Skip to content

GHCN’s Arctic Adjustments

October 19, 2012
tags: ,

By Paul Homewood



                        Before Adj                                                               After Adj


We have already seen that GHCN have been busy making large adjustments to the temperature records provided by the Icelandic Met Office. By cooling the past record up to about 1965 and/or increasing more recent temperatures, they have added about a degree of warming to the trend since the 1930’s at every Icelandic station. In doing so, they have flattened out the extremely cold spell in the 1960’s and 70’s, (well known in Iceland as the “Sea Ice Years”), and also done the same with the extremely warm period between 1930 and 1960. (Full story is here.)

It is important to recognise, though, that this climatic pattern was not simply a local effect limited to Iceland. HH Lamb, for instance, reports that in 1965 “sea temperatures around the Faroe Islands fell by 1.0C and were as cold as at any time in the previous 100 years.” (This series of sea temperatures is one of the longest in the world, dating back to 1867). Dickson and Osterhus, in their paper “100 Years In The Norwegian Sea”, find exactly the same pattern around Norway.

This is all borne out in the original temperature records for a wide range of Arctic stations that GISS still maintain, from GHCN V2, for instance Nuuk in Greenland below, or Ostrov Dikson in Siberia above.





Yet, just as in Iceland, the algorithm used by GHCN has decided that these climatic shifts never really occurred and adjusted them out of history. From Greenland, going east via Iceland and Norway, through to Siberia there are 21 GHCN stations above 64oN, which have current and long term records. In addition to the six Icelandic ones already reviewed, there are the following:-







Jan Mayen









Kanin Nos

Ust Cilma

Malye Karmaku


Ostrov Dikson



Of these 15 stations, the records remain unadjusted at six. The adjustments for the other nine are shown below.


GHCN ARCTIC ADJ_htm_m6d32fb7c

GHCN ARCTIC ADJ_htm_71469042

GHCN ARCTIC ADJ_htm_7fdbb6ba


GHCN ARCTIC ADJ_htm_m3ffa182[7]

GHCN ARCTIC ADJ_htm_4425230b[4]

GHCN ARCTIC ADJ_htm_m63a26424

GHCN ARCTIC ADJ_htm_m7b3aed51

GHCN ARCTIC ADJ_htm_5581af68

GHCN ARCTIC ADJ_htm_718aba76

As you will see, every station has had the same sort of adjustments as the Icelandic ones – either the past is cooled or the present is warmed, or both. Most stations have had adjustments of a degree or more.

GHCN have never provided any justification for these adjustments, and yet their effect is significant. The adjusted temperatures in GHCN V3.2 are used by GISS as the basis for their global temperature calculations. It is well known that neither GISS or GHCN have very little temperature data inside the Arctic Circle, as the map below shows. (This is based on 250km smoothing).




And yet, by projecting temperature trends over 1200km, they manage to colour in the whole Arctic.





So it appears that adjustments at a small handful of stations can have a significant and disproportionate effect across the entire Arctic region, and in turn on GISS’s global temperature datasets.

Whatever the cause of these unjustified adjustments in the first place, it is the case that GHCN have known about this issue since February. Their failure to either remove or justify the adjustments suggests at the very least a degree of incompetence. Others may view this opinion as generous!

It has been suggested that I take the GHCN software codes apart to see what is wrong with them! If my car breaks down on the motorway with steam coming out of the bonnet, I don’t have to take the engine to bits to know it is buggered! On the contrary, it is up to GHCN to check their software, and either prove that it is working correctly or put right what is wrong with it.

It is also worth questioning the position of GISS in all of this. They too, to my personal knowledge, have been aware of these problems for many months. Their attitude has been to wash their hands and pass off the problem to GHCN. Although there may, of course, have been discussions behind the scenes, of which I have not been made aware, such an attitude raises questions about their own competence and scientific rigour.

We already know that satellite data shows less Arctic warming since 1979 than does GISS. One wonders what Jimmy Hansen’s map above would look like if we had had satellites back in 1940!

While such doubts remain about the GHCN V3.2 dataset, the GISS Surface Temperature Analysis becomes ever more suspect.



1) All unadjusted GHCN V2 temperatures are available at

2) The latest GHCN V3.2 temperatures can be accessed at

3) GHCN before and after adjustments can be seen here  (Station numbers can be obtained from the GISS database).

  1. Edward. permalink
    October 19, 2012 11:56 pm

    HadCRUT, GISS, NCDC all are bent data sets, ergo, how can GHCN be ‘more honest’ if the source/ base data sets are flaky……………sorry er………readjusted?

  2. October 20, 2012 6:47 am

    Proper protocol:
    1. Show the original data
    2. Explain why it is likely faulty
    3. Suggest and explain adjustments to correct it
    4. Detail and apply such adjustments
    5. Show the result, and justify why it is better.

    None of the above have been followed, except for a crippled version of #5.

  3. Jeffery permalink
    October 20, 2012 12:33 pm


    I’ve been reading along for a number of months and quite enjoy your site. Full of great information to which I’ve regularly pointed global warming acolytes of my acquaintance. Many thanks for your work.

    I have no doubt the political fix is in. Recently it was revealed that America’s political prophet of global warming doom, Al Gore, has achieved a personal worth of over 100 million US, (from a starting point in the 90’s of about 2 million.) so some motives are easily seen at the top.

    That said, I also believe valid studies show atmospheric CO2 is rapidly increasing. Do you have any response to the likely effects of this shift?

    • October 20, 2012 5:58 pm

      I tend to keep away from the science side of things, as I am not qualified to comment. There is of course a wide variety of scientific opinion,, despite what we are told, as to the effects of CO2.

      I prefer to concentrate on the empirical side, what is actually happening and how this fits into what has gone on before.

      Nothing I have seen so far suggests that global warming or climate change will be “catastrophic” in the foreseeable future.

  4. October 21, 2012 5:24 am

    I am amazed how widespread these unexplained GHCN adjustments of basic data seem to be! I recognise that sometimes there are changes in location of stations that require some adjustments to retain homogeneity, but any such adjustments must be justified if the results are to be at all meaningful. Surely it is illogical to adjust recorded data just because another gauge some hundreds of km away has a different variation in annual temperature. If, on the other hand, the other station was nearby (say 10km away) an adjustment of one record may be acceptable.

    I note that the GHCN adjustments shown only extend back to 1930. Many stations have much earlier data and I wonder if GHCN makes any adjustments to earlier data (eg Vardo, Norway since 1829). My analysis of raw temperature data (from the KNMI website) for far-northern stations indicates a consistent pattern of warmer periods around 1930 and 2000 and cooler periods around 1900 and 1970 (Google: brigun Vardo). It would therefore appear that GHCN would choose to use near-zero adjustments prior to about 1920! But how could they justify that?

  5. John G permalink
    October 21, 2012 7:54 pm

    what do you mean there isn’t “justification” for the homogeneity adjustments? why don’t you download the homogeneity code (isn’t the old version from USHCN still on the website somewhere?) and have it print out the statistics of every breakpoint it detects. Or better yet, print out all the data about what stations were paired with the ones here and do a thorough forensic analysis.

    all you continue to illustrate is that the homogeneity software finds large step changes in the data. isn’t that the point of it?

  6. John G permalink
    October 22, 2012 12:20 am


    “Until GHCN can justify them, they are “unjustified”.”

    Uh-huh. You’ve cited menne/williams/09 before. That’s the paper that details the homogenization algorithm. The justifications are in there – they’re statistical and mechanical. AND they’re validated against synthetic test data where the authors can specifically control where breakpoints are.

    “Just because there are step changes does not mean they are not real. If the software is adjusting for events that are real, there is a big problem with it.”

    Of course. but there’s more than enough middle-information in the code to tamp down that problem. Maybe if you were to have a list of the stations which were used to pinpoint a particular breakpoint and then line them all up and see what’s going on it might be more obvious. or you could see that the statistics is poorly tuned and gives lots of false positives.

    Point is you ain’t done none of that yet. why? it’s not hard, just a tiny bit more time consuming. You claim that the adjustments aren’t “justified.” But you haven’t bothered to show us the network of data that was processed in the first place. Is it because in context, there are obvious breakpoints and anomalies in the raw data? Skeptic in me says “yessir”

    “I also gather that the software code, which should be on the GHCN website, is missing!!”

    Then email ncdc. it’s public domain.

    • October 22, 2012 9:58 am

      I’m not sure if the algorithm in Menne 2009 is still being used. I do know that GHCN have changed their software twice in the last year, V3.1 and then V3.2. Before these updates, these Arctic adjustments were not being made.

      I have highlighted these adjustments which do not appear to reflect what was happening in reality. I still believe that it is up to GHCN to justify them, not for anybody else to disprove them. Having said that, any help in untangling the software would be gratefully accepted!

    • October 22, 2012 12:51 pm


      A further thought. I accept that the algorithm has probably done what it was designed to do, search out abrupt shifts. However, according to the GHCN techlog, the reason for adjustments is to remove “non climatic influences such as relocations, changes in instrumentation and observation practices”.

      So have they found any evidence of such changes in any of these stations?

      Indeed, they admit they don’t really understand any of the adjustments they have made, saying “The reason for the larger number of cold step changes is not completely clear, but they may be due in part to systematic changes in station locations from city centers to cooler airport locations that occurred in many parts of the world from the 1930s through the 1960s.”

      I already have assurance from the Iceland Met that there were no such relocations or other changes that would justify the adjustments made in Iceland.

  7. John G permalink
    October 22, 2012 1:17 pm

    > So have they found any evidence of such changes in any of these stations?

    You’re neglecting the point of the algorithm, which is very explained at ( and in a peer-reviewed article, menne and williams in 2005 (

    The entire point is that there isn’t metadata documenting some strong anomalies in the timeseries data. But very obviously, when you throw standard timeseries analysis techniques to the data, you come up with strong evidence that there are anomalies or breakpoints in the data. what you are asking for is something which can never and will never be provided for you – you are NEVER going to find a random interview with some random inuit in the Arctic who is going to confirm that “yeah, a polar bear ate the observer that one year so we had to train a new observer and they measured things too high”

    the homogenization process DOES use metadata where available to identify breakpoints. Why don’t you run the code and see for yourself? The old version is on the previous link on ncdc website. its not too hard to run.

    > I already have assurance from the Iceland Met that there were no such relocations or other changes that would justify the adjustments made in Iceland.

    Then why hasn’t the Iceland met office made a fuss with the ncdc for corrupting their data? Perhaps you have misinterpreted their response to your inquiries?

    • October 22, 2012 3:35 pm

      On the contrary, the metadata in Iceland is very well recorded and the original temperatures have already been adjusted where necessary by the IMO for change in location etc.
      And according to one of their senior meteorologists, the GHCN adjustments are” a blunder”.

  8. John G permalink
    October 22, 2012 11:04 pm

    The metadata in the US is also very well recorded. and yet there are myriad instances of obvious flaws in the data.

    “And according to one of their senior meteorologists, the GHCN adjustments are” a blunder”.”

    i’d love to see an attributed quote and the work that the person making the quote did to back up this assertion.

    • October 23, 2012 9:59 am

      I raised the whole matter as a legitimate concern with GHCN back in February, and have chased several times. As you suggest, it should be pretty simple for them to run the code, explain the adjustments and list the stations which were used for comparison.

      But they have not done so.

      The quote is from a private email to me, which I can forward to you if you want.

  9. John G permalink
    October 23, 2012 1:36 pm

    you could always run the code. you’ve clearly got the chops to interpret whatever falls out of those experiments i suggested. I think there is a real opportunity to do some interesting, sound, and valuable science, and the code really isn’t that hard to run or hack. you don’t need to forward any private e-mail; generally, if I send something in confidence to someone, I wouldn’t want it forwarded! But at the same time, if all you have is a private e-mail, then you really shouldn’t be traipsing around claiming that the iceland met office has a stance or opinion on GHCN.

    good example might be from the presidential debates last night. pres flubbed something with a weak rebuttal to the fact he hasn’t visited israel since he took office. now, I’m sure that david axelrod is making the rounds talking about how he effectively defended his record or whatever, but I’m sure that behind closed doors, when they focus group the message and realize that they botched things, his opinion will be different. That doesn’t necessarily mean that Axelrod’s cautious, careful interpretation should be aired to the public.

    Of course, you can do whatever you want. But you shouldn’t represent a private, personal correspondence as an official stance or policy. At best, you’re misrepresenting the organization. At worse, you’re creating a liability or poor cicrcumstances for the person corresponding with you in confidence.

    • climate sceptic permalink
      October 23, 2012 2:04 pm

      No, he can’t run the code, because despite what they say in the FAQ document, GHCN have not released it on their FTP site. Paul has already told you this.

    • October 23, 2012 3:37 pm

      You are going round in circles. It should be very simple for GHCN to run the code and explain what has happened to everybody.

      I am not going to waste any more time on this, so please do not keep repeating yourself.

      As for the e-mail, the person in question, who is the Senior Meteorologist at the Iceland Met and specialises in climatology, has already given permission to publicise previous emails, which made similar comments and were sent in his official capacity. I will leave the final comment to him.

      In 1965 there was a real and very sudden climatic change in Iceland (deterioration). It was larger in the north than in the south and affected both the agriculture and fishing – and therefore also the whole of society with soaring unemployment rates and a 50% devaluation of the local currency. In the questions above the year 1965 is mentioned twice. It is very sad if this significant climatic change is being interpreted as an observation error and adjusted out of existence.

      I have been working for more than 25 years in the field of historical climatology and have been guilty of eager overadjustments in the past as well as other data handling crimes. But as I have lived through these sudden large climatic shifts I know that they are very real. “

      I think I would rather trust him than an algorithm.

  10. John G permalink
    October 24, 2012 4:06 am

    “No, he can’t run the code, because despite what they say in the FAQ document, GHCN have not released it on their FTP site. Paul has already told you this.”

    An older version of the code is right here: We’ve already established that menne and williams 2009 is a good baseline for talking about homogenization, and the coded algorithm in that paper is right there at that link. we could quibble details later on whether the results of analyzing that code extend to the new GHCN, but we still have to look at >>something<< first.

    "You are going round in circles. It should be very simple for GHCN to run the code and explain what has happened to everybody."

    [SNIP – I am not prepared to be lectured on what I should and should not do. And I am also not prepared to have my motivations impugned]

    "I think I would rather trust him than an algorithm."

    in science you don't trust people. you trust work. even if he has more domain expertise than, say, a lay person, your quote indicates that he is throwing his authority around instead of solid science. that's as bad as Hansen stepping out of line or Gore pretending to be an expert in paleoclimate stuff

    • climate sceptic permalink
      October 24, 2012 8:19 am

      John, you really should read some of Paul’s earlier posts, or even GHCN’s own documentation. What’s the point of looking at an old version that is no longer used? In fact if you read the GHCN Tech reports you will see that they say that v3 was introduced to ‘correct software coding errors’ in earlier versions.

    • October 24, 2012 9:59 am


      I have made myself clear on this. Any further repetitions will be spammed.

      (And please check some of the IMO’s work as I have done before you make such ridiculous remarks).

  11. October 24, 2012 6:28 pm

    I am working on a more detailed post for Iceland and the 1960’s cooling. Watch out – I think you will all find it interesting!!

Comments are closed.

%d bloggers like this: