A major error in an alarming study published in Nature on October 31 suggesting that “ocean warming is at the high end of previous estimates, with implications for policy-relevant measurements of the Earth response to climate change, such as climate sensitivity to greenhouse gases and the thermal component of sea-level rise.” How much higher? Using a novel technique to measure the accumulation of heat in the oceans, Princeton geoscientist Laure Resplandy and her team calculated that the amount of heat being absorbed by the oceans is more than 60 percent higher per year than the estimates offered by the United Nations’ Intergovernmental Panel on Climate Change in 2014.
However, British climate researcher and statistician Nicholas Lewis re-crunched the numbers in the study and found that Resplandy and her team had made significant errors in their calculations. I noted in my reporting on the controversy that I had reached out to Resplandy and had not heard back from her or her colleagues yet. I added that I expected that she and her colleagues need time for a careful evaluation of Lewis’ arguments. Now co-author Scripps Institution of Oceanography climate scientist Ralph Keeling has acknowledged that Lewis is at least partially right and the reseachers are preparing a correction to their original article (apparently not yet published).
The San Diego Union-Tribune article, “Climate contrarian uncovers scientific error, upends major ocean warming study,” is reporting that Keeling now accepts that Lewis is right and that the study’s findings are far more uncertain than they had claimed in their Nature article. “When we were confronted with his insight it became immediately clear there was an issue there,” Keeling said to the Union-Tribune. “We’re grateful to have it be pointed out quickly so that we could correct it quickly.”
“Our error margins are too big now to really weigh in on the precise amount of warming that’s going on in the ocean,” Keeling said. “We really muffed the error margins.”
But getting the error margins wrong is not the only problem with the research, suggests Lewis. Keeping in mind that the correction has not yet been peer-reviewed or published, Lewis responded to an email asking for his reaction to these developments:
In general terms, if [Keeling] is only saying that they acknowledge that their study underestimated the uncertainty in their ocean heat uptake estimate, that is not enough. They should also acknowledge that another consequence of their mishandling of the treatment of uncertainty was that their central estimate of ocean heat uptake was overstated by approximately 30%.
So far as I can tell (from statements on their websites), the authors hope to alter an assumption that affects one aspect (that relating to constancy of the land oxygen-carbon exchange ratio) of the input data used to derive their ocean heat uptake estimate, in such a way that will increase its level, when correctly calculated, to a value close to their originally published estimate. It would seem a little surprising that a valid adjustment made after publication happened, conveniently, to have the effect of almost cancelling a statistical methods error.
Unfortunately their work involves many assumptions where there scope for subjective choices by the authors, so it is difficult to validate those assumptions. I would hope that Nature will have any changes made by the authors to their assumptions examined carefully by peer reviewers who are experts in the same field as Resplandy and Keeling, as well as by statistically expert peer reviewers. However, the failure of the original peer review and editorial process to pick up the fairly obvious statistical problems in the original paper do not engender confidence in Nature’s approach.
The upshot, says Lewis, is that “If you calculate the trend correctly, the warming rate is not worse than we thought – it’s very much in line with previous estimates.”
Of course, peer review is not perfect. However, cynical folks might be forgiven for thinking that this ocean warming paper is an example of a study confirming the prevailing scientific shibboleths being subject to far less scrutiny than those that might challenge them.