My Enquiry to GHCN

New commenter Colin asks:

So what is the purpose of changing the data?

We can only speculate about that. The only people who can provide a definitive answer are the curators of the data at GHCN. I have sent them the following email and eagerly await their reply.


Dear Sir / Madam,

By way of brief introduction I am a British Energy and Climate blogger based in Aberdeen Scotland. I have taken an interest in recent press and blog reports about modifications being made to temperature records and have been conducting a number of region based comparisons that include Central Australia and Iceland. It is Iceland that I wish to make the focus of this query. My observations can be found at the following link:

Re-writing The Climate History of Iceland

The key points I feel that GHCN need to address in public are as follows:

1) The Icelandic data are apparently very well curated and already adjusted for non-climatic influences. Furthermore homogenisation is unlikely to be effective in an isolated system like Iceland. Can you therefore please explain the rationale for homogenising these stations when it seems quite clear that it would be inappropriate to do so? The effect I believe has been to cool the past creating an impression of more warming than the raw data actually supports.

2) Figure 7 in my post displays the arithmetic difference of the mean temperatures from all 8 Iceland stations. From this it can be seen that centred on 1939 larger than usual adjustments to temperature have been made in a narrow band. This has been achieved by significantly reducing temperatures in this narrow band in the V3.1 data. The raw data indicate that 1939 was the warmest year on record in Iceland and the effect of these adjustments is to remove this temperature peak all together. Can you please explain the justification for these adjustments and how they were made?

3) Staying with Figure 7, it can be seen that there is another narrow band of temperature adjustments centred around 1967, this time in the opposite direction. This time it seems that the adjustments have been achieved in part by systematic deletion of temperature records in this narrow band. See Figure 8 where deleted records are shown in yellow. The effect of this has been to substantially reduce the significance of the well-doucumented cold spell in Iceland at this time. Can you please explain how and why these records came to be deleted?

4) Staying with Figure 8, in green are cells where V3.1 data seems to exist that does not have V2 equivalent data. Given that V3.1 is supposed to be a homogenised derivative of V2, how is it possible to have V3.1 data where no V2 data exists?

There is a fair amount public concern about the veracity of the official temperature records that are used by many governments and the United Nations to guide energy policy and the allocation of $trillions in energy infrastructure. I would therefore be very grateful if you could arrange for a senior official at GHCN to answer the questions detailed above so that public concern may be put to rest.

Yours faithfully

Dr Euan Mearns (Geologist)

This entry was posted in Climate change, Energy, Political commentary and tagged , , . Bookmark the permalink.

22 Responses to My Enquiry to GHCN

  1. bobski2014 says:

    My support may not be worth much (anything) scientifically, but you’ve got it anyway, in spades.

  2. A C Osborn says:

    Euan, yet another study of Ausy Temperatures, this time using historic Published data that shows once again the past was cooled by 0.4C.

    Add to that the known problems with incorrect UHI adjustments and problems with the new electronic Thermometers reading high we have a real mess that virtually proves that “The Warmest Ever Year” is absolute nonsense, along with their error bars.

  3. Graeme No.3 says:

    A form letter won’t be enough so you may have to wait quite some time for an intelligent (and intelligible) reply.

  4. Andrew Thickpenny says:

    Hi Euan. Looking forward to the answer from GHCN.
    FYI, here is an explanation for temperature alteration (including cooling the past) from the skepticalscience website:
    I suspect that you’ve already seen this or something similar.

    • Euan Mearns says:

      Andrew, thanks for this I had a flash scan read. Its really not worth wasting time on that junk. If a new thermometer is introduced that is biased compared to the old, then in any record that is a simple one off step change adjustment to the record when the new instrument is introduced.

      Imagining that pairwise homognenisation across the whole record can correct for that one off change to the instrument record is zombie land.

      And they are several plays behind the main game which is here on EM. In Oz I showed that homognenisation did not in fact introduce significant bias. The big story is why in central Oz and southern Africa there is close to zero warming and the big question is why temperature reconstructions based on flat records create a warming trend.

      • Yvan Dutil says:

        Old solar shield were less efficient than modern one. This lead to higher temperature in the past compare to now. This is pretty important in very sunny place like Australia.

        • Euan Mearns says:

          Yvan, like Kevin’s, your continued involvement here is highly valued. But you are going to have to push the boat out an awful long way to convince me that perceptions of global warming are founded upon corrections for faulty instrumentation.

        • A C Osborn says:

          Yvan, so what do you say about the inadequate UHI adjustments made to the data and also the error found with the new MMTS over-reacting to spikes in temperature, which is NOT corrected for at all.
          So current final temperatures are too high when compared to the old thermometers, whcih are not raised to suit , but lowered evn more.

    • manicbeancounter says:

      Skepticalscience gives a very good explanation of why adjustments are done. But there is nothing about validating whether those adjustments bring the temperature data towards what the temperature measurements would show without the biases, or create additional anomalies in the data. Euan’s letter asks for explanation of adjustments introducing additional patterns in the data (e.g. in 1939), or deletion of certain records.
      For instance, it is important to allow for station moves, but when I compared the frequent station moves in Reykjavik with the adjustments by GHCN, there was no correlation.
      Skepticalscience should try to compare some of the metadata for stations with larger adjustments and see if they can give a good explanation. Euan has given plenty of examples that go either way. For fun I have looked at Isfjord Radio and Svarlbard Luft (the airport.) on Sptizbergen. The former has for 1917 an average warming adjustment of 4.0C, (between GHCN raw and GISS homogenised) yet had -1.7C cooling adjustments in the mid 1950s. The latter data set started in the mid 1970s with a +1.7C adjustment, falling to nil adjustment by around 2000.
      Raw data from
      GISS Homogenised from
      Can anyone find bigger adjustments.

      • Euan Mearns says:

        Kevin, I hope you hang around here a while at least because I see you tour a number of blogs and are trying to bring a degree of objectivity to the debate. I loose patience rather too easily.

        The view I have developed right or wrong is that homogenisation algorithm has to balance. And to make it so extreme values are applied to some stations.

        Try Windhoek in SA, almost 5˚C adjustment.



        • manicbeancounter says:

          I cry foul – the Windhoek example has an average 4˚C adjustment throughout the 65 years of data. 🙂
          My example went from +4˚C to -1.7˚C in 40 years. 🙂
          Actually, is provides a good counter example to your theory of a homogenisation algorithm adjusting to a preconceived view. This idea I was also developing from the Iceland, Paraguay and your many Southern Hemisphere examples.
          The Windhoek example provides evidence of a number of different people constantly tinkering with the data, making adjustments and deleting “anomalous” data. But as you have said elsewhere, there are adjustments on adjustments. The tinkerers have no doubt very good reasons for adjusting – and maybe have peer-reviewed articles to back this up – but they have lost sight of what real world data. The Windhoek example demonstrates this – and helps disproves the idea of a conspiracy to defraud. You see, the overall adjustments nicely cancel out a 0.8˚C warming between the 1920s and the 1980s.

          I wrote up the Spitzbergen data. There does seem to be an homogenisation algorithm at work there, adjusting to something on a par with the adjusted trend for Reykjavik.
          One oddity I had not spotted before. The “raw” data for Isfjord Radio contained estimated data during WW2. The British destroyed the weather station in 1941 (to stop it falling into German hands) and it was not replaced until 1946. There was no proxy temperature data available in hundreds of miles.

          We need to compare and contrast the raw and adjusted data more fully. Approaching it from different angles, and learning from each other is the best way forward to understand the real situation. In the meantime, it will be fun to try to find the biggest trend adjustment.

          • Euan Mearns says:

            Kevin, I have post on Southern Africa and then will move on to S America. And then I run out of S hemisphere. I will then likely move on to Arctic.

          • A C Osborn says:

            I posted this over at Kevin’s Forum.

            From the work that I did last year on USA temperatures involving the “E” Estimated values, as I have said before, it appeared to me that it was not a simple case of lowering the past to increase the warming.
            It also included reducing the Step changes that were apparent from things like ENSO and AMO changes. These step changes do not fit in with the Gradual, continuous rise due to CO2 increases.
            I am not sure if either of you have seen Steve Goddard’s graph of Adjustments v CO2 , the adjustments are an almost perfect fit, whereas the Temperatures are not.

  5. Pingback: Windhoek Temperature adjustments | ManicBeancounter

  6. Euan Mearns says:

    Actually, is provides a good counter example to your theory of a homogenisation algorithm adjusting to a preconceived view.

    Kevin, the early view, developed in part by Paul Homewood, was that homogenisation was adding warming to records. I set off to do epic job on Australia and showed this was not the case. Some records are warmed, some are cooled and the net effect is zero. I have post on southern Africa that should fly Friday that shows same.

    Interesting post on Windhoek. What you should see tomorrow is that homogenisation straightens out all the natural variance – as your post shows. I think its still possible that such massive adjustments are in there to make the homogenisation algorithm balance. You make a point that homogenisation in this case has removed a warming trend – don’t forget that it also added 4˚C 😉

  7. Hi Euan,

    In case you’re interested, in February 2014, we actually did a very detailed review & assessment of the homogenization algorithms used by NOAA on the GHCN and also the one used by NASA GISS. We’ve written two papers on them (one each) as part of our series on studying the urbanization bias problem, which we uploaded to our Open Peer Review Journal forum.

    They’re both quite long, but if you (or your readers) have the chance, they might provide some insight… FWIW, Paul Homewood mentioned he found them useful in his analysis which you refer to above.

    1) For NASA GISS’s “urbanization bias” adjustments, see here:

    2) For NOAA NCDC’s GHCN homogenization (& a discussion of the extent of UHI in the GHCN), see here:

    • A C Osborn says:

      I have just read the first of the papers and find it of great interest, especially the part about identifying Rural & Urban stations and also the overall reduction of Rural stations in the dataset. With so many Rural stations to choose from the original data sets it makes you wonder why they chose to use more Urban ones instead.
      The second item would automatically add UHI bias if the bias is not handled properly.
      The second item of interest is the difference between Trend and Step changes and it makes you wonder how their (NCDC) algorthims handle actual Climatic Step Changes like ENSO or Volcanic eruption events as against non climatic step changes like lnstrument or location changes.

      • A C, I’m glad you found them of interest.

        With regards to why GISS chose to try & correct, then use, urban stations, rather than just using rural stations, as we discussed in the GHCN paper, the rural station records tend to be relatively short & incomplete, with large data gaps (often associated with non-climatic jumps).

        This is not too surprising when you think about it: Before the era of automated weather stations, if you wanted a continuous station record you needed it staffed & the more isolated the station was the harder it was to find staff to maintain the records continuously!

        Unfortunately, this means that, aside from exceptions like the USHCN, most of our best-kept, long-term weather records are from urban areas.

        It’s truly an insidious problem.

  8. A C Osborn says:

    I have now read the second paper and like the first it is very thorough and very interesting.
    Urban blending is an obvious problem, especially for stations like Valentia, where it’s climate is controlled by the Sea and prevailing winds. To compare it to London or France is quite laughable.
    So, not only do they not adequately correct for UHI, the corrections that they do do smear the UHI all over the place.

    • Exactly. On our blog, we made the analogy that it’s like putting strawberries and bananas in a blender and “homogenising”. After homogenizing, your smoothie may well be a “homogenous blend”. But, its properties are a mixture of both strawberries & bananas.

      NOAA throwing urban & rural stations into “the blender” and thinking this will “remove” the urbanization bias, is like expecting your smoothie to be 100% pure strawberries!

Comments are closed.