Readers were kind enough to point to the newest revision of global temperature data from the Hadley Centre/Climate Research Unit in the U.K., the HadCRUT5 data (a revision of the HadCRUT4) data.
The HadCRUT4 data were in disagreement with the other surface temperature data sets, namely those from NASA, Berkeley Earth, and NOAA. But the new HadCRUT5 data set agrees with them excellently:
All except HadCRUT4 show evidence that global temperature increased faster since the year 2000 than it did during the interval 1980-2000, but that the evidence doesn’t reach “statistical significance.”
At least, not with the data as is. However, much of the fluctuation is due to known causes, so we can estimate it and remove it for the HadCRUT5 data, just as we did for the others. Then we can compare the warming rate 1980-2000 to that since 2000, like this:
The evidence does reach statistical significance for NASA, Berkeley, NOAA, and HadCRUT5 as well. It does not do so for HadCRUT4, RSS, and UAH.
A note: someone asked about the error bars in the graph like this one, for the RSS and UAH data (satellite data of lower-troposphere temperature). I had indeed plotted the wrong analysis for those two data sets; I’ve corrected it in the above graph.
Other note: before adjustment, the error bars for satellite data are much bigger than for surface data. This is because they respond so much more dramatically to those known fluctuation factors, el Nino and volcanic eruptions. When those are removed, the uncertainty in trend from satellite data is comparable to that from surface data.
This blog is made possible by readers like you; join others by donating at My Wee Dragon.