From UC Berkeley Earth comes this paper that tries some new statistical techniques to get “the pause” to go away, following on with the infamous Karl et al paper of 2015, that played tricks with SST measurements done in the 40’s and 50’s to increase the slope of the warming. This aims to do the same, though the methods look to be a bit more sophisticated than Karl’s ham-handed approach.
The paper link is below, fully open sourced. I invite readers to have a look at it, and judge for yourselves. Personally, it looks like ignoring the most current data available for 2016, which has been cooling compared to 2015, invalidates the claim right out of the gate.
If a climate skeptic did this sort of stuff, using incomplete data, we’d be excoriated. yet somehow, this paper using incomplete data gets a pass by the journal, and publishes with 2015 data at the peak of warming, just as complete 2016 data becomes available.
The results section of the paper say:
From January 1997 through December 2015, ERSSTv3b has the lowest central trend estimate of the operational versions of the four composite SST series assessed, at 0.07°C per decade. HadSST3 is modestly higher at 0.09°C per decade, COBE-SST is at 0.08°C per decade, whereas ERSSTv4 shows a trend of 0.12°C per decade over the region of common coverage for all four series. We find that ERSSTv3b shows significantly less warming than the buoy-only record and satellite-based IHSSTs over the periods of overlap [P < 0.01, using an ARMA(1, 1) (autoregressive moving average) model to correct for autocorrelation], as shown in Fig. 1. ERSSTv3b is comparable to ERSSTv4 and the buoy and satellite records before 2003, but notable divergences are apparent thereafter.
What’s missing? Error bars showing uncertainty. Plus, the data only goes to December 2015. They’ve missed an ENTIRE YEAR’s worth of data, and while doing so claim “the pause” is busted. It would be interesting to see that same graph done with current data through December 2016, where global SST has plummeted. Looks like a clear case of cherry picking to me, by not using all the available data. Look for a follow up post using all the data.
