Britain’s Met Office welcomes audit by Australian researcher about HadCRUT errors, by Graham Lloyd.

Britain’s Met Office has welcomed an audit from Australian researcher John McLean that claims to have identified serious errors in its HadCRUT global temperature record.

The audit said the HadCRUT temperature record, the main global temperature set used by climate models, showed exaggerated warming and was not fit for global studies. …

Dr McLean’s audit found data prior to 1950 suffered from poor coverage and very likely multiple incorrect adjustments of station data. …

Anomalies identified in the McLean paper include at St Kitts in the Caribbean, the average temperature for December 1981 was zero degrees, normally it’s 26C.

For three months in 1978 one place in Colombia reported an 82 degrees Celsius average — hotter than the hottest day on Earth.

In Romania one September the average temperature was reported as minus 46°C.

Sometimes ships would report ocean temperatures from places up to 100km inland.

If they’re that careless with the data, how can we trust them? More importantly, the adjustments they make exaggerate the trends — which inform the climate models.

The Met Office didn’t dispute any of McLean’s charges, and avoided mentioning the temperature adjustments at all.

By the way, Joanne and I helped promote and publicize this first ever audit of the temperature data. (Why hasn’t it been audited before?) We helped set up set up Robert Boyle Publishing to distribute it.

From the press release:

Freezing Tropical Islands? First global temperature audit finds 70 major errors

The fate of the planet is at stake, but the key temperature data set used by climate models is riddled with mistakes. Governments have spent billions of dollars because of predictions based on this data — yet even the simplest of quality control checks have not been done.

At St Kitts in the Caribbean, the average temperature for December 1981 was zero degrees, normally it’s 26C. Thus, it would have been below zero every night in the tropics that month, which makes no sense. For three months in 1978 one place in Colombia reported an extraordinary 82 degrees Celsius average – hotter than the hottest day on Earth.  While in Romania the average temperature one September was reported as minus 46°C. In other comic mistakes, sometimes ships would report ocean temperatures from places up to 100km inland.

The dataset is the main global temperature set used by the IPCC and climate models are tuned to fit it. Despite that, it has never been properly audited. In an extension of his PhD thesis, Dr John McLean examined the HadCRUT4 data managed by the UK Met Office Hadley Centre and the Climate Research Unit at the University of East Anglia.

One of the most serious flaws he identified is the shortage of data. For the first two years, from 1850 onwards, the only reporting station on land in the Southern Hemisphere was in Indonesia. According to the HadCRUT4 calculation of coverage it was almost 1950 before we had data from even half of the Southern Hemisphere.

Dr McLean points out how influential this data is: “The draft agenda for this year’s UN COP24 proposes that the HadCRUT4 average from 1850 to 1899 be used as an ‘indicative’ temperature. This is just nonsense; the average global coverage of the dataset from 1850 to 1899 was just 30%. In May 1861, the global coverage was just 12%”.

Probably the worst systematic error is the practice of correcting data after a site moves. Thermometers are usually placed in open urban spaces but buildings and roads are gradually constructed around them which can make sites artificially warmer. Replacement thermometers are placed further out of town, in cooler locations. All the older data is adjusted downwards, or made cooler, to match what is being recorded at the new site. While the adjustment might be correct for recent data, older data will often be overcorrected which makes it colder than it should be, and makes warming seem greater.

McLean explains:  “The data for some stations might have been incorrectly adjusted five or more times, making the oldest data much colder than it should be.  This is another reason why the early HadCRUT4 data is very unreliable and why it shouldn’t be used for the Paris Climate Agreement.”

The 135-page audit and the data files it’s based on are available from Robert Boyle Publishing (http://www.robert-boyle-publishing.com).

hat-tip Stephen Neil