Sunday, February 12, 2017

Too fast or too slow?

The discussion around Karl et al., 2015 continues in certain segments of the media.

Last week, the allegations were that the paper was rushed to influence the Paris agreement. That the multi-year multi-party talks could be swayed by a single paper is of course pure bunkum (as an aside Bunkum arises from the behaviour of a politician from Buncombe County where NCEI is located). The Paris climate agreement was the culmination of thousands of individual studies and painstaking, careful, assessments of a wealth of scientific evidence that led to a conclusion by the governments of the world that action was imperative. That any single study was, or could ever be, a ‘clincher’ is pure fantasy. The weight and breadth of evidence is what convinced all parties and is the work of many thousands of experts.

This week, there has been a volte face (about face) and the allegation instead is that Karl et al. led to a delay in the release of ERSSTv4 which was ‘unacceptable’. You’ll forgive me for a second while I sweep up the remains of my shattered irony-meter strewn about over the floor.

Right, where was I? Well, firstly, the new allegation fundamentally clears up one allegation in that the ERSSTv4 product was in no sense experimental and had undergone full internal review as well as having (at the time) two peer-reviewed published papers describing it.

Huang et al. put out a technical paper, published in the Journal of Climate in February 2015, on the new dataset, accompanied by a paper describing its uncertainties by Liu et al.  The Huang et al. paper covered in detail the dataset specification.  It did not address the implications of the dataset related to temperature trends over the past several decades. It was a highly technical paper focused on explaining the details of new corrections to sea surface temperatures. It was not intended for general public communication.  

At the same time, NOAA and ISTI had in late 2014 released a major upgrade to monthly global land data holdings source that is updated each month in NOAA’s regular global temperature monitoring.  It has extra stations in many regions of the world and improved coverage , and that is why Karl et al. took advantage of this for a snapshot data set analysis that ended in 2014. The envisaged upgrade to GHCNM will regularly add new data and process these data through the existing land analysis processing suite.

Whenever NOAA puts out new data sets, that will be updated each month for tracking global temperatures they always get numerous questions that they have to be able to answer. So, NOAA waited on releasing this new ERSSTv4 data set until the Karl et al. paper was out because the Karl et al. paper presented the implications of the new corrections and new land data sources to previously reported trends appearing in prominent works such as the Fifth IPCC assessment including, but not limited to, the recent behaviour. NOAA took extra steps to ensure the impact of the new corrections and data could be readily explained. It took an extra 4 months to accomplish this. If the Karl et al. paper had not been in the pipeline, arguably NOAA would have had to write up a stand-alone paper to explain the implications of the updated data set, essentially the “Karl et al. paper.”

And there is still nothing in these process complaints that substantiates anything but high-quality science that has already been reproduced by other scientists in peer-reviewed scientific articles. Those studies validated NOAA's work. They validated the science, the data provided to them by NOAA, and the quality of the work.


Phil said...

The allegation, as far as I can see from David Rose's twitter feed, is that the withholding of ERSST.v4 data from release in January 2015 allowed the data to be subsequently "manipulated" to come to the predefined conclusions of Karl et al. This is despite the fact that, as you say, the methodology paper was to be published the next month and also that, as I understand it, Karl et al. had already been submitted for publication in December 2014.

Given this timeline, it seems such a bizarre accusation, but then again it is made by someone working to the tight deadlines of a tabloid who appears to feel no need to spend any time fact checking.

Vaytw said...

There is a new line going around (see here ) that the scientists lost some data because of a computer failure. Can't confirm it anywhere. Do you know anything about this?

PeterThorne said...


if this is the data loss in the initial accusation then John Bates has walked it back and everyone else has denied it. The data from the paper are still on the ftp server when I looked a week or so ago (on the road so unable to check now) as are the underlying data in their respective pages.

Post a Comment

Please note all comments are moderated. Comments containing profanities, unwarranted accusations, deemed off-topic etc. shall be deleted. There may be a substantial delay between posting a comment and its acceptance owing to moderator availability.