Sea Heat

Just a quick note, that when it comes to the heat that’s been building up in the oceans, Zeke Hausfather got right to the point:


He’s one of the authors of a new paper (Cheng et al. 2018) which brings together all that we’ve learned lately about ocean heat content, to show that the best estimate is going up faster than we thought before. In fact it’s following what the computer models predicted it would do, with surprising fidelity. You can find plenty of press reports, including in the New York Times.

The data in his graph (available here) are from Lijing Cheng, the paper’s lead author. They’re monthly data, and look like this (Hausfather’s graph uses a different baseline, doesn’t start until 1955, and shows a 12-month moving average rather than monthly values):

The rapid and seemingly inexorable rise in the heat of the oceans is a sign of trouble. If it continues, it will bring us another foot of sea level rise by 2100 from thermal expansion alone, quite apart from the melting of land ice. Since ocean heat is the energy source for storms like hurricanes, they can be even more destructive. Ocean species will have to migrate and adapt to the new conditions; they’re already on the move (ask fishermen), and their future is uncertain. That means our future is uncertain.

What strikes me most about the data is the sharp turn about 1990. It’s easily confirmed statistically, and if we model the data as two straight lines, choosing the optimal “turning point” by change point analysis, we get this:

With it, we can estimate the average rates during the two episodes, pre- and post-1990. Two things strike me about this. First, the increase in the speed of ocean heat content rise is quite large, going from 2.8 ZJ/yr to 9.5 ZJ/yr, three and a half times as fast. Second, it’s probably not a coincidence that the year 1990 is the same at which sea level accelerates. Since thermal expansion is one of the root causes of sea level rise, this is to be expected.

In closing, I’ll mention that my wife’s idea of “sea heat” is Jason Momoa.


This blog is made possible by readers like you; join others by donating at My Wee Dragon.


13 responses to “Sea Heat

  1. Timothy (likes zebras)

    1990? About when the Communist East collapses, their heavy industry shuts down and stops adding sulphate aerosols to the atmosphere?

  2. The late 80’s was also the time that the sulfate cap and trade to kill the acid rain problem really kicked in North America . As atmospheric sulfates are cut out, clouds go down and more solar insolation is absorbed.

  3. Trevor Woodward

    Tamino, please forgive me for posting this off-thread question; I’ve searched for your email address but cannot find it:
    With revelations of Climate-Denier-in-Chief and his Russian connections coming under scrutiny by the FBI, I was wondering whether it was possible to take a climate-event attribution studies approach to determining the statistical likelihood of the Russian interference on the 2016 US elections swinging it for the Trump campaign?
    A WaPo article suggests it is possible but it would require more information from social media to understand how minority groups were targeted in votr-suppression campaigns. https://www.washingtonpost.com/opinions/2019/01/11/theres-way-know-if-russia-threw-election-trump/?noredirect=on&utm_term=.d7185a458a63
    I’m no mathematician but isn’t there a simpler way by looking at what would had been the outcome had the DNC hack not been released in such a timely manner, compared to what really happened, using historical aggregated polling data such as from FiveThirtyEight.
    Given the rise in emissions under this government and their toxic, anti-science stance, It would be a wonderful smack-down by the climate science community if it could establish the percentage likelihood of Russian interference swinging it.
    I just can’t imagine how one would determine what the polling data may have been had there been no interference.
    I hope I have given you a good challenge. Keep up the good work. My email address is supplied in the next field.
    Regards,
    Trevor

    • Trevor, This is off topic, but what I have seen showed that the main reasons the election swung were:
      1) Hillary going into the last month of the election has only the slimmest of leads–especially in electoral votes.
      2) James Comey’s inept handling of the “new” emails.

      No doubt Russian trolls amplified these effects, but all they really did was repeat memes that the Republican lie machine had been screaming for years. That does not diminish the crimes of the colluder in chief. Were we talking about any other politician, people would be asking what he knew and when he knew it–and not knowing that one’s campaign chairman is a Russian agent ought to be nearly as damning as knowing it.

      • FWIW, and briefly as it is indeed off topic, it wasn’t my experience that the memes were purely GOP old hat; I was quite startled to see unknown persons popping up to claim that Hillary was likely to provoke war with Russia. How effective that particular tactic was, I don’t know, but it was definitely out there, and rather persistently.

      • Trevor Woodward

        Thank you Doc & snark, I apologise for going off thread. I guess there is no statistical way to identify likelihood.

      • Hi Doc,
        I considered the suggestion that Hillary would provoke a war with Russia to be an extension of the long-standing knock that she was a hawk/warmonger. This was more prevalent on the left than the right, actually, but the ‘Murica-First wing of the Rethugs were quite vocal on this score, in part because they knew it hurt her with her base.
        Trevor,
        It is not that you couldn’t assess such a hypothesis, but that any conclusions would be model dependent, and any model would be a gross oversimplification of the state of the electorate from Sep.-Nov. 2016.

  4. If you model the data by two lines that intersect at some point, how many degrees of freedom is that? And can you do better with that many degrees of freedom via some other model?
    I find change point analysis compelling, and the suggestions above that a sudden drastic reduction in sulfate emissions also seems reasonable. But I have nagging doubts…

    • John, each line has 2 parameters, so a piecewise line-segment model has 4 parameters. You might think that the fact that the lines intersect at some changepoint would decrease the number of parameters, but the changepoint is itself a variable.
      A quadratic has 3 parameters, and so would be a more parsimonious model unless you have good reason to expect slope to change where it does.

  5. Pingback: Deep data state of the climate: How the world warmed in 2018 | Red, Green, and Blue

  6. In IPCC WG1 AR5, page 29, they estimate the radiative imbalance of the Earth between 1993 and 2010 to 163 ZJ, i.e. 9.6 ZJ/year, the number being similar to Cheng.