RealClimate logo

‘The Discovery of Global Warming’ update

Filed under: — group @ 14 July 2006

If you haven’t already seen the American Institute of Physics website by Spencer Weart on the ‘The Discovery of Global Warming’, we heartily recommend it. It provides both a summary of science, and more importantly, a history of how an obscure speculation from over one hundred years ago has become the scientific consensus of today. It has recently been updated with many more references from 1873 to the present, and so is even more worth reading. Spencer is very keen on getting feedback on the project, so don’t hesitate to let him know what you think.

78 Responses to “‘The Discovery of Global Warming’ update”

  1. 51
    Lloyd Flack says:

    #43 Martin,

    The best indicator that the mechanism of temperature change in the past 150 years is different from what occurred earlier would be sustained rates of temperature increase that were unprecedented. I think high temperatures are less of an indicator than high rates of temperature increase. And yes we do need better reconstructions of past temperatures.

    But by themselves temperature changes will only be indicative. To draw conclusions about causes one needs to have an idea of the processes involved and what they can and can’t do. What the paleoclimate record potentially can do is to provide a reality check on models.

    In climate science one cannot get the conclusive proof that one can from say a mathematical theorem or from a designed experiment. It has been pointed out on this blog that we have only one world and cannot do the required experiments on it. What we can do is come up with physically possible explanations and test them. If a hypothesis is consistent with the available data and no one has been able to come up with an alternative hypothesis that fits then we provisionally accept the hypothesis. It is always possible that someone will come up with an alternative hypothesis in the future but we don’t hold our breath waiting.

    Some people claim that climate is too complicated and detailed modeling is a hopeless task. This would leave us without a way of telling whether temperature rises are being driven by greenhouse gases or not. It is a position that fails if anyone comes up with a model that adequately explains what is happening.

    The general circulation models appear to to be a reasonable but not perfect fit to what has been happening in the past century. No one has been able to create a model that fits the data that is not sensitive to greenhouse gases. In particular there does not appear to have been a solar forcing that would explain the temperature increases of the past quarter century.

    I think the argument that the fingerprint of changes is what would be expected of greenhouse gas rather than solar forcing depends on the most robust aspects of the models. Gavin, Rasmus could you do a post comparing what you would expect from greenhouse gas forcing with what you would expect from solar forcing? And then could you compare both with what is happening?

  2. 52

    Re: response to #43,

    Gavin, the 1% globally averaged surface albedo errors would correspond to 1.7W/m^2, the average albedo errors of the AR4 models Roesch’s IPCC diagnostic project reported were 0.016-0.019, which corresponds to 2.6-3.1 W/m^2. Now in Hansen (2005), I believe you used GISS-EH. Its global albedo came in over 1 SD below the model average, almost spot-on the higher of the satellite observations. In that paper you were reporting that the earth was now absorbing 0.85+/-0.15W/m^2 more than it was radiating into space, and that of the 1.8W/m^2 of forcing increase since 1880, 0.85 W/m2 remains, i.e., remains in the pipeline corresponding to 0.6 degrees C of committed future increase. The actual albedo errors to that contribute to the global average that Roesch reports, are quite large and concentrated in regions of snow cover, times of snow melt and in the tropical deserts. The coupling between the errors and the annualized net energy balance will be nonlinear, but the global average of the errors suggest their magnitude is too large to dismiss as unlikely to make a big difference.

    The whole AR4 suite of models is biased against solar forcing by on average over 1.5 times the forcing increase we are trying to simulate, and more than an order of magnitude larger than the accuracy we need balance the energy budget for attribution and climate commitment studies. With models being so central to attribution studies and even “independent observational” studies, we should not be “hesitant”, to point out that it is premature to make important policy decisions based on model projections.

    Any models managing to balance their energy budgets to under 1W/m^2, with these surface albedo errors, must have other compensating errors in forcings, sensitivities or internal feedbacks. We know of many more problems with models than just this albedo bias. Of particular importance, this diagnostic study, showed that all the models have this systematic positive bias to a greater or lesser degree, so the practice of combining them into ensembles to somehow reduce their errors and increase their credibility is called into question.

    There is no doubt that models are much improved since the TAR, and that these diagnostic studies will contribute to further progress. Unfortunately, premature “confidence” in the models has damaged the reputation of this science which has already made important qualitative contributions and has promise for useful quantitative results in the future.

    Despite the problems we will always have judging how to spin up the oceans and climate to uncertain past conditions, I look forward to studies with corrected models, accurate enough to carry climate commitment properly forward into attribution studies of the recent warming, ideally all the way from the Maunder minimum, to the historically high levels of solar forcing of the last 60 years.

    James Hansen, Reto Ruedy, Larissa Nazarenko, Makiko Sato, Josh Willis, Anthony DelGenio, Dorothy Koch, Andrew Lacis, Ken Lo, Surabi Menon, Tica Novakov, Judith Perlwitz, Gary Russell, Gavin A. Schmidt, Nicholas Tausnev (2005). “Earth’s Energy Imbalance: Confirmation and Implications”. Science. DOI:10.1126/science.1110252

    Roesch (2006) “Evaluation of surface albedo and snow cover in AR4 coupled climate models”

    [Response: You are making a fundamental error. Just because two quantities are in the same unit, it doesn’t mean they can be usefully compared. There is a huge difference in the net imbalance and an error in two compensating fluxes. Think of this as trying to model a lake – the ‘control’ lake level is a function of the water coming in and going out which we know must be balanced. Now we add some extra water (or slow the rate of outflow), and we calculate the rate of lake level rise. The lake level rise can be estimated very clearly, despite not being absolutely sure of what the actual level in the lake is. Thus the rise estimated is largely independent of the absolute level. This is exactly what happens in the AR4 models. They are tuned to a particular ‘level’ (i.e. global albedo) and are in balance, and the imbalances that occur when GHGs are added are largely independent of that level. I will point out also that the AR4 models are mainly tuned to ERBE data, not CERES – which was not available during the model development process.

    Your statements regarding solar forcing biases make no sense. An increase in solar irradiance (say from solar min to solar max, which is about 1.3 W/m2) need to be converted to a TOA forcing to see it’s effect and the albedo comes into that calculation F_solar = 1.3 * (1-a)/4, so the difference in F_solar as a function of a 1% error in ‘a’ is tiny – around 0.003 W/m2. How this can be described as a major bias is beyond me. -gavin]

  3. 53
    fFreddy says:

    Re #47, Response from Gavin
    [Response: Steve Milloy again. Complete cr*p again. Neither CO2 nor water vapour nor CH4 nor O3 nor N2O spectra are saturated. – gavin]

    As a matter of interest, do you happen to know how saturated they are – 20%, 80%, whatever ? Particularly for CO2, of course.

    [Response: I’m not sure a simple percentage is a useful diagnostic. The net forcings increase at different rates (logarithmically for CO2 at least up to 1000ppm, like sqrt(conc) for methane and N2O), and the saturating bands are included in those estimates. -gavin]

  4. 54

    Re #48 and “increasing CO2 can warm the surface. Doing so decreases the “optical depth” of the atmosphere”

    Don’t you mean increasing CO2 INCREASES the optical depth of the atmosphere?

  5. 55

    #51 Lloyd,

    There is a lag in climate response to an increase in forcing. Solar forcing has been at a plateau of high activity for over 60 years. This current level of activity has been characterized as the highest in 8000 years, with only an 8% probability (based on the paleo record) of continuing another 50 years. (Solanki, 2004, 2005, correspondence w/Muscheler 2005)

    Climate commitment studies by Wigley (2005) and Meehl (2005) showed that due to the thermal inertia of the oceans that most of the temperature response to an increase in forcing took place in the first 100 years, and that sea levels may continue to rise for 1000 years before equilibrium is achieved.

    Hansen (2005 above) reported that the lag was a function of the climate sensitivity, “The lag could be as short as a decade, if climate sensitivity is as small as 0.25-C per W/m2 of forcing, but it is a century or longer if climate sensitivity is 1-C perW/m2 or larger.”

    Obviously we need models that can balance the energy budget and represent the heat flux into the oceans to accurately reproduce the climate commitment from all the increase in solar forcing from up to 100 years or more previous to the recent warming. One also must consider that equilibration to this solar forcing would have been delayed by the aerosol cooling. How does one properly attribute this? Is the recent rapid warming a rebound from the ending of negative aerosol forcing, or a delayed equilibration to the still persisting plateau of solar forcing?

    While the GHGs probably also made a significant contribution, keep in mind that past attribution studies were probably made with models with the systematic surface positive albedo biases I discussed above. This reduction in the solar contribution to the energy budget would have to be made up with increased forcings or sensitivities elsewhere in the models (compensating errors), in order to match the 20th century climate. This perturbs the whole climate model, including probably attribution of the warming earlier in the century. Since the albedo bias reduced the sensitivity of the models to solar forcing, it would also reduce the equilibration lag time, so that less climate commitment from increases in solar forcing would be carried through to later years. I don’t think we can know whether most of the recent warming is attributable to GHGs or to solar (including prior commitment) until better studies are done with corrected models. I look forward to that.

    Meehl G. A., et al. Sciencexpress, 10.1126/science.1106663 (2005).

    Raimund Muscheler, Fortunat Joos, Simon A. Müller and Ian Snowball (2005). “Climate: How unusual is today’s solar activity?”. Nature 436: E3-E4.

    S.K. Solanki, I.G. Usoskin, B. Kromer, M. Schussler, J. Beer (2004). “Unusual activity of the Sun during recent decades compared to the previous 11,000 years.”. Nature 431: 1084-1087. DOI

    S. K. Solanki, I. G. Usoskin, B. Kromer, M. Schüssler and J. Beer(2005). “Climate: How unusual is today’s solar activity? (Reply)”. Nature 436: E4-E5. DOI

    Wigley T. M. L., et al. Sciencexpress, 110.1126/science.1103934 (2005).

  6. 56
    fFreddy says:

    Re #53, Response from Gavin
    Couldn’t you say that, for the relevant wavelengths, there is a certain amount of energy that leaves the surface of the earth, and a certain amount that escapes to space ? Why wouldn’t the ratio of the two give a reasonable figure for saturation ?
    (Of course this would only be saturation by wavelength. If I understand it correctly, there is a fair amount of overlap between different GHGs in terms of the wavelengths they grab. So “saturation by individual GHG” would take a bit of thought to define meaningfully.)

  7. 57

    R:: 52 response from Gavin,

    I am not talking about a change in solar irradiance from the solar cycle when discussing the model bias. The change is between the climate and the model. A 0.01 positive surface albedo error, would be applied to the solar flux at the surface of 169W/m^2, which gives 1.69W/m^2. This is the average flux that is in the climate, but not in the model. To avoid any TOA confusion, remember that at the bottom of the atmosphere, per Hansen again “The observed annual mean rate of ocean heat gain between 1993 and mid-2003 was 0.86 +/- 0.12 W/m^2 per year for the 93.4% of the ocean that was analyzed.” These figures are comparable. I admit land based albedo errors are not as tightly coupled to the oceans as an ocean albedo error, but we can’t be comfortable dismissing the large land errors, in the non-linear climate system. It is possible for instance, that the increased local warming pins or shifts climate features to certain regions for longer periods of time, in a way that increases the coupling.

    [Response: You stated there was a bias against solar forcing, which I took to mean the change of climate due to solar forcing. I apologise if that wasn’t what you meant. But my first point still stands – the imbalance can be seen to high accuracy despite uncertainties in the individual fluxes – and we do not know them to better than a few W/m2 in any case. We do know that they must have been close to balance, and now they are not. I can measure the lake level rising with an accuracy of millimeters even I don’t know the depth of the lake to better than a few meters. – gavin]

  8. 58
    Grant says:

    Re: #54

    Doh! I stand (sit, lie down, and run in place) corrected.

  9. 59

    Re: Response to #52, the lake part

    Gavin, I have tried to be fair to your lake analogy and see if it can apply somehow. I knew it would be a copout to say it was an oversimplification, because all analogies are. But I couldn’t get my mind around balancing models around the albedos, because I could see how it would be done, since albedos are also a function of climate due to snow cover and vegetation changes. So, I translated your lake analogy to balancing heat flows and the energy budget instead. If all we care about is the rate at which the lake is rising, that is equivilent to just caring about the net energy flux, and you are right, it the absolute level of the water and the total amounts of energy don’t matter. If the systems behavior is l linear and the inputs and outputs are not changing, we can even make predictions. However, the absolute levels of the inflows and outflows and the mechanisms become important once we start trying to attribute relative importance to each, or need to estimate what the rise will be in response to changes in inputs, etc. For instance, if a 1 acre lake is rising 1 ft per year, because of an inflow of 1 acre ft per year, and the outflow is dammed. That is a different situation than a lake rising 1 ft per year, because of a 2 acre ft per year inflow balanced by a 1 acre ft per year outflow. In the case of the dam, a doubling of the inflow, doubles the rate of rise, until the dam is topped. In the case without the dam, a doubling of the inflow, triples the rate of rise, assuming the rate of outflow doesn’t change, however, we don’t know the response function of the outflow.

    In the case of the climate, attribution studies were attempted back when models were not fitted to much more than temperature. With the advent of coupled models, with studies attempting to simulate climate commitment and future sea level rise from heat flows, models actually do have to balance the energy budget. But I don’t think we can assume there will be a simple counter balancing of flux errors about this balancing of the energy budget, because models today are expected fit far more than just the energy balance. Trying to fit temperatures, cloud cover, ocean currents, snow cover, etc. all at the same time, brings in a lot of other constraints, which will force some elements of the models closer to the absolute physical values. Both the heat flow into the oceans, and the albedo error we have been discussing, are net heat fluxes at the surface, averaged both globally and annually. So they are directly comparable in terms of the energy budget, although there is quite a contrast in temporal and geographical distribution of the details of the two quantities. With this size of errors, one has to question how the model balances the energy budget, and yet still matches so many other diagnostic statistics. Has it recovered the solar heat it lost through surface albedo error, through increased sensitivity to GHGs? Given the range of model sensitivities the particulars of the distribution of compensating errors must vary.

    [Response: The models are usually tuned in two specific ways – firstly to have a reasonable global albedo (target range around 30% to 32% given the uncertainties in the data), and secondly to have a net long term surface heat budget close to zero. Both of these things are usually accomplished by adjusting minor cloud parameters. If the albedo was clearly shown to be exactly 30.5%, we would have no problem adjusting the models to match – these kinds of adjustments are made all the time as a function of changes in the cloud routines, boundardy layer or surface code. Generally, the sensitivity to increasing CO2 is not affected by these changes, however, there are obviously compensating adjustments to the other fluxes – generally within their uncertainties except in some key problematic locations. You need to appreicate that these changes in the fluxes are really small compared to the absolute flux, on the order of a few percent at most – thus they don’t generally change the big picture climate in the models and I have seen no evidence that there is a systematic relationship to the sensitivity. It certainly doesn’t show up in the AR4 models – try plotting sensitivity or feedabck strengths against the albedo offset (Soden and Held, J. Climate, 2006 for instance). – gavin]

  10. 60
    Chris Rijk says:

    #51 Lloyd,
    Maybe you’d find this useful:

  11. 61
    Hank Roberts says:

    Gavin, if the artwork linked in #60 is sound, you all might consider recommending the artist’s site– that one ends in 1990 and is based on a DOE model. I wish more scientific illustrators did such clear charts.

  12. 62
  13. 63

    Re: response to 59,

    Gavin, correcting a positive surface albedo bias with a negative cloud albedo bias of unrelated geographical distribution just doubles the error in the model. You don’t know whether you are correcting the solar sensitivity, or just unrealistically altering cloud feedbacks to who knows what forcings. If we did know this a apriori, then we wouldn’t need the models.

  14. 64

    Re: my 63 comment

    I reported earlier that the GISS-EH model probably used in Hansen (2005), had a globally averaged albedo over 1 SD below the mean and close to the more positive of the satellite observations. Perhaps attempting to compensate for one error with another, explains another reason Roesch (2006) found this model notable. It was one of only 4 models to show a positive Snow Cover Area trend at at time when the climate was showing a negative Snow Cover Area trend.

    [Response:It was GISS-ER in Hansen et al 2005 – gavin]

  15. 65
    Leon Pahler says:

    I did not see the reference to the ‘Wegman’ article and I have reviewed all of your comments. Someone please provide a web address for this article. Thanks.

  16. 66

    Gavin, Martin,

    In addition to this discussion, one may not forget that there is an observed inverse correlation between TOA solar irradiance and (low) cloud cover, see the caption in Fig. 1 of Kristjansson e.a.. During a solar cycle, this causes a +/- 1-2% change in global low cloud cover. SST temperature differences can go to up to 0.3 C within a sun cycle. Thus anyway, the -relative- small changes in TOA solar irradiance seems to be strengthened by changes in cloud cover.

    What that means for long-term changes in solar irradiance is quite uncertain, but strengthens the possibility that these are underestimated in current climate models.

  17. 67
    Hank Roberts says:

    65, Leon: There’s a link on this page that’ll download a copy of Wegman:

  18. 68

    Ferdinand (Re: 66)

    Thanx for the reference. The Roesch paper also emphasized the importance of SSTs: “The time slice simulation using ECHAM5 with prescribed SST and sea-ice suggests that accurate SSTs and ice coverage are fundamental for a correct prediction of SCA trends.” I presume, that whereas the subtropics were important in your cite, that it is the higher latitude SSTs are what is more influential snow cover. Although, it is undoubtedly all coupled at some time scale. Modelers have a difficult task of validating their models against so many qualitatively different observations, all the while their models are expected to also display the internal variability of the natural climate.

  19. 69
    Lloyd Flack says:

    The suggestion by Wegman et al. that long memory time series should be tried in modeling paleoclimate seems reasonable to me. This is going to need someone who is familiar with both climate processes and with new methods in time series. This is going to be needed in order to come up with models that are both physically plausible and interpretable and mathematically tractable. If you can’t find someone with the necessary experience working on climate data then I would suggest someone with experience working on climate related data. Perhaps hydrological data.

    What we need to know is not so much whether current temperatures are unprecedented in the last thousand years but whether the current rate of temperature change is unprecedented in that period. A re analysis with better methods might answer that question. At the least it would give us a better idea of the limits of our knowledge.

    It’s time to admit that mistakes were made in the paleoclimate reconstructions and that improvements can and should be made.

    #39 Eric we can’t prove the mechanism of global warming from observational data alone. We can’t do controlled experiments and have only one planet to work with. We can just say that the observations are consistent with certain mechanisms.

    #45 Hank deterministic models always give the same results given the same inputs. Stochastic models usually don’t give quite the same result when repeated with the same inputs. The first are taught in applied mathematics departments. The second are taught in statistics departments.

    Which is the more useful type of model depends on the questions asked, the processes involved and the resources available. Not many people have a lot of skill with both. It is not easy to derive the random behavior of a process from its deterministic aspects if one cannot perform experiments. I suspect that the creation of deterministic general circulation models does not necessarily help their creators as much as might be expected in predicting the random aspects of weather and climate. One necessarily ignores or tries to average out random elements when creating these models. This could leave one without guidance when one is confronted by randomness such as that in paleoclimate estimation.

    These are my observations and impressions and I haven’t seen things put this way elsewhere.

    [Response: Improvements can always be made, but the choices made in the MBH report were reasonable judgment calls. It’s obviously important to know what impacts they have, but for the centered PCA change, the impact is very small (see or the Wahl and Amman (2006) paper (their scenario 5d) (the fourth panel in this figure.). – gavin]

  20. 70
    Hank Roberts says:

    Here’s an image I haven’t seen before that seems rather good.
    I’d like climatologists’ opinions on — accurate? convincing?

  21. 71
    Henri says:

    I first learned about global warming when I was in elementary school. There were many other students who were interested in science and so we then had the chance to really hear about this. I then also read about global warming in a nice book about our planet that I read in 1993 titled, Save The Earth. When I learned about global warming I was learning about how it was a difficulty in our planet’s atmosphere where a big round entrance had formed above the Antarctica region, being the south pole area. If I remember right, apparently this gap in outer space and in the atmosphere somehow created the continual warming of our planet.

  22. 72
    lars says:

    Ok, if we have CO2 fluctuations over the years and 5 in the last 400,000 years What caused them to be periodic? Periodic volcanic activity? Or something else or combinations of something else?

    [Response:Orbital forcing – see Hays, Imbrie and Shackleton (1974) and all subsequent references…. – gavin]

  23. 73
    Hank Roberts says:

    Lars, look among the links in the footnotes here:

  24. 74

    Re #71 and ” When I learned about global warming I was learning about how it was a difficulty in our planet’s atmosphere where a big round entrance had formed above the Antarctica region, being the south pole area. If I remember right, apparently this gap in outer space and in the atmosphere somehow created the continual warming of our planet.”

    No, you’ve conflated two issues there — ozone depletion and global warming. The widespread use of chlorofluorocarbons in the 20th century depleted stratospheric ozone and created the ominous “hole in the ozone” over Antarctica. The problem, while still serious, is under control because the 1987 Montreal Protocol (revised a couple of times since then) has pretty much banned the use of CFCs.

    Global warming is due primarily to the release of carbon dioxide (CO2) when fossil fuels are burned. The amount of CO2 in the air has risen by over a third since the industrial revolution started (from about 280 parts per million by volume to about 380 ppmv). As a result, the world has warmed by about 1 degree Kelvin. Fossil fuel use is increasing so quickly that warming of 1.5-6.0 degrees is likely in the next century. This is enough to have a major, probably negative, effect on human agriculture.

  25. 75

    Re #70,

    Hank, if that picture is true, we are heading for temperatures higher than during the Cretaceous, a period with CO2 levels 4-10x higher than pre-industrial and a different (warmer) continents location… Seems a little overblown…

  26. 76
    Hank Roberts says:

    This should change something — now we have rapid changes far faster than the human fossil fuel release, in geologic history. Any temperature records that would go with this or similar?

    Found here, a press release:

    “… the Bishop tuff, a volcanic layer tens to hundreds of meters thick that is exposed at the earth’s surface as the Volcanic Tablelands in eastern California. This massive deposit represents what is left of the estimated 750 cubic kilometers of magma ejected during the formation of the Long Valley supervolcano caldera some 760,000 years ago.
    …. Alfred Anderson of the University of Chicago and his colleagues studied the size of the bubbles under a microscope to estimate how long it took the magma to leak out. Based on these and other experiments and field observations from the 1990s, geologists now think that the Bishop tuff–and probably most other supererupted debris–was expelled in a single event lasting a mere 10 to 100 hours. “

  27. 77

    Re #76 “This should change something — now we have rapid changes far faster than the human fossil fuel release, in geologic history. Any temperature records that would go with this or similar?”

    Hank, Have you read “The two mile time machine” by Richard Alley? That tells of rapid warming much faster than now with no greenhouse gas forcing! At least not methane or carbon dioxide. It happened only 10,000 years ago.

  28. 78

    […] Warming?? update??? …. Notice that a recent warming has occurred, starting at 2000m in 1985, … NEWS Health Global warming &aposmay cut deaths&aposA deadly UK heatwave may be imminent, but […]