Why It’s Getting Hot

The Berkeley Earth Surface Temperature project concludes: carbon dioxide concentration and volcanic activity suffice to explain most of the changes in earth’s surface temperature from 1751 to 2011. Carbon dioxide increase explains most of the warming; volcanic outbursts explain most of the bits of sudden cooling. The fit is not improved by the addition of a term for changes in the behavior of the Sun!

For details, see:

• Robert Rohde, Richard A. Muller, Robert Jacobsen, Elizabeth Muller, Saul Perlmutter, Arthur Rosenfeld, Jonathan Wurtele, Donald Groom and Charlotte Wickham, A new estimate of the average earth surface land temperature spanning 1753 to 2011, Geoinformatics and Geostatics: an Overview 1 (2012).

The downward spikes are explained nicely by volcanic activity. For example, you can see the 1815 eruption of Tambora in Indonesia, which blanketed the atmosphere with ash. 1816 was called The Year Without a Summer: frost and snow were reported in June and July in both New England and Northern Europe! Average global temperatures dropped 0.4–0.7 °C, resulting in major food shortages across the Northern Hemisphere. Similarly, the dip in 1783-1785 seems to be to due to Grímsvötn in Iceland.

(Carbon dioxide goes up a tiny bit in volcanic eruptions, but that’s mostly irrelevant. It’s the ash and sulfur dioxide, forming sulfuric acid droplets that help block incoming sunlight, that really matter for volcanoes!)

It’s worth noting that they get their best fit if each doubling of carbon dioxide concentration causes a 3.1 ± 0.3°C increase in land temperature. This is consistent with the 2007 IPCC report’s estimate of a 3 ± 1.5°C warming for land plus oceans when carbon dioxide doubles. This quantity is called climate sensitivity, and determining it is very important.

They also get their best fit if each extra 100 gigatonnes of atmospheric sulfates (from volcanoes) cause 1.5 ± 0.5°C of cooling.

They also look at the left-over temperature variations that are not explained by this simple model: 3.1°C of warming with each doubling of carbon dioxide, and 1.5°C of cooling for each extra 100 gigatonnes of atmospheric sulfates. Here’s what they get:

The left-over temperature variations, or ‘residuals’, are shown in black, with error bars in gray. On top is the annual data, on bottom you see a 10-year moving average. The red line is an index of the Atlantic Multidecadal Oscillation, a fluctuation in the sea surface temperature in the North Atlantic Ocean with a rough ‘period’ of 70 years.

Apparently the BEST team places more weight on the Atlantic Multidecadal Oscillation than most climate scientists. Most consider the [El Niño Southern Oscillation](http://www.azimuthproject.org/azimuth/show/ENSO) to be more important in explaining global temperature variations! I haven’t seen why the BEST team prefers to focus attention on the Atlantic Multidecadal Oscillation. I’d like to see some more graphs…

26 Responses to Why It’s Getting Hot

  1. Peter Best says:

    You would expect the residuals after removing greenhouse and volcanic contributions to depend on other key modes such as ENSO and NAO as well as the AMOC.

    The fit in the early 60’s is not very good. Regime shifts happened then, see the work of Roger Jones 2012 and others.

    I suspect that recent studies showing that non-linear solar influences are important will gain more backing, but not in the way dissenters might expect.

    Regional temperature analyses will show the influence of climate modes, as we find from reanalysis studies. The key issue from a seasonal and decadal forecasting viewpoint – needed for agriculture and insurance applications – is how climate change alters natural variability and the frequency of regime shifts.

    Some good mathematical and signal processing challenges here!

    • John Baez says:

      I just added something pointing out that most climate scientists consider the ENSO (El Niño Southern Oscillation) to be more significant than the BEST team does. I haven’t read the papers—presuming they exist—where the BEST team studied ENSO and decided the AMO (Atlantic Multidecadal Oscillation) was more important.

      I suspect that recent studies showing that non-linear solar influences are important will gain more backing, but not in the way dissenters might expect.

      Regional temperature analyses will show the influence of climate modes, as we find from reanalysis studies.

      Are these two sentences supposed to be related? Are you hinting the climate modes arise from ‘non-linear solar influences’?

      The key issue from a seasonal and decadal forecasting viewpoint – needed for agriculture and insurance applications – is how climate change alters natural variability and the frequency of regime shifts.

      Some good mathematical and signal processing challenges here!

      I wish the Azimuth team (like, for example, me) could boost our competence in this area to the point of doing something really useful here.

      • Pete Best says:

        John, the two sentences were not meant to be connected.

        Gerard Meehl and Julie Arblaster and now others have shown that there is some amplification of solar variability on various timescales.

        The second sentence was based on recent unpublished work; I can send you a powerpoint that links this quest with the viability of using climate modes for index-based insurance.

        I see from other BEST papers that they have looked at ENSO, PDO but not annular modes yet (maybe because of the lack of observational data).

        Re getting into the applied area, could I suggest that you get in contact with David Stephenson’s group at the University of Exeter and Dan Osgood at Columbia/IRI. Both are heavily involved in the area I outlined and working on development projects. David is one of the world experts on climate variability and is very analytic; the Exeter group has a wide range of expertise and strong connections with the Hadley centre. Dan is the mainstay of getting alternative insurance schemes going in East Africa, achieving heaps but often with very little human resources.

        Granger causality for extreme copulas would seem a very attractive and useful area for a mathematician interested in climate issues at the present.

        I will send on a list of references if you are interested.

  2. arch1 says:

    Is there any idea as to why the short dips from 1850 to 1950 aren’t reflected in the historical temperature record? Otherwise the black/red match does seem impressive, given the meager inputs and (if I understand the model) only two free parameters

  3. Harvey Brown says:

    I understand that there was little if any warming between 1998 and 2008; Kaufmann et al (2011) explain this on the basis of a combination of the rise of sulphate pollution resulting from Asiatic coal-powered power stations, the shift to the La Niña cycle, and the eleven-year solar cycle. I cannot tell if this hiatus is consistent with the Berkeley findings. Also, it is noteworthy that volcanic emissions seem to explain the dips in temperature post 1750; but does the significant warming around 1770 likewise require an explanation?

  4. Arrow says:

    Correlation is not causation.

    Besides the fit is mediocre at best. Before 1900 errors in temperature reconstructions invalidate any agreement. After 1900 at this level of averaging the graph is almost featureless and the few features it does have are not reproduced, only the general shape agrees.

    You could just take an exponential function and get a similarly good fit, but that doesn’t mean that an exponential function is a good model of global temperature and that from now on the temperature will keep increasing indefinitely.

    • John Baez says:

      Please tell the folks at the BEST project. They have some statisticians there who’d be interested to hear this news. But maybe you should read their paper first, to see what it claims. They are very careful to avoid the language of causation, but they do claim, quite interestingly, that adding corrections due to changes in solar forcing is unable to improve the fit to the data.

      Personally, from my readings on this stuff, I’m confident that carbon dioxide and volcanic activity are causing the temperature changes. We can at least say that if it’s not solar forcing (an explanation most work seems to disfavor), nor carbon dioxide increases, nor volcanic activity, a better theory is needed. And for this we need to think about the physical mechanisms involved. If for example we simply claim that the Earth’s temperature is increasing exponentially, for no particular reason, that’s not good enough.

      I don’t expect to convince you, though.

    • davidtweed says:

      When you say “you could”, it would be interesting if you actually did this and reported on what statistical measures you obtained on the fitted result.

  5. Robert says:

    For comparison, another recent peer reviewed paper (http://www.earth-syst-dynam-discuss.net/3/561/2012/esdd-3-561-2012.html) finds no statistically significant correlation between anthropogenic forcing and temperature.

    From the abstract: “Therefore, greenhouse gas forcing, aerosols, solar irradiance and global temperature are not polynomially cointegrated. This implies that recent global warming is not statistically significantly related to anthropogenic forcing. On the other hand, we find that greenhouse gas forcing might have had a temporary effect on global temperature.”

    I’m not familiar with the particular statistical tests used, but they seem to be saying they can separate out transient effects, and that the non-transient component isn’t significant. The BEST paper, on the other hand, doesn’t seem to separate out the transient component.

    Transient effects are certainly to be expected – if, hypothetically, the CO2 level were to double overnight, the global temperature wouldn’t instantly adjust, and might well overshoot the final stable temperature – so anything that can estimate the transient component is good.

    • Frederik De Roo says:

      Their conclusions are even better. An excerpt (emphasis mine):

      The implication of our results is that the permanent effect is not statistically significant. Nevertheless, there seems to be a temporary anthropogenic effect. If the effect is temporary rather than permanent, a doubling, say, of carbon emissions would have no long-run effect on Earth’s temperature, but it would increase it temporarily for some decades. Indeed, the increase in temperature during 1975–1995 and its subsequent stability are in our view related in this way to the acceleration in
      carbon emissions during the second half of the 20th century
      (Fig. 2). The policy implications of this result are major since
      an effect which is temporary is less serious than one that is
      permanent.
      The fact that since the mid 19th century Earth’s temperature is unrelated to anthropogenic forcings does not contravene the laws of thermodynamics, greenhouse theory, or any other physical theory. Given the complexity of Earth’s climate,and our incomplete understanding of it, it is difficult to attribute to carbon emissions and other anthropogenic phenomena the main cause for global warming in the 20th century. This is not an argument about physics, but an argument about data interpretation.

      • Frederik De Roo says:

        Regardless of whether their statistics is right, if they claim that a doubling or carbon emissions (and hence, a doubling of atmospheric carbon dioxide concentration) has no long-term influence on the Earth’s temperature and that this is in accordance with greenhouse theory, that’s incorrect. I wonder how they look upon the atmosphere of Venus.

        If their statistics is done well and they want to conclude from their statistics that doubling carbon dioxide has no long-term influence, they should have the honesty/courage to claim that their experiment/data analysis does indeed contradict greenhouse theory.

        • Dan says:

          Or to put it in a slightly less diplomatic way, the last sentence you quoted (i.e., “This is not an argument about physics, but an argument about data interpretation.”) seems to be patently false in that one cannot argue for their interpretation of the data without also arguing against established physics.

          I noticed that the two lead authors on the cited paper are in schools of economics. Now, I have no beef with economics or economists, but I wonder if statistical techniques which I presume were developed for studying economic time series are generically applicable to physical time series, like global temperature data? In the latter case, there are well-known physical mechanisms at work which it seems to me should be considered in the analysis (for example, surface temperature is bounded and thus cannot really be a sort of random walk, which it seems they are implying by looking for a unit root). I wonder if any experts (i.e., statisticians, climate scientists, or economists) care to weigh in on this paper or the validity of the techniques used in this domain of inquiry?

        • John Baez says:

          Dam wrote:

          Or to put it in a slightly less diplomatic way, the last sentence you quoted (i.e., “This is not an argument about physics, but an argument about data interpretation.”) seems to be patently false in that one cannot argue for their interpretation of the data without also arguing against established physics.

          Right: trying to understand global warming without using any knowledge of physics amounts to tying both hands behind our backs. This is true even at the simplest level. The rising temperature of the oceans and atmosphere mean that their energy is increasing enormously. In fact the power required for this temperature increase is approximately 1,000,000 gigawatts (that is, 1015 watts). Conservation of energy says this energy must be coming somewhere! If we simply wave our hands and blame ‘natural variations in climate’ we are being as silly as someone who believes in every perpetual motion machine that comes along.

          We need to use all the knowledge we have to get a good understanding of global warming.

        • Dan says:

          Okay, I’ve been trying to give this paper an honest, open-minded read since my last post and I have to say the logic is eluding me. They have time series of things like green house gases and global surface temperature that aren’t stationary (because they have trends in them maybe?) and, rather than attempting to remove these trends by modeling them, they just apply first or second differences to the series. That’s like taking derivatives, which in turn is like multiplying the Fourier transformed functions by factors of frequency, which will suppress the low frequencies (i.e., all of the long term effects) and amplify the high frequencies (i.e., all of the transient/noise-like stuff). Then, once they have noisy’d up all of the time series, they “use the stochastic energy balance model to motivate” (section 2.4) their fancy polynomial cointegration tests. Why wait until you’ve filtered out everything but the transients to use physics? And how can you say that any results you get after that shows there isn’t any long-term (i.e., low frequency) signal? My mind is boggled. I’d still be interested in the opinions of experts, but I felt morally obliged to point out that the paper might not be worth the time it takes to click on the link….. That’s the conclusion I’m coming to, anyway. Please correct me if it is a misguided one.

        • The underlying math seems to be an invention of economists, hence smells fishy to me. (E.g. what is stochastic drift ?)

          We have a smooth process (CO2 concentration) and a stochastic process (global temperature). Thus, the authors seem to conclude, these processes cannot be related.

          The peer review comments smell like economist inbreeding.

        • Arrow says:

          A single glance at past temperature reconstructions should be enough to tell anyone that much higher temperature swings took place in the past, many without homo sapiens even being there to witness them. So yes, natural variations can easily dwarf human contributions.

        • John Baez says:

          Yes, but ‘natural variations’ is not an explanation of what’s going on now. If you say current-day global warming is due to ‘natural variations’, you need to present a theory of what’s causing these. Natural variations always have a physical cause.

          This cause can be hard to figure out for the climate long ago. But given the detailed data we have about the current-day climate system, it would be absurd to suggest that 1015 watts of power are going into the oceans and atmosphere for some unknowable reason. We should—and can—account for that power.

          Scientists are also making progress on understanding climates long ago, but since there wasn’t a huge network of weather stations, satellites, etc., and it takes work to even figure out basic things like the shape of the continents, this will take longer.

        • Robert says:

          A doubling of carbon emissions doesn’t imply a long term doubling of atmospheric carbon. Depending on the long term behaviour of carbon sinks, it could conceivably leave atmospheric carbon unchanged in the long term. or quadruple it – unchanged if there are sufficiently large sinks which work on a long enough time scale to have been overlooked, quadrupled if increasing temperatures diminish the capacity of existing sinks causing the release of stored carbon.

          If their statistics are correct, this would be one possible explanation: the increase in atmospheric carbon is a transient, as is the consequent global warming. The extra 10^15 Watts is real, but isn’t here to stay.

          However, the first question should really be whether their statistical tests are reliable. If they are, the climatologists will have to explain how this can be – scientists don’t get to ignore inconvenient maths. If they’re unreliable, we can tell the economists they don’t understand maths – which a lot of people probably already suspect.

          Some economists do deal with broadly similar problems in their field, so we can’t blithely assume they’ve got the maths wrong. It does need checking, by an expert in statistics. (If the price of petrol goes up, how does that affect the price of potatoes? There are transient effects, from the increased transport cost, but in the long run there are all kinds of knock on effects that may end up partially cancelling this out,)

          What I’d like to see then, is a thorough rebuttal on grounds of statistical theory.

        • Frederik De Roo says:

          What I’d like to see then, is a thorough rebuttal on grounds of statistical theory.

          Yes, I’d like to see that too.

          If their statistics are correct, this would be one possible explanation: the increase in atmospheric carbon is a transient, as is the consequent global warming.

          Thanks for the clarification. Well then, since apparently you, I and the authors all seem to agree that through greenhouse physics there is a link between atmospheric carbon concentration and global warming, I think they should have run their statistical tests directly between the amount of atmospheric carbon and the human carbon emissions, to find out if there’s a long-term effect here, instead of blending it with temperatures.

          A doubling of carbon emissions doesn’t imply a long term doubling of atmospheric carbon.

          Of course not! (a) I suppose you mean “net” emissions, and (b) in the short term the relation between total atmospheric carbon and total net emissions is additive (not proportional) so without feedback mechanisms (like the sink term is proportional to the atmospheric carbon, in a very simple model) it would be additive in the long term too. The emissions are linearly related to the rate of change of atmospheric carbon.

          Depending on the long term behaviour of carbon sinks, it could conceivably leave atmospheric carbon unchanged in the long term. or quadruple it

          Coming back to the statistics. This is physics/biology/chemistry, but not data analysis. And conceivably, yes, anything can happen, but without a good explanation (better than: we can’t know so why not) the simpler solution is to be preferred, unless the statistics is very strong indeed.

        • Graham Jones says:

          The cointegration method they use comes from economics, and was apparently very popular in the 80s and 90s. This paper severely criticizes the method:

          • Imad Moosa, The Failure of financial econometrics: assessing the cointegration “Revolution”, Applied Finance.

          Abstract. One aspect of the failure of financial econometrics is the use of cointegration analysis for financial decision making and policy analysis. This paper demonstrates that the results obtained by using different cointegration tests vary considerably and that they are not robust with respect to model specification. It is also demonstrated that, contrary to what is claimed, cointegration analysis does not allow distinction between spurious relations and genuine ones. Some of the pillars of cointegration analysis are not supported by the results presented in this study. Specifically it is shown that cointegration does not necessarily imply, or is implied by, a valid error correction representation and that causality is not necessarily present in at least one direction. More importantly, however, cointegration analysis does not lead to sound financial decisions, and a better job can be done by using simple correlation analysis.

        • Adrian Burd says:

          I’ve not read the paper in question, but there have been long discussions about the validity and relevance of these techniques for studying climatic time series. A nice place to start is Bart’s blog

          http://ourchangingclimate.wordpress.com/?s=unit+root

          My feeling is that the above commenters here at Azimuth are correct: one cannot divorce the statistical interpretation from the underlying physics, chemistry, biology, geophysics etc..

          If one starts from a position of maximal ignorance and one just uses these techniques to analyze the time series, then I think that they are correct and give useful information on what can be said about the underlying processes (e.g. are they like Brownian motion) and whether one has enough data (just data alone, not data and science) to be reasonably sure about this identification. In other words, one can analyze a time series without knowing anything else about it and say that the underlying process causing this is like Brownian motion, or Brownian motion plus something else.

          However, we’re not in a state of maximal ignorance; we have a pretty good understanding of the basic principles of what drives climate, and this knowledge has to be taken into consideration when interpreting the data.

          Now, one might justifiably argue that this is not strictly science (i.e. not strict hypothesis testing) because we are mixing our hypotheses with the interpretation of the data. But that epistemological discussion is for another day when I don’t have to rush away to teach about the fluid mechanics of swimming fish and drag reduction.

          Just my ha-penny’s worth.

          Adrian

        • Thanks for the link to Bart Verheggen’s blog.

          So this idea is nothing new. And I guess someone has already done a good writeup of the absurdity of these fancy economic-Nobel statistics. In our case I bet the stuff is easiest refuted with simple numerical experiments.

          It is quite an assumption that a time series be autoregressive (of finite, fixed order). That needs some proof (model) and you can forget about statistical tests. As Tamino put it: The whole “unit root” idea is nothing but a “throw some complicated-looking math at the wall and see what sticks”

You can use HTML in your comments. You can also use LaTeX, like this: $latex E = m c^2 $. The word 'latex' comes right after the first dollar sign, with a space after it.

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 2,827 other followers