Skip to content

Berkeley Earth Surface Temperature: Good Data, Poor Conclusions

Dr David Whitehouse

The second series of papers from the BEST consortium analysing global land temperatures have been published. They contain some interesting analyses, and the good thing is that all the working and code is posted online even if the papers have not yet passed peer-review! It includes the first analysis of global land temperature 1750 – 1850. However, the papers suffer from being too ambitious in the face of sparse data, and the conclusions reached by Professor Muller in this accompanying article in the New York Times is too far reaching considering the data he has been studying, and the analysis he presents.

Considering only land data, as he did in the original BEST study, he extends his analysis back to 1753 concluding that over the past 250 years there has been a 1.5 deg C increase and 0.9 deg C over the past 50 years. This is a stronger statement than the IPCC makes. It is also considerably larger than combined land-sea data.  Hadcrut3 or Hadcrut4 data gives 0.4 deg C since 1980 and no increase between 1940 – 80. However, looking at the errors in Muller’s temperature estimates (which should have been included in the New York Times article) we see that the increase in temperature is between 0.1 and 2.9 deg C.

Post-1850 the new analysis is not different from the previous one. Professor Muller says that the flattening of the past decade or so is not statistically significant. In the first analysis he said it was not seen in the data, which is a somewhat different thing from seen and not significant. Note that our own analysis shows that it is significant.

By comparing the temperature curve to that of rising CO2 Muller concludes that; “Humans are almost entirely the cause.” This is not a wise statement to arrive at based on curve fitting and lining up two variables. Consider the amount of CO2 in the atmosphere between 1750 – 1850. Click on image to enlarge.


Previous work, based on many more parameters, have concluded that anthropogenic CO2 is an important factor in the past-50 years, but not before.

Climatologists know there are other significant influences over the timescales considered in this analysis that Muller has not considered. Muller’s statement that the Sun’s role since the Little Ice Age (LIA) is insignificant is not shared by many other scientists who have studied the cause of and the emergence from the LIA.

Large Uncertainties

In addition it is clear that the pre-1850 uncertainties in land temperatures are very large and quantitatively and qualitatively different from the more modern parts of the land data. A very interesting part of the new analysis, and in my opinion its main finding, is the interplay of intense volcanic activity and temperature seen prior to 1850. There is a clear case of a regularity there of about 25 years that is unexplained. It is indeed curious that whilst there are named volcanic eruptions in the first 150 years of the graph they seem to be far more powerful than recent ones. Click on image to enlarge.


In conclusion, the new analysis tells us nothing new about global temperatures post-1850 (although there is some interesting data about changes to the diurnal temperature range in the 20th century) and its claims of attribution due solely to mankind is unsupportable. There is some fascinating data about volcanic eruptions and global temperatures prior to 1850. Click on image to enlarge.