Is there a basis for global warming alarm?
Richard S. Lindzen
Alfred P. Sloan Professor of Atmospheric Science
Massachusetts Institute of Technology
Yale Center for the Study of Globalization
October 21, 2005
These slides are available as a pdffile from firstname.lastname@example.org
The reality of global warming is frequently attested to by reference to a scientific consensus:
Tony Blair: "The overwhelming view of experts is that climate change, to a greater or lesser extent, is man-made, and, without action, will get worse."
Elizabeth Kolbertin New Yorker: "All that the theory of global warming says is that if you increase the concentration of greenhouse gases in the atmosphere, you will also increase the earth's average temperature. It's indisputable that we have increased greenhouse-gas concentrations in the air as a result of human activity, and it's also indisputable that over the lastfew decades average global temperatures have gone up."
These references fail to note that there are many sources of climate change, and that profound climate change occurred many times both before and after man appeared on earth.
Given the ubiquity of climate change, it is implausible that allchange is for the worse.
Moreover, the coincidence of increasing CO2and the small warming over the past century hardly establishes causality.
Nevertheless, for the most part I do not personally disagree with the Consensus (though the absence of any quantitiveconsiderations should be disturbing).Indeed, I know of no serious split, and suspect that the claim that there is opposition to this consensus amounts to no more than setting up a straw man to scoff at.
However, I believe that people are being led astray by the suggestion this agreement constitutes support for alarm.
What is truly agreed (albeit with some controversy)
1. The global mean surface temperature is always changing. Overthe past 60 years, it has both decreased and increased. For the past century, it has probably increased by about 0.6 ±0.15degrees Centigrade (C). That is to say, we have had some global mean warming.
2. CO2is a greenhouse gas and its increase should contribute to warming. It is, in fact, increasing, and a doubling would increase the greenhouse effect (mainly due to water vapor and clouds) by about 2%.
3. There is good evidence that man has been responsible for the recent increase in CO2, though climate itself (as well as other natural phenomena) canalso cause changes in CO2.
Let us refer to the above as the basic agreement. Consensus generally refers to these three relatively trivial points.
Is there any reason to consider this basic agreement as being alarming?
Relatedly, is there any objective basis for considering the approximate 0.6C increase in global mean surface temperature to be large or small.
The answer to both questions depends on whether 0.6C is larger or smaller than what we might have expected.
A useful approach to this question is to determine how the current level of man made climate forcing compares with what we would have were CO2to be doubled (a common reference level for GCM calculations).
In terms of climate forcing, greenhouse gases added to the atmosphere through mans activities since the late 19thCentury have already produced three-quartersof the radiative forcing that we expect from a doubling of CO2. The main reasons for this are
1) CO2is not the only anthropogenic greenhouse gas -others like methane also contribute; and
2) the impact of CO2is nonlinear in the sense that each added unit contributes less than its predecessor. For example, if doublingCO2from its value in the late 19th Century (about 290 parts per million by volume or ppmv) to double this (i.e., 580 ppmv) causes a 2% increase in radiative forcing, then to obtain another 2% increase in radiative forcing we must increase CO2by an additional 580 ppmvrather than by another 290 ppmv. At present, the concentration of CO2is about 380 ppmv.
It should be stressed that we are interested in climate forcing,and not simply levels of CO2.
8 200400600800100012001400CO2 (ppmv)024681012Percentage increase in radiative forcingLinearLogarithmicThe CO2forcing follows the logarithmic rather than the linear curve. Note that the logarithmic curve rises ever more slowly as the CO2increases.The easiest way to understand this is to consider adding thin layers of paint to a pane of glass. The first layer cuts out much of the light, the next layer cuts out more, but subsequent layers do less and less because the painted pane is already essentially opaque.
This brings us, finally, to the issue of climate models. Essential to alarm is the fact that most current climate models predict a response to a doubling of CO2of about 4C (which is much larger than what one expects the simple doubling of CO2to produce: ie, about 1C). The reason for this is that in these models, the most important greenhouse substances, water vapor and clouds, act in such a way as to greatly amplify the response to anthropogenic greenhouse gases alone (ie, they act as what are called large positive feedbacks). However, as all assessments of the Intergovernmental Panel on Climate Change (IPCC) have stated (at least in the text –though not in the Summaries for Policymakers), the models simply fail to get clouds and water vapor right. We know this because in official model intercomparisons, all models fail miserably to replicate observed distributions of cloud cover. Thus, the model predictions are critically dependent on features that we know must be wrong.
Here we see that treatment of clouds involves errors an order of
magnitude greater than the forcing from a doubling of CO2
Let me summarize the main points thus far:
1. It is NOT the level of CO2that is important, but rather the impact of man made greenhouse gases on climate.
2. Although we are far from the benchmark of doubled CO2, climate forcing is already about 3/4 of what we expect from such a doubling.
3. Even if we attribute all warming over the past century to man made greenhouse gases (which we have no basis for doing), the observed warming is only about 1/3-1/6 of what models project.
We are logically led to two possibilities:
1. Our models are greatly overestimating the sensitivity of climate to man made greenhouse gases, or
2. The models are correct, but there is some unknown process that has cancelled most of the warming.
Note that calling the unknown process "aerosols"does not change this statement since aerosols and their impact are unknown to a factor of ten or more; indeed, even the sign is in doubt.
In arguing for climate alarmism, we are choosing the second possibility. Moreover, we are assuming that the unknown cancellation will soon cease.
How is the second possibility supported?
The IPCC TAR made use of a peculiar exercise in curve fitting using results from the Hadley Centre.
14 15 In the first, we are shown an observed temperature record (without error bars), and the results of four model runs with so-called natural forcing for the period 1860-2000. There is a small spread in the model runs (which presumably displays model uncertainty –it most assuredly does not represent internal variability). In any event, the models look roughly like the observations until the last 30 years. In the first, we are shown an observed temperature record (without error bars), and the results of four model runs with so-called natural forcing for the period 1860-2000. There is a small spread in the model runs (which presumably displays model uncertainty –it most assuredly does not represent internal variability). In any event, the models look roughly like the observations until the last 30 years.
16 We are then shown a second diagram where the observed curve is reproduced, and the four models are run with anthropogenic forcing. Here we see rough agreement over the last 30 years, and poorer agreement in the earlier period. We are then shown a second diagram where the observed curve is reproduced, and the four models are run with anthropogenic forcing. Here we see rough agreement over the last 30 years, and poorer agreement in the earlier period.
17 Finally, we are shown the observations and the model runs with both natural and anthropogenic forcing, and, voila, there is rough agreement over the whole record. It should be noted that the models used had a relatively low sensitivity to a doubling of CO2of about 2.5C. Finally, we are shown the observations and the model runs with both natural and anthropogenic forcing, and, voila, there is rough agreement over the whole record. It should be noted that the models used had a relatively low sensitivity to a doubling of CO2of about 2.5C.
In order to know what to make of this exercise, one must know exactly what was done.
19 The natural forcing consisted in volcanoes and solar variability. Prior to the Pinatubo eruption in 1991, the radiative impact of volcanoes was not well measured, and estimates vary by about a factor of 3. Solar forcing is essentially unknown. Thus, natural forcing is, in essence, adjustable. The natural forcing consisted in volcanoes and solar variability. Prior to the Pinatubo eruption in 1991, the radiative impact of volcanoes was not well measured, and estimates vary by about a factor of 3. Solar forcing is essentially unknown. Thus, natural forcing is, in essence, adjustable.
20 Anthropogenic forcing includes not only anthropogenic greenhouse gases, but also aerosols that act to cancel warming (in the Hadley Centre results, aerosols and other factors cancelled two thirds of the greenhouse forcing). Unfortunately, the properties of aerosols are largely unknown. In the present instance, therefore, aerosols constitute simply another adjustable parameter (indeed, both its magnitude and its time history are adjustable). Anthropogenic forcing includes not only anthropogenic greenhouse gases, but also aerosols that act to cancel warming (in the Hadley Centre results, aerosols and other factors cancelled two thirds of the greenhouse forcing). Unfortunately, the properties of aerosols are largely unknown. In the present instance, therefore, aerosols constitute simply another adjustable parameter (indeed, both its magnitude and its time history are adjustable).
21 Science, 2003This was remarked upon in a recent paper in Science, wherein it was noted that the uncertainty was so great that estimating aerosol properties by tuning them to optimize agreement between models and observations (referred to as an inverse method) was probably as good as any other method, but that the use of such estimates to then test the models constituted a circular procedure. This was remarked upon in a recent paper in Science, wherein it was noted that the uncertainty was so great that estimating aerosol properties by tuning them to optimize agreement between models and observations (referred to as an inverse method) was probably as good as any other method, but that the use of such estimates to then test the models constituted a circular procedure. The choice of models with relatively low sensitivity, allowed adjustments that were not so extreme. The choice of models with relatively low sensitivity, allowed adjustments that were not so extreme.
New uncertainties are always entering the aerosol picture. Some are quite bizarre.
Of course this is the beauty of the global warming issue for many scientists. The issue deals with such small climate forcing andsmall temperature changes that it permits scientists to argue that everything and anything is important for climate.
In brief, the defense of the models starts by assuming the modelis correct. One then attributes differences between the model behavior in the absence of external forcing, and observed changes in ‘global mean temperature’to external forcing. Next one introduces ‘natural’forcing and tries to obtain a ‘best fit’to observations. If, finally, one is able to remove remaining discrepancies by introducing ‘anthropogenic’forcing, we assert that the attribution of part of the observed change to the greenhouse component of ‘anthropogenic’forcing must be correct.
Of course, model internal variability is not correct, and ‘anthropogenic’forcing includes not only CO2but also aerosols, and the latter are unknown to a factor of 10-20 (and perhaps even sign). Finally, we have little quantitative knowledge of ‘natural’forcing so this too is adjustable. Note that the Hadley Centre acknowledges that the "aerosols"cancelled most of the forcing from CO2.
Yet, the ‘argument’I have just presented is the basis for all popular claims that scientists now ‘believe’that man is responsible for much of the observed warming!
It would appear that the current role of the scientist in the global warming issue is simply to defend the ‘possibility’of ominous predictions so as to justify his ‘belief.’
To be fair to the authors of Chapter 12 of the IPCC Third Scientific Assessment here is what they provided for the draft statement ofthe Policymakers Summary: From the body of evidence since IPCC (1996), we conclude that there has been a discernible human influence on global climate. Studies are beginning to separate the contributions to observed climate change attributable to individual external influences, both anthropogenic and natural. This work suggests that anthropogenic greenhouse gases are a substantial contributor to the observed warming, especially over the past 30years. However, the accuracy of these estimates continues to be limited by uncertainties in estimates of internal variability, natural and anthropogenic forcing, and the climate response to external forcing.
This statement is not too bad –especially the last sentence. To be sure, the model dependence of the results is not emphasized, butthe statement is vastly more honest than what the Summary for Policymakers in the IPCC’sThird Assessment Report ultimately presented:
In the light of new evidence and taking into account the remaining uncertainties, most of the observed warming over the last 50 years is likely to have been due to the increase in greenhouse gas concentrations.
In point of fact, the impact of man remains indiscernible simplybecause the signal is too small compared to the natural noise. Claims that the current temperatures are ‘record breaking’or ‘unprecedented’, however questionable or misleading, simply serve to obscure the fact that the observed warming is too smallcompared to what models suggest. Even the fact that the oceans’heat capacity leads to a delay in the response of the surface does not alter this conclusion.
Moreover, the fact that we already have three quarters of the climate forcing expected from a doubling of CO2means that if one truly believes the models, then we have long since passed the point where mitigation is a viable strategy. What remains is to maximize our ability to adapt.
That the promotion of alarm does not follow from the science, isclearly illustrated by the following example.
According to any textbook on dynamic meteorology, one may reasonably conclude that in a warmer world, extratropicalstorminess and weather variability will actually decrease. The reasoning is as follows. Judging by historical climate change, changes are greater in high latitudesthan in the tropics. Thus, in a warmer world, we would expect that the temperature difference between high and low latitudes would diminish. However, it is precisely this difference that gives rise to extratropicallarge-scale weather disturbances. Moreover, when in Boston on a winter day we experience unusual warmth, it is because the wind is blowing from the south. Similarly, when we experience unusual cold, it is generally because the wind is blowing from the north. The possible extent of these extremes is, not surprisingly, determined by how warm low latitudes are and how cold high latitudes are. Given that we expect that high latitudes will warm much more than low latitudes in a warmer climate, the differenceis expected to diminish, leading to less variance.
Nevertheless, we are told by advocates and the media that exactly the opposite is the case, and that, moreover, the modelspredict this (which, to their credit, they do not) and that the basic agreementdiscussed earlier signifies scientific agreement on this matter as well. Clearly more storms and greater extremes are regarded as more alarming than the opposite. Thus, the opposite of our current understanding is invoked in order to promote public concern. The crucial point here is that once the principle of consensus is accepted, agreement on anything is taken to infer agreement on everything advocates wish to claim.
At this point, it is doubtful that we are even dealing with a serious problem. If this is correct, then there is no policy addressingthis non-problem that would be cost-effective. Even if we believe the problem to be serious, we have already reached the levels of climate forcing that have been claimed to be serious. However, when it comes to Kyoto, the situation is even worse. Here, there is widespread and even rigorous scientific agreement that complete adherence to the Kyoto Agreement would have no discernible impact on climate regardless of what one believes about climate. Thus, the theme of this meeting is, at least onthis count, appropriate.
What about the first possibility: namely that the models on which we are basing our alarm are much too sensitive? Not only is this the possibility that scientists would normally have preferred on the basis of Occam’s famous razor, but it is also a possibility for which there is substantial support.
I will focus on one line of this evidence: tropical warming in the 90's has been associated with much greater outgoing long wave radiation than models produce. This discrepancy points toward the absence of a strong negative feedback in current models.
The discrepancy has been confirmed by at least four independent groups.
Chen, J., B.E. Carlson, and A.D. Del Genio, 2002: Evidence for strengthening of the tropical general circulation in the 1990s. Science, 295, 838-841.
Del Genio, A. D., and W. Kovari, 2002: Climatic properties of tropical precipitating convection under varying environmental
conditions. J. Climate, 15, 2597–2615.
Wielicki, B.A., T. Wong, et al, 2002: Evidence for large decadal variability in the tropical mean radiative energy budget. Science, 295, 841-844.
Lin, B., T. Wong, B. Wielicki, and Y. Hu, 2004: Examination of the decadal tropical mean ERBS nonscannerradiation data for the iris hypothesis. J. Climate, 17, 1239-1246.
Cess, R.D. and P.M. Udelhofen, 2003: Climate change during 1985–1999: Cloud interactions determined from satellite measurements. Geophys. Res. Ltrs., 30, No. 1, 1019, doi:10.1029/2002GL016128.
Clement, A.C. and B. Soden(2005) The sensitivity of the tropical-mean radiation budget. J. Clim., 18, 3189-3203.
The preceding papers attempted to either attribute the discrepancy to circulation changes or to ‘unknown’cloud properties –except for the last paper.
Clement and Soden(2005) showed that changes in dynamics could not produce changes averaged over the tropics. They showed this using 4 separate models, but it had been shown theoretically in
Chou, M.-D. and R.S. Lindzen (2004) Comments on "Examination of the Decadal Tropical Mean ERBS NonscannerRadiation Data for the Iris Hypothesis". J. Clim.18, 2123-2127.
Clement and Sodenalso showed that the discrepancy could be resolved by allowing convective precipitation efficiency to increase with surface temperature.
Such a dependence is at the heart of the iris effect which was first found by
R.S. Lindzen, M.-D. Chou, and A.Y. Hou(2001) Does the Earth have an adaptive infrared iris? Bull. Amer. Met. Soc.82, 417-432.
and was theoretically predicted by
Sun, D-Z. and R.S. Lindzen (1993) Distribution of tropical troposphericwater vapor. J. Atmos. Sci., 50, 1643-1660.
In LCH, we attempted to do examine how tropical clouds responded to changing surface temperature, and found that existing satellite data was only marginally capable of dealing with this issue. The results, however, suggested that there were strong negative feedbacks --counter to what models suggest. It was moreover, easy to show that models in no way replicated the cloud behavior that was observed.
Ground based radar allows the almost continuous measurement of precipitation and the separation of convective precipitation from stratiformprecipitation. In the tropics, both types of precipitation originate in condensation within cumulus towers. However, condensation that does not form precipitation is carried aloft as ice which is detrained to form cirrus from which the condensate eventually falls as stratiformprecipitation.
Precipitation efficiency is given by the relation
It now turns out that the rigorous measurement of precipitation can be done with ground based radar.convective precipitationconvective precipitation + stratiform precipitatione=
38 Using data from Kwajalein Atoll in the western Pacific, we were able to study how evaries with sea surface temperature.
In addition, the Kwajalein radar allows one to explicitly look at the area of stratiformrain per unit of convective mass flux.
We see that eincreases about 7.1% per degree C increase in SST (compared with 7.5% estimated by Sun and Lindzen, 1993), and that this increase is associated with a decrease in normalized stratiformarea of about 25% per degree C (which is a bit larger than what was estimated from space observations by Lindzen, Chou and Hou, 2001).
This basically confirms the iris effect, and the fact that models have greatly exaggerated climate sensitivity because, in contrast to models, nature, itself, acts to limit rather than exaggerate the influence of added greenhouse gases.
The primary implication is that for over 25 years, we have based not only our worst case scenarios but even our best case scenarios on model exaggeration.
As far as I can tell, the main question we ought to be confronting is how long the momentum generated by this issue will prevent us from seeing that it has been an illusion based on model error.
What is the implication of these simple results?
In the mean time, we can continue to play our parts in the modern version of "The Emperor’s New Clothes."Our descendents will be amused for generations to come.