The GWPF 2019 Temperature Prediction Competition

Benny Peiser writes:

It would be great if you would encourage your readers to participate in our 2019 Global Temperature Prediction Competition.

Here the the description from The GWPF website

Date: 08/02/19
Global Warming Policy Forum

With GWPF readers having trounced the Met Office at predicting temperatures for 2018, it will very interesting to see if you can do just as well for 2019.

So we hereby announce the 2019 HadCRUT temperature prediction competition. Once again, the opportunity is there to win some magnificent prizes: more whisky, and your choice of a book from the growing range of GWPF titles.

Of course the real prize on offer is to do better than the boys in Exeter. The Met Office are again being very aggressive on the warming front. They are predicting a 0.19°C warming next year (!), plus or minus 0.12°C. So their predicted range is 0.67-0.91°C.

So will carbon dioxide sweep all before it as they think? Will temperatures creep back further, shoot back up again, or will they keep sliding away? Will El Nino kick in, or will La Nina dominate?  Your guess is probably as good as mine, but – if experience is anything to go by – probably better than the Met Office’s.

Enter here


HT/Benny Peiser


81 thoughts on “The GWPF 2019 Temperature Prediction Competition

      • cdm should add that little snippet of info back in. The whole thing is rather out of context and confusing without it.

        0.19C +/- 0.12C or 0.19C +/- 63%

        so it’s like: here’s out best guess but it could be three times smaller.

        at some stage on the way to making predictions approaching +/- 100% you need to admit you have no idea. 😉

        • Don’t forget that we are now in the climate disruption extreme weirding regime. Everything will happen now. Cats and dogs living together!

          Remarkable though that they are predicting a MINIMUM warming rate of 0.7C per decade. It seems extremely likely, almost certain, that the true answer will be well below the bottom of their range.

    • The Met Office are again being very aggressive on the warming front. They are predicting a 0.19°C warming next year (!), plus or minus 0.12°C. So their predicted range is 0.67-0.91°C.

      They are predicting an increase of .19°C over last year which was 0.596°C over the 1850 to 1900 average. (i.e. the official value of the 1850 to 1900 HadCRUT4 global temperature average). Add .569 + .19 = .759 and you get the average increase next year of .759°C.

      However their range is .19°C +/- .12°C or a range from +.07°C to +.31°C over last year.
      Therefore the range is .569 +.07 = .639°C and .569 + .013 = .879°C

      If you round off the .569°C to .6°C (not sure why they did that sloppy rounding) then the range is from .6 + .07= .67°C to .6 + .031 = .91°C

      And that is how they got the numbers .67°C to .91°C increase over the 1850 to 1900 global average.

      • I got dyslexic on the actual .596°C value and used .569°C above. That also explains the “sloppy rounding” comment. Other than that….

      • The linked MO page makes no mention of 0.19°C. It predicts 2019 as being 1.1 ­± 0.12 °C above the 1850-1900 period. By my calculation this would be about 0.79°C compared with their normal base period of 1961-1990, about 0.01°C cooler than 2016.

        • Bellman

          By my calculation this would be about 0.79°C compared with their normal base period of 1961-1990, about 0.01°C cooler than 2016.

          Yes, and it would need to be at least 0.67°C to fall within the MO’s lower estimate range.

  1. “…They are predicting a 0.19°C warming next year (!), plus or minus 0.12°C. So their predicted range is 0.67-0.91°C…”

    Eh? What am I missing?

    0.19° +/- 0.12° produces a range of 0.07° to 0.32°

    • Making predictions is a bit of a mug’s game, but with a bit of wiggle matching based on the 22 year solar magnetic cycle, which features prominently in the both hemispheres’ spectral composition I get
      thus my guess (in no way better than anyone else’s) for the 2019 HadCRUT4 global temperature anomaly is +0.59 degree C.

  2. ” They are predicting a 0.19°C warming next year (!), plus or minus 0.12°C. So their predicted range is 0.67-0.91°C.”
    Shouldn’t it be 0.31- 0.07 “C ?

  3. Predicting the HADCRUT4 anomaly is hard because not only do you have to get the temperature right, you have to figure out how much thumb they have on the scale.

  4. I don’t know what they smoke at Met Office but I would like some of it because it looks strong.

    +0.19°C is a huge increase. From 2014 to 2015 the huge El Niño added +0.18°C. So they are predicting an even bigger El Niño for 2020. They won’t get it. Such large Niños drain the subsurface heat so much that they can’t be repeated for quite a few years even when a big La Niña helps with the recharge. And we didn’t have any big Niña since 2016.

    Adding ±0.12°C on top of that big exaggeration indicates they are heavily UID. We need to abolish the laws of physics or get a large asteroid to warm the Earth +0.31°C in a calendar year. There is only one precedent in the entire HadCRUT 4 database. It happened in 1877 when temperature measurements couldn’t be trusted much, and when one the biggest El Niño in recorded history killed millions of people all over the world. It was +0.306°C.

    Such is the idiocy we face everyday now.

    • The lower end of their estimate works out at 0.98C above the 1850-1900 average, which is 0.66-0.67C on their 1961-90 anomaly base. That’s just 0.07C warmer than 2018 and cooler than 2015/16/17.

  5. It staggers me that people worry about a harmless gas that’s .03% of the atmosphere of which we contribute a fraction of, when the magnetic poles of this planet are thousands of klm from where they were a couple of hundred years ago. Maybe we need a “pole model” so these crackpots can earn a quid modelling?

  6. Wondering why the prediction is for only 2019. It seems that a random guess would have as good a chance of winning. With all those billions spent on supercomputers surely they could predict the average anomaly for 2029. Then the real betting could start.

    • It seems that a random guess would have as good a chance of winning.

      Nope. It is a normal distribution. Most of the results fall between -0.15 and +0.15°C.

      The correct range would be 0.45-0.75°C. That has a disproportionate chance of winning.
      The 0.67-0.91°C range from Met Office is too high. Only the lower values of the range have a high chance.

      If I was Adam Scaife, head of long range prediction at the Met Office, and I wanted to forecast warming, I would have forecasted a warming of 0.08°C ±0.07 for a 0.68°C forecast with a 0.61-0.75°C range. If there is warming there is an 85% chance that it will be within that range. But of course that doesn’t look scary. It is clear that they don’t pretend to be correct.

      • when predicting a conserved quantity under a normally distributed random change, this year’s average is the best estimation for next year’s average, without applying any climate knowledge whatsoever.

        The estimated uncertainty would be +/- one standard dev.

        It seems that UK Met Orifice take a wider spread since there is less obvious shame is issuing an enormous uncertainty than there is in getting it wrong outside your declared margin. This also allows them to edge towards “a warmer world” and provide a hotter average to pet jounro hacks for alarmist “expert” opinion.

        Since UAH looks like we are hitting the bottom of the post El Nino cooling curve, I’d take the end of last year as the best estimation for next year’s mean.

        • Where have you ever seen an explanation of how the Met Office derives any +/- terms?
          The Australian BOM has declined for a year now, with several reminders, to reveal its +/- for daily temperatures from customary weather stations.
          You should have a parallel competition to see who can make the best forecast of their error terms. At the moment, error terms often look like convenient numbers with little scientific validity, shown to pay lip service to hard science conventions. Geoff

  7. Mods, one of my comments went missing. I only talked about drugs. I promise there was no sex or profanities, nor racial or gender offensive comments. Drugs are OK, aren’t they?

    Jeez. This comment is also going to be sequestered by the filter police.

  8. The winner will be the one with the highest temperature prediction and actual temperature readings will be adjusted accordingly.

  9. While not climate come weather, I am reminded of what Joseph Stalin is said to have said.

    Its not the number of votes which matters, its who counts the votes”.

    We have over so many years seen blatant example of fiddles being carried out by taxpayers funded bodies, that its very hard to know what is good data, or just adjusted stuff.


    • Depends on what type and how much you want to spend. A liquid in glass (mercury or alcohol) is marked to nearest degree and the observer is supposed to estimate to the nearest tenth (this is how it was done up until the ’70’s to ’80’s. Now you can get a digital one to even the nearest thousandth of a degree, but it will cost you a *lot* of money. The digital thermometers used for observation read to the nearest 1/10. But when 2/3’s of your 200 year long term record is only precise to 1deg and estimated to 1/10deg, you can’t claim any warming (or cooling) any closer than +/- 0.1F (since Englands records were recorded in F until the 70’s or 80’s.

      • The temperatures may have been estimated to the one tenth in order to round them when they were read, but they are recorded in integer values which makes them +/- 0.5 degrees. Any temperature inside that range has an equal probability of occurring. Any temperature projection less than that range is simply noise since there is no way to predict what the correct value was.

        • You are more than likely right. But recording it to the nearest degree was before my era. We recorded it to the nearest 1/10 degree which as I said was estimated. The point is, and I believe you agree with me, is that to say that it is the warmest (or coldest) by anything closer than +/- 0.3d (or even 0.5deg) is comparing apples to bananas. The problem is that many people still believe that if you take a lot of imprecise, inaccurate measurements and average them together you will get a precise, accurate result.

          • The real question is “a precise, accurate result” of what? Someone has to explain something to me about how you can take one measurement of temperature or anything then average it with one additional reading of something else, i.e., a temperature reading a day later and get something more accurate that the measurement error of the original readings.

            I am working on an essay that discusses some of this. Hopefully it will at least make some folks think about what they are doing.

    • You don’t need one. Just average a few thousand readings from thermometers calibrated to one decimal place and you can quote the result to as many decimal places as you like. The precision is the result of the averaging process, not the calibration of the instruments.

      • Did you mean this to be sarcastic?

        You may get a very accurate mean value but it also means nothing. Calculate the standard deviation to 3 sigma’s and see what your error range really is.

  10. You mean, a “Guess how lead-meltingly hot HadCRU and the Met Office are going to claim 2019 was” competition?

    I’ll go for Hottest Year Evah. How that compares to Dr Spencer’s more reliable data is another question.

  11. The existing data is fiction so its impossible to test any theory. There is a strong 60 year oscillation in the NH, with an amplitude that has been reduced with adjustments over the years. What has happened in the SH before 1980 is fiction so we really have no idea of how any theory fits what really happened, and unlikely to have reliable measure of the change from this year. More than likely that they would pull out the jigger if a cool year.

  12. Anybody can make predictions. But what is the underlying physical theory for such a prediction? People love to do statistics of random numbers. The probability of success can be quantified just like in games of chance. That is interesting math but not interesting physics.

    • Indeed.
      On the other hand there are some physical processes which are clear in the data but there is no available supporting theory. In my post above I used simple 22 year (solar magnetic cycle) repeating pattern as seen
      to make a prediction (with the proviso it’s a mug’s game). The repeating pattern it is very unlikely to continue much longer, but it is a good base as any to start from.

  13. “Will El Nino kick in, or will La Nina dominate? ”

    El Niño already kicked in.

    After La Niña built up domi nance over 2 years – following > 6 years El Niño dominance.

    Interestingly “homicide” is “dominance:

    The base word is”home”, domus.


Leave a Reply

Your email address will not be published. Required fields are marked *