The Ocean ate my heat and global proxies

And interesting discussion has been triggered about the use of proxies to measure global temperature. It developed but the key point seems to be this comment from catweazle666
in response to William Connolley.

October 12, 2013 at 7:27 pm

“A skeptic actually goes out and reads the scientific literature where this is all described in excruciating and boring detail.”

[Citation required]

Actually, i suspect you are making stuff up, as there is not to my knowledge currently no coherent, credible hypothesis – never mind theory – as to precisely how and why serious quantities of thermal energy suddenly decided to change from warming the atmosphere to warming the deep oceans – entirely ignoring the principle of convection, i might add, and further, without producing a characteristic increase in sea level rise commencing at the time it suddenly decided to alter its behaviour.

Not to mention that the relative thermal capacities of air and water tend to indicate that any such increase in temperature would be lost in the noise, of course.

It then moved onto the use of proxies and the hockey stick.

William Connolley says:

> MBH97 comes to mind.

MBH98 is trivially publically accessible (, so that can’t be the one you mean. I don’t know what you mean by ’97. And like it or not, the results of the paper do indeed follow directly from the described methodology and datasets.


> One is left with guessing how they actually measured that.

I’m sorry, I can’t parse that. What did you mean?

Given such a contentious range of issues, and as it was already becoming hard to follow, I decided to elevate the discussion to an article.


This entry was posted in climate. Bookmark the permalink.

25 Responses to The Ocean ate my heat and global proxies

  1. There seems to be a fair amount of confusion in the “lay public” about how accurate you can expect temperature measurements to be. People will say, correctly, “its pretty hard to read a mercury thermometer to within 0.2 oC” but then go on to say, incorrectly, “so you can’t trust the temperature record to within better than 0.1 oC”. This is to forget the effects of averaging. This is trivial to demonstrate mathematically: if the true temperature is T, then you measure T’, and we assume that T’ = T + e, where e is the measurement error, and the measurement error is normally distributed, then the expected error for the average of n repeat measurements is e/sqrt(n). If you want to care about the details (as it applies to temperature records; what I’ve said is just basic stats, of course) you’ll need to read one of the foundational papers on the temperature records; CRU, GISS, whatever; BEST comes to hand: They’ll tell you things like “The land temperature rise from the 1950s decade to the 2000s decade is 0.90 ± 0.05°C (95% confidence)”. If you want to criticse numbers, criticise those, not made-up ones like “99.99% confident they could measure temperatures globally to 0.0001C”.

    Note also that the statements (a) “you can’t measure the (global, decadally averaged) temperature accurately” and (b) “we believe that the true temperature trend is IPCC minus 0.2/whatever” are mutually inconsistent. If you believe (a), then you can’t believe (b); instead, you have to adopt “we don’t know how the temperature has changed” (note also that your uncertainty in measurement is +/-, so you can’t just reduce the trend for that uncertainty).

    • I went to B&Q today. The thermometers were were pretty good for B&Q being within a range of 1C. Does this mean if I average more and more the temperature will get closer and closer to the “actual temperature”?

      Of course not. Not only is there variation, but there is also bias.

      A family member who used to take temperature readings for the Met Office explained one of these biases. “When it was cold some people didn’t bother going out to read the thermometer but just used the last reading”. But when it was a nice day they would prefer to be outside than in.

      Working in a factory, I have seen whole sheets of paper with supposed measurements that were all EXACTLY THE SAME. In other words, only the first was an actual measurement … I was even told “there was nothing wrong so we just put the same reading”.

      To be frank, you cannot possibly imagine how people behave until you have gone into a factory at 3am when none of the management are around.

      • > I went to B&Q today. The thermometers were were pretty good for B&Q being within a range of 1C. Does this mean if I average more and more the temperature will get closer and closer to the “actual temperature”?

        The Met Office doesn’t get its thermometers from B+Q. It buys very carefully constructed and calibrated ones, and it gets them recalibrated on a regular schedule.

        > there is also bias.

        Indeed there is. But it depends on the bias; if you’re averaging across a number of different thermometers, the bias in any one thermometer merely appears as an additional source of noise in the overall result. Think about it.

        Problems arise if the thermometers, overall, are biased. In the case of mercury-in-glass, they aren’t, in and of themselves. You do have to take account of changing practices in shelter design, or canvas buckets on ships; this is all part of the tedious details; you need to read that stuff if you want to discuss it.

        > A family member who…

        You can’t do science by anecdote. The people that did the temperature readings were dedicated; I find it unpleasant to see you demeaning all their hard work.

        To be frank, you cannot possibly imagine how much effort people have put into creating a viable temperature record unless you bother read their papers.

        To be frank, you cannot possibly imagine how much effort people put into measuring temperature at Met stations unless you’ve talked to them. I have. You haven’t.

  2. how can the sun warm up the bottom of the ocean and bi-pas the top 700m ? miracle

    • In most large bodies of water where the temperature ever drops below freezing (0C) the bottom layer is near 4C because colder or hotter water rises. This bottom layer forms a distinct zone which doesn’t easily mix with upper layers.

      One area this does happen is in the Arctic where rapid evaporation of the warm North Atlantic drift current (falsely called gulf stream) increases salinity and this and the cold creates water at about 4C which is dense enough to fall to the bottom of the ocean.

      Another type of area is one where the crust is thin and heat from the core raises the temperature. So water is heated above the 4C and tends to rise.

      Another way, is that ocean currents (caused by trade winds diverted by Coriolis force) can push deeper water toward shallower areas and cause mixing.

      These are either nothing to do with “global warming” or easily measured (if there was the will to measure anything sensible)

      But otherwise, there is no way that the lower layers can change unless there is a change at the surface.

      • 4C (39F) is the heaviest – simple gravity controls

        one litre of water, or 10cm3 at 4C is only one kilogram heavy – that’s how measurement of a kilogram was invented – nothing new – they cannot say that they overlooked something

        gravity makes water to circulate and take oxygen deep down, to replenish……

        • Hint: try looking up the variation of density with temperature in freshwater and seawater.

          • plus: salinity is not evenly distributed – otherwise freshwater or salty, behavior is the same; 4C is the magic number – all other water is less heavy than 4C.

          • Billy, at a kilometer dept, fresh and salt water have great density – if you are trying to say that: you know that saltwater is heavier – bravo; you know what every child knows; but tell me I’m open for learning

            • JBL says:

              If you’d actually tried to understand what WMC is saying, you would have put the words “density of water by salinity and temperature” into your favorite search engine. This brings up a variety of resources, like (the first (not necessarily best) hit for me). Then, you can play around and notice the following features:
              1) for pure water, the largest density is indeed at 4C;
              2) salt water is denser than pure water; and
              3) salt water is *not* densest at 4C. (I leave it to you to find more detailed information than this.)
              This took me about 5 minutes to do. Perhaps in the future, you could try a similar technique — it helps prevent one from looking foolish.

              • JBL, if you have to consult Google for those such a simple things; we are talking on different frequency – be happy, cheers – I’m going through giving up smoking at the moment
                hint, relay on yourself, not on Google, even Google can deceive
                originally the subject I think was, the temperature below 700m and ad density there…

                • JBL says:

                  Use whatever resource you like — perhaps your local library has some reference works with tables of density by salinity and temperature. But at some point, if you want to pontificate on questions relating to the density of seawater, you should probably have spent more than 0 minutes studying the question. (Of course, it’s not on me if you want to make a fool of yourself by refusing to understand basic issues.)

  3. Jeff Norman says:

    Just for clarification, I sed:

    ” This skeptic has also experienced the frustration of eventually accessing a paper to find that the claims made in the paper are not supported by the secrets hidden in the paper. MBH98 comes to mind. One is left with guessing how they actually measured that.”

    Please note, I changed the “7” I originally typed to an “8” in this quoted section, since W.C. has already graciously pointed out my error.

    Of course “how they measured that” was northern hemisphere land surface temperature anomalies going back to 1400 AD, that gave then a 95% confidence that their plotted data was within ~0.2 C of the true figures.

    Once again W.C. stepped into the breach and helpfully pointed out they had used proxies.

    At the time (1999) this was new and exciting to me but I was greatly perplexed that many of the features of the climate history I was familiar with could not have shown up in this new and improved record. It seemed to me that extraordinary claims required extraordinary proof.

    To me, this proof would have involved a careful examination of each proxy record to confirm a good correlation with locally recorded temperature anomalies. This correlation would be used to derive a confidence interval for each individualy proxy that would be used in calculating the overall confidence of the reconstruction. This information however was not recorded in the paper or any subsequent corrigendum. Apparently there were too many other concerns with this paper, like using proxies that were not temperature proxies as temperature proxies.

    I have heard people say that MBH98 was a good start for trying to reconstruct past temperatures but I also believe it poisoned the well making people like me expect much more from future attempts. Fool me once…

    • You really need to try actually reading the paper. Its available as You want the top R of page 780. Or p 781. And so on.

    • Jeff Norman says:

      Once again W.C. has stepped up to the plate with a helpful a pleasant explanation.

      Having followed his suggestion, I read again that:

      “Regions not directly represented in the trainee network may nonetheless be indirectly represented through teleconnections with regions that are.”

      This is an interesting hypothesis. I don’t see any of the listed cites that explicitely or implicitely justify this kind of correlation. But surely this would reduce the confidence interval of the data sets so impacted. Which ones were they again?

      • > indirectly represented through teleconnections

        I’m afraid you’ll have to think. If you do that, it will be obvious. If you can’t work it out, you don’t understand the paper, or the maths behind it. In this you are by no means alone.

  4. Jeff Norman says:

    World ocean heat content and thermosteric sea level change (0–2000 m), 1955–2010

    Levitus et al 2012

    In part the abstract sez:

    ” Our estimates are based on historical data not previously available, additional modern data, and bathythermograph data corrected for instrumental biases. We have also used Argo data corrected by the Argo DAC if available and used uncorrected Argo data if no corrections were available at the time we downloaded the Argo data. The heat content of the World Ocean for the 0–2000 m layer increased by 24.0 ± 1.9 × 1022 J (±2S.E.) corresponding to a rate of 0.39 W m−2 (per unit area of the World Ocean) and a volume mean warming of 0.09°C.”

    Remarkable claims require remarkable timing, just before the AR5 cut off. Widely cited in WG1.

    Not being associated with an academic or research organization, I would have to pay Wiley a sum of money that would be revealed to me after I had registered before I could view the entire document, but even then would I have learned how they had measured a volume mean warming of 0.09°C?

    • Jeff Norman says:

      And if I’m reading that correctly, that is 0.09 ± 0.0075 °C.

    • JasonB says:

      Or you could just Google for “Levitus et al 2012”.

      The very second link it returns is to the paper itself:

    • Jeff Norman says:

      The webpage at might be temporarily down or it may have moved permanently to a new web address.

      • JasonB says:

        Still works for me; just clicked on the link above using a completely different computer and downloaded it successfully again.

        Did you try the Google search? It still comes up as the second result, and if you click on the little down arrow on the end of the URL you can just click on “Cached” and obtain Google’s cached copy of the document.

        BTW, I’ve almost always been able to find a free copy of a paper this way. If a direct search doesn’t work, look for the lead author’s homepage because they will often list their publications with links to download them (or at least a just-before-acceptance draft). If all else fails, email the corresponding author — most journals allow authors to give out copies on request (or at least, turn a blind eye).

        • Eli Rabett says:

          A simpler way is to find the paper on Google Scholar. There is a link to all the versions at the bottom of the item listing. Click on that. Look for links that have an adobe acrobat file on the right side.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s