Winners and losers in the GHCN estimation derby

Guest essay by John Goetz

As noted in an earlier post here, approximately 66% of the GHCN record is estimated after processing by the GHCN adjustment models. In the current GHCN data set, there are 226 unique country codes represented. Following are the top and bottom stewards of temperature data, by country, based on the frequency that the GHCN adjustment models replace their raw data with an estimated value.

The top 10 countries with the most pristine temperature record:

Country % Estimated
DJIBOUTI 0.0%
TROMELIN ISLAND (FRANCE) 0.0%
QATAR 0.0%
COCOS ISLANDS (AUSTRALIA) 0.0%
WESTERN SAMOA / SAMOA 0.0%
BELARUS 0.0%
KAZAKHSTAN 0.0%
ARGENTINE BASE IN ANTARCTICA 0.0%
PEOPLE’S DEMOCRATIC REPUBLIC OF YEMEN 0.4%
MOLDOVA 1.0%

The 10 bottom-feeding countries requiring the most adjustment are:

Country % Estimated
BARBADOS 97.8%
CHRISTMAS ISLAND (AUSTRALIA) 95.7%
NIUE (NEW ZEALAND) 95.6%
CYPRUS 94.2%
SOUTH GEORGIA (U.K.) 93.3%
MACAU (PORTUGAL) 93.2%
UNITED STATES OF AMERICA 92.5%
BERMUDA (U.K.) 92.3%
BAHAMAS 92.2%
CAPE VERDE 91.9%

There are several things of note in these rankings, not the least of which is the fact that the United States ranks 7th in having its temperature record adjusted. One is that three of the top 10 countries come from the former Soviet Union (Belarus, Kazakhstan, and Moldova). Taken as a whole, the countries comprising the former Soviet Union have data estimated just 22% of the time.

Also of note are the two middle eastern countries of Qatar and Yemen making the top 10 (The winner, Djibouti, is just a stone’s throw away from Yemen across the Red Sea). With all of the turmoil in the region, it is surprising the countries were able to keep their records so polished. Overall, the countries comprising the traditional definition of the middle east have their temperature records estimated a modest 43% of the time.

Other countries of note, the percentage of their record that is estimated, and their global ranking (from least estimated to most estimated) out of 226 are:

  • North Korea, 28.3% (43rd)
  • China, 36.9% (53rd)
  • Canada, 41.2% (61st)
  • Brazil, 56.7% (106th)
  • Australia, 59.5% (119th)
  • India, 74.5% (167th)
  • Mexico, 75.7% (171st)

Of final note, the countries that make up the current European Union have 45% of their temperature record adjusted.

Note: This post was updated to set the featured image.

Advertisements

62 thoughts on “Winners and losers in the GHCN estimation derby

    • According to the US CRN, the actual rate of mean temperature change in the US is a cooling rate for the duration of the CRN record. I doubt that is the case for the adjusted GHCN record.

  1. Barbados?
    Barbados is #1??
    I would have thought they would use Barbados and extrapolate temperatures 1200 Km in all directions.
    Maybe the trade winds coming in from the Atlantic are not warm enough. But, just guessing, I would have thought Barbados is plenty warm. Barbados, really.
    (For the record, open disclosure: when there, I buy my local supplies at a little shopping center known as “The SuperCenter” in Ostins, in the parish of ChristChurch, right on Maxwell Coast Road.)
    For whatever reason, I guess they just do not like the Barbados Weather Service, or it’s thermometers.
    Here is a puzzle for all at WUWT:
    I measure the ocean temp off the beach at 86 Deg. F. (fairly constant year around), and measure 84 Deg F. in the pool (also constant). The swimming pools run consistently 1 deg. C cooler than the ocean. WUWT?
    And NO, they are not cooled. They do not do that on that island.

    • Since Barbados is the furthest east Car Island it would be a proxy for a good part of the Atlantic ocean so adjusting it would affect 100’s of thousands of square miles,
      re pool, must be the ground temp, assuming an in ground pool.

      • OK, not a bad answer, I did not think of that myself. But after 6000 years of equlibration time, why would the ground still be cooler?

      • The pool in the ground is exposed to cool temperature at the bottom. The island’s rock won’t have warmed to match the ocean surface, as the underlying rock is being cooled by deeper cold water.

      • In the Spring at our beach when the water is still cool, I notice a thin layer of warm water brought in by an on-shore wind. Could it be the limited area of the pool, vs area of the ocean?

    • > Here is a puzzle for all at WUWT:
      I measure the ocean temp off the beach at 86 Deg. F. (fairly constant year around), and measure 84 Deg F. in the pool (also constant). The swimming pools run consistently 1 deg. C cooler than the ocean. WUWT?

      Clarity.

      • Also, the swimming pol is likely filled with fresh groundwater to replace splash-out and evaporation.
        Groundwater in Florida is no where warmer than about 76 F

    • One factor to consider (of several): evaporation of fresh water in swimming pools is more efficient at cooling than salt water.

    • A couple of things I can think of as possibilities but depends on how far off shore your are measuring. Have you walked on the sand at the beach without shoes? Kind of hot isn’t it? Depending on how far away from the beach you are measuring the temperature, you could have hot sand influencing shore water temperature and as suggested, if the pool is in ground, you could have heat loss to ground. Stick a thermometer in the ground near the pool and also one in the sand at the beach. See what the difference is

    • @TonyL, I like some have already wondered if you have a nice stratification going on at the beach. Google epilimnion and Hypolimnion. So a reading at the top and bottom down at the beach might shed some light. The pool being small and probably has people moving around in it mixing it up continually so it doesn’t stratify as much as the beach.

    • @ Tony, I have a really great set of thermometers ( accurate to .25 C ), a good model on my computer and if you pay the air fair and breakfast, ( maybe a few drinks and a dinner as well maybe an escort to show me the “sights”) I’l be more than willing to come and do a study! ( sounds familiar?)

    • Either oceanic currents, or more likely ocean overturning is less effective than your swimming pool pump such that ocean surface temperature is warmer since it is not as well mixed with cooler deeper water..

    • The swimming pools run consistently 1 deg. C cooler than the ocean. WUWT?

      The thermal energy transfer (temperature) within and/or out of a small volume of water is more easily measured than is that for a large volume of water.

  2. So 2/3 of temperatures are estimated yet global temperatures are reported to the hundredths of a degree. After an explanation by Mr. Mosher a few articles back that there are multiple reasons for all these adjustments, maybe he can enlighten us on how such an accuracy can be achieved?

    • My understanding is some devices report to 1/10 of a degree, if you had 3 devices and 1 reported 60, a second 60.2 and the third 60.2 the average would be 60.13 (you can state it even more “accurately” if you want as 60.3333333333333333333333333333333333333333333333333333333. Of course if the device has been recently calibrated it is likely only accurate to 0.2 of a degree or so..

      • Good point, but there is a typo, it should be 60.133333333333333333333333333333333.
        And, in any case, no reputable mathematician would pretend that the mean of any series of numbers that are truncated to the first decimal place can be accurate to more than one decimal place. The average (mean) of the three numbers 60, 60.2, 60.2 is 60.1. But even that may be wrong. For instance the three numbers 60.04, 60.24 and 60.24 would be correct to the first decimal place 60.0. 60.2 and 60.2 but their average (mean) would be 60.17 and, correct to the first decimal place 60.2.
        The fiddle factor for those averaging, homogenising, estimating or generally playing around with temperature measurements, whether current or historic, is enormous. Starting with accurate data, which itself is a great assumption, a true believer can prove anything.

      • One of the first things I learned in engineering school is a result cannot be more accurate than the least accurate data point or more specifically, if there are temp readings to the degree and others to the tenth of a degree then whatever mathematical manipulation is done to them, accuracy is only to the degree.
        Actually, in this case; manipulated data is only inaccurate to the degree.
        Claiming 2014 or 15 or 16 is the hottest year on record by 1 or a few / 100ths. of a degree is bogus.

      • Menic old soul,
        I wish you were right: –
        “Reporting any result to more significant figures than the least number of sig figs in the numbers being manipulated, would get you a big fat zero in on any science exam at any University anywhere.”
        In some places, perhaps not a million miles removed from West Anglia, you may get tenure.
        All very sad.
        A descent into opinion, witch-doctoring and malice.
        Auto.

      • Expat:

        One of the first things I learned in engineering school is a result cannot be more accurate than the least accurate data point or more specifically, if there are temp readings to the degree and others to the tenth of a degree then whatever mathematical manipulation is done to them, accuracy is only to the degree.

        Um… what? If you really were taught that in engineering school, you went to a terrible school. What you describe is true when dealing with a small number of data points, making it an okay rule of thumb for some types of problems, but it is a terrible lesson for any engineer.
        Any engineer, or anyone who is expected to work with data in a serious manner, should know the precision and accuracy of your results can often be increased by increasing the amount of data you have even if you do not increase the precision/accuracy of that data. If you have five data points measured to a single degree, your accuracy is only to the degree. If you have five thousand data points measured to a single degree, your accuracy should be greater.
        Menicholas:

        Reporting any result to more significant figures than the least number of sig figs in the numbers being manipulated, would get you a big fat zero in on any science exam at any University anywhere.

        Auto:

        In some places, perhaps not a million miles removed from West Anglia, you may get tenure.
        All very sad.
        A descent into opinion, witch-doctoring and malice.

        There’s nothing sad or wrong about it. There are many problems which get solved where individual data points have less accuracy than is needed but there’s sufficient amounts of data to reach the required levels. There are tens of thousands of practical examples in the world around you which show your claims are baseless.

      • “If you have five thousand data points measured to a single degree, your accuracy should be greater”
        NO !! Not if those data point are not measuring exactly the same thing.. Which they aren’t.
        The “one on square root n” criteria ONLY applies for repeated measurements of the same thing.
        A measurement to the nearest degree is essentially a uniform distribution between +/- half the measuring unit.
        Averaging uniform distributions does NOT change that error. (easily checked using random uniform distributions in Excel)
        If you have 5000 measurements read to the nearest degree, the read error is still +/- 0.5 degrees….
        …… then add on all the other error possibilities.

      • “My understanding is some devices report to 1/10 of a degree, if you had 3 devices and 1 reported 60, a second 60.2 and the third 60.2 the average would be 60.13 (you can state it even more “accurately” if you want as ….”
        But your actual average is anywhere between 60.183…. and 60.083…. so writing 60.13 is basically meaningless, and certainly 60.1333333 is total nonsense.!

      • Then of course you have the issue that adjustments have proven that many older measurements are to an accuracy of +/- 2 or 3 C
        So your final average comes with a MASSIVE range, which makes it essentially USELESS for any purpose whatsoever.. let alone science.!!!!
        ie GISS and HadCrut !!

      • Shollenbergergergerger,
        Wrong, wrong, wrong, a thousand times wrong. You are talking about the so-called “Law of Large Numbers” which does not apply here. Guess what: temperatures vary every yard or two, every second or two, worldwide year-round. “L of LN” only applies to repeated measurements of the SAME QUANTITY! Read, and understand…

  3. It would be very helpful if authors would definf acronyms in the first sentence/paragraph or so. For example GHCN = The Global Historical Climatology Network (GHCN) is a database of temperature, precipitation and pressure records managed by the National Climatic Data Center, Arizona State University and the Carbon Dioxide Information Analysis Center.

    • I agree strongly. Now that Government does “everything” we have “zillions” of acronyms. Some are used to describe multiple things.
      Bet the NCDC is described as the NCDC until “they” come up with a new name or acronym. Then it will be a different string of letters. “they” need to keep the peasants confused and in awe.

    • Not truly ‘acronyms,’ but ‘initializations.’ Acronyms are pronounceable; i.e. ‘NASA,’ ‘NOAA,’ ‘OSHA’

    • Yes, when done properly, the first reference to any abbreviation contains both the long and shortened versions.
      The way I was taught, anytime one is reading something, and one comes across a word or phrase that you do not know the meaning of, you must find out what it means, or you might as well stop reading.
      Because you are then reading something with no chance of complete comprehension.

  4. “The demand for high-value environmental data and information has dramatically increased in recent years. To improve our ability to meet that demand, NOAA’s former three data centers—the National Climatic Data Center, the National Geophysical Data Center, and the National Oceanographic Data Center, which includes the National Coastal Data Development Center—have merged into the National Centers for Environmental Information (NCEI).” http://www.ncdc.noaa.gov/about

  5. The torturing of the Australian data by the Bureau of Meteorology was covered by the Australian newspaper. The government under a then very fragile leader Tony Abbott backed away from an inquiry into the matter. Malcom Turnbull, the new leader is a supporter of climate change so there will be no appetite to revisit this. The Australian paper FOI request around the decision is actually interesting reading
    https://www.dpmc.gov.au/sites/default/files/foi-log/FOI-2014-206%20-%20Documents%20for%20publication.pdf
    It looks like the Bureau of Meteorology got off by agreeing to increase the expert panel to include external experts, data users, regional experts and statistics experts to further look at torturing the data.
    I guess for Australia the good news is even with the officially improved climate change friendly version of the data, the temperature rise was 1 degree for the century, most of that since 1950 and no hockey stick. So even taking the higher rate from the 1950’s the increase would be around 1.5 degrees at the year 2100.
    The climate trends section of:
    http://www.bom.gov.au/climate/change/acorn-sat/#tabs=Exper-review

  6. The very fact that we must dicker over hundredths of a degree indicates strongly that there is no threat to humanity from the production of CO2. This leaves anyone with common sense trying to deduce why there is a panicked push to hobble the western economies just in the name of science.
    Then take a look at the campaign to intertwine the science-cloaked ideological agenda with religion and the public conscience, to inspire faith in the “electronic climate prophets” and “overwhelming opinions of authority” presented to them.
    To compel respect from those who refuse the religion, the climate anthro-doomists call for a censorship of any views or data that they don’t agree with, and legal enforcement of that ban.
    How can this go on without great public outcry?

  7. I just can’t buy Yemem being pristine;; with all the rockets’ red glare, the bombs bursting in air, the UHI effect should substantially raise temps——–nevermind.

    • And it is nearly always the same ‘hot, hot and more hot’ there.
      I keep saying, ‘global warming’ should be tracked in only one place: Hudson Bay. This is the epicenter of all major glaciation events. If it is colder than normal and is cooling down faster and faster, this is warning sign we are going into another Ice Age.

      • It occurs to me that the best indication of CO2 induced warming should be in the deserts at night. Where there is very little water in the air it is up to the minor GHGs to warm at night.

    • “…the UHI effect should substantially raise temps…”
      It may be that the UHI effect works by removing the moderating effects of moisture containing ground surfaces, plants and shrubs and trees, and replaces them with surfaces that are more like bare rock or dry sand. Some of the effect comes from actual heat emissions, as from AC units, motor vehicles and other machinery, heat escaping from buildings, etc.
      So perhaps areas that are dry deserts and bare rocky soil to begin with, will have less, perhaps much less, alteration with urbanization.

      • Plus, i have noticed they end to use a lot of white paint in such places. Perhaps this increases albedo sufficiently to counteract UHI to some degree.

      • I can tell you it makes a noticeable difference in the ability to cool the top floor of a classroom building when the roof is white membrane, instead of black and covered with brown river rock.

Comments are closed.