the turn of the century. I have produced a couple of charts that illustrate the problem.

In the first graph, the long term impact on temperatures of the CO2 rise from 2003-2012 is 2.5 times that from 1953-1962. Similarly, from the second graph, the long term impact on temperatures of the CO2 rise from 2000-2009 is 2.6 times that from 1950-1959. It is a darn funny lagged response if the rate of temperature rise can significantly slow down when the alleged dominant element causing them to rise accelerates. It could be explained by rising GHG emissions being a minor element in temperature rise, with natural factors both causing some of the warming in the 1976-1998 period, then reversing, causing cooling, in the last few years.

Kevin Marshall