Over the past few years, I have written about problems with the data related to global temperature measurements (see here and here). It is very difficult to get a handle on how the planet’s temperature has changed over the past century or so, because the only long-term data we have come from thermometers that are placed at various spots throughout the world. Since 1979, we have much more accurate global temperature data, which come from satellite measurements. However, those satellite measurements are not consistent with the thermometer measurements.
This is an important issue, because climate models (which make projections about future temperatures based on different emission scenarios) are “calibrated” against the known temperature data in an effort to make them more realistic. Since the satellite data have only been collected since 1979, they are rarely used. Instead, the longer temperature record (based on thermometers) is generally preferred. The two commonly-used thermometer records are GISS TEMP (maintained by NASA) and HadCrut4 (maintained by the University of East Anglia and the UK Met Office). Those two data sets are in good agreement with one another, but once again they do not agree with the satellite data.
Are these thermometer data reliable? Based on the PhD thesis of John D. McLean at James Cook University, the answer is “no.” He did what he claims is the first audit of the reliability of the Hadcrut4 data, and he has found 25 areas of concern. I will discuss only three. First, he finds many instances of anomalous data. One station in Colombia, for example, reports that the 1978 average monthly temperatures in April, June, and July were 81.5 oC, 83.4 oC, and 83.4 oC. In case you aren’t familiar with the Celsius temperature scale, that’s about 180 oF. Given that the highest temperature ever recorded on earth was 134 oF, it’s safe to say that the report from Colombia is simply wrong. He lists many other examples of anomalous data that cannot possibly be correct.
Second, he finds that for the vast majority of the record, the global coverage was incredibly poor. From 1850 to 1852, for example, there was only one land station (in Indonesia) and a few ships traveling a trade route that reported temperature data for the entire Southern Hemisphere. As time went on, coverage got better, but as he writes:
Even as a notional concept the global coverage has varied greatly over time, being less than 50% of the Earth’s surface for about one-third of the data record from 1850 to 2015 and only having been greater than 75% for only the last 60 years (since 1956). Even at the end of 2015 there was no coverage of 16.6% (i.e. 1/6th) of the Earth’s surface.
As a measure of global temperature, then, Hadcrut4 is only valuable for the last 60 years or so, at best.
Finally, what I see as the biggest problem he found is something that has also been seen in the GISS TEMP data as reported through the NOAA. In order to account for various changes that occur over time, temperature data are “adjusted” to make them more homogeneous. There is nothing wrong with doing that, but as you can see from the image at the top of the page, the adjustments tend to affect the earlier data more than the later data. In the example shown above (for seven stations from New Zealand) the effect is to make the past cooler relative to today. As far as I can tell, there is no justification for that. The author seems to agree:
There is no reasonable explanation of why earlier data should need a greater adjustment because temperature observations should have the least interference and distortion when the station commences operation at any given site.
In other words, the adjustments are supposed to take into account problems that develop after the station is operational, such as heat-trapping buildings being constructed near the station. Thus, adjustments should affect the recent data more than the past data, but this audit found the opposite. The problem, of course, is that only the adjusted data are used by climate scientists trying to understand global climate. So climate scientists are using data that overemphasize any warming that has happened over the past 150 years.
The author does more than simply list his many (justified) criticisms of the Hadcrut4 data set. In Section 8.7 of his thesis, he proposes a method to construct the best possible data set for global temperatures. I am not sure I am qualified to assess the value of his proposal, but based on what he has found in his audit, something must be done.