Prof. Richard Lindzen (MIT) is often described as the most respectable of the climate ‘sceptics’ and is frequently cited in discussions here and elsewhere. Lindzen clearly has many fundamentally important papers under his belt (work on the QBO and basic atmospheric dynamics), and a number of papers that have been much less well received by the community (the ‘Iris’ effect etc.). Last year, he gave evidence to and answered questions from, a UK House of Lords Committee investigating the economics of climate change, in which he discoursed freely on the science. I’ll try here to sort out what he said.
Firstly, it is clear that Lindzen only signs up to the first point of the basic ‘consensus’ as outlined here previously, that the planet has indeed warmed significantly over the 20th century. While he accepts that CO2 and other greenhouse gases have increased due to human activities, and that this should warm the planet, he does not accept that it is necessarily an important component in the 20th century rise. His preferred option (by process of elimination) appears to be intrinsic variability, but he provides no support for this contention.
In terms of scientific content, his testimony covers a few basic topics: the greenhouse effect, climate sensitivity, aerosol forcing and water vapour feedbacks. We have discussed these topics previously (here, here and here), and so my critique of Lindzen’s comments will come as no surprise. He intersperses his comments with references to ‘alarmism’ which I will get to at the end.
Lindzen accepts the main principle of the greenhouse effect, that increasing greenhouse gases (like CO2) will cause a radiative forcing that, all other things being equal, will cause the surface to warm. He uses an odd measure of its effectiveness though, claiming that a doubling of CO2 will lead to a ‘2%’ increase in the greenhouse effect. How has he defined the greenhouse effect here? Well, a doubling of CO2 is about a 4 W/m2 forcing at the tropopause, which is roughly 2% of the total upward longwave (LW) (~240 W/m2). But does that even make sense as a definition of the greenhouse effect? Not really. On a planet with no greenhouse effect (but similar albedo) the upward LW would also be 240 W/m2, but the absorbed LW in the atmosphere would be zero, so it would make much more sense to define the greenhouse effect as the amount of LW absorbed (~150 W/m2). In which case, doubling of CO2 is initially slightly more*, but as soon as any feedbacks (particularly water vapour or ice albedo changes) kick in, that would increase. Due to the non-linearities in the system, you certainly can’t multiply the total greenhouse effect of ~33 C by 2% to get any sensible estimate of the climate sensitivity. So it’s not clear what relevance the ‘2%’ number has except to make the human additions to the greenhouse effect seem negligible.
*Update: The initial post had an arithmetic error which I have excised (see comment 76 below).
That leads in to Lindzen’s main theme in his evidence – how sensitive climate will be to increasing CO2. He starts off by giving the standard Stefan-Boltzmann no-feedback value for the climate sensitivity: “A doubling of CO2 should lead (if the major greenhouse substances, water vapour and clouds remain fixed), on the basis of straightforward physics, to a globally averaged warming of about 1°C”. But he couples this with an extremely misleading statement: “The current increase in forcing relative to the late 19th Century due to man’s activities [by which he means greenhouse gases alone] should lead to a warming of about 0.76°C, which is already more than has been observed, but is nonetheless much less than current climate models predict.” He repeats this point in the Q&A session as well. However, Lindzen is undoubtedly well aware (having written papers on the subject i.e. Lindzen, GRL, 2002) that lags in the surface temperature due to ocean thermal inertia imply that the transient response is always smaller than the equilibrium response, and that additionally, there are other forcings in the system (specifically land-use change and aerosols) that counteract the forcing from greenhouse gases alone. Since he does not mention these two factors in connection with this statement, a listener could be left with a rather misleading impression. He combines this with a (deliberate?) overstatement (Q130) of the ‘consensus’ value for the sensitivity as being 4 to 5°C (while the actual consensus is between 1.5 to 4.5°C, best guess around 3°C), misleadingly giving the impression that the mainstream is way off. Similarly, his claim that models overpredict the 20th Century temperature rise is easily shown to be false.
Later he states: “Attempts to assess climate sensitivity by direct observation of cloud processes, and other means, which avoid dependence on models, support the conclusion that the sensitivity is low. More precisely, what is known points to the conclusion that a doubling of CO2 would lead to about 0.5°C warming”. One wonders which attempts he is referring to, since it can’t be Lorius et al (1991), or Forrest et al (2004) or Andronova and Schlesinger (2002), given that they give ranges that are all significantly higher than this, and indeed, Gregory et al (2002) specifically rules out anything less than 1.6°C. A more recent estimate (Annan and Hargreaves, in press) using multiple lines of observational constraints places the sensitivity well within the value estimated by the models (i.e. around 2 to 4°C).
Actually, I think it is quite easy to rule out a sensitivity as low as 0.5°C by considering the last glacial period about 20,000 years ago. At that time the temperatures were globally around 5 or 6°C colder than the pre-industrial, and the forcings (from ice sheets, vegetation, greenhouse gases and dust) are estimated to be around 6 to 11 W/m2 (a slightly broader range than I previously quoted, updated from some of the PMIP2 results). This implies a sensitivity of between 1.8 and 4°C for a doubling of CO2, with a most likely value of around 3°C. If however, the sensitivity really was as low as 0.5°C, that would imply that either the forcings estimates are 3 to 8 times too low, or the temperature changes are 3 to 8 times too high. Since around 3 W/m2 of the ice age forcing is directly related to greenhouse gases and is well accepted (even by Lindzen), it would require the ice sheets to impart an enormous forcing even to get anywhere near a level consistent with his sensitivity estimate. That does not appear even remotely plasuible. On the other hand, it is unlikely that we have mis-interpreted the proxy evidence for temperature since it comes from very many different sources – snow lines, foraminefera, alkenones, Mg/Ca, pollen records, ice core isotopes, speleothems, faunal assemblages etc. To be sure, some of these data do not completely agree, but none would imply that global temperatures were only 1.5°C cooler (which is the minimum that would be required).
In summary, Lindzen’s testimony regarding on climate sensitivity is idiosyncratic at best, and certainly not supported by the literature.
He goes on to describe the attribution study of Stott et al. (2000) who showed that both natural (solar and volcanic) and human-related forcings (GHGs and aerosols) were necessary for a climate model to match the 20th Century temperature changes. This is seen in every model (for instance) and so is not the result of some individual model quirk. Lindzen goes on to claim that uncertainty in the forcings (particular solar and aerosols) imply that the result is somehow ‘fixed’ to give the observed result. Since we all agree that there are uncertainties in the forcings (which preclude strong statements about climate sensitivity being derived from the 20th century records for instance), is this criticism valid?
In the absence of any other constraints on either of these forcings and if their value was being defined a posteriori then he may have had a point. However, timeseries for solar forcing have been produced by groups unaffiliated with any modelling group, and the modellers have simply taken the values from the literature (Lean et al, 1995;2000;2005, Hoyt and Schatten, 1998) – any ‘fudging’ to produce the ‘correct’ answer would be immediately obvious. For aerosols, models are needed to produce the 3-dimensional distribution based on independently-derived emission data sets, but the validation is based not on the transient studies over the last 100 years, but on the satellite data and observations over the last 25 years. Once the various unknown parameters have been constrained as much as possible, they are fixed before the transient runs are started. However, aerosol modelling is indeed fraught with uncertainty, and so no group can claim that their resultant transient forcing is the unique best representation of the value found in the real world. Thus in papers such as Hansen et al (2005), it is clearly stated that the results are merely consistent simulations that match the surface temperature response and ocean heat content changes (as well as many other observations) but that this does not rule out a different combination of climate sensitivity and aerosol forcing having as good a match (see this post for more details). The point that needs to be emphasised is that all of these forcings are all very close to the ‘best guesses’ of the aerosol and solar communities.
In the question session (Q143), Lindzen goes into more detail on the reason why he feels that climate sensitivity is so low – specifically, he believes that water vapour feedbacks are not only less positive than models suggest, but actually negative. That is he feels that the amount of longwave aborbtion by water vapour will go down as the planet warms due to increasing GHGs. This implies that actual water vapour amounts will decrease with increasing temperature. On the face of it this is a rather odd claim to make in general – the amount of water vapour that can exist in the atmosphere depends on the Clausius-Clapyeron equation that goes up with temperature. However, it is conceivable that convective processes might cause more extensive drying due to increased areas of subsidence (the basis of the so-called Iris effect), but this applies mainly to the upper troposphere and in the tropics only. As a general effect, reductions in water vapour as temperature increases in general seem rather unlikely.
But we can do better than simply speculating on the issue – we can look at the data and compare that to the models. The best examples to test this idea come from large and relatively rapid changes in the climate such as El Nino events, the eruption of Mt Pinatubo and the trends over the last few decades. In each case (Soden 1997; Soden et al 2002; Soden et al 2005), water vapour increases with warming, and decreases with cooling. There is some uncertainty about exactly how much it increases in the very uppermost troposphere (Misnchwaner and Dessler, 2004), but even those results show a positive feedback. So in summary, the data and the models both agree that not only is the water vapour feedback positive, it is quite close to the value suggested by the models – Lindzen’s insistence on the converse (while it has generated increased attention on the subject) seems increasingly perverse.
In general, I think it is incumbent on scientists when speaking to non-specialists to clearly deliniate what one’s personal opinion is, and what is generally accepted. That is not to say one should not state one’s opinion, but when a panelist specifically asks ‘how far your view of the role of water vapour is shared by other scientists?’ (Q144), one cannot honestly answer ‘That is shared universally’ when no other scientist in the field has made a case for a negative water vapour feedback. This is probably the most egregious mis-statement in the whole testimony and is deeply misleading.
Throughout his testimony, Lindzen refers to the global warming ‘alarmists’. In my dictionary an ‘alarmist’ is defined as ‘a person who alarms others needlessly’. However, Lindzen appears to define as ‘alarmism’ anything that links human activities to climate change. For instance, when discussing the statement from the NRC (2001) report (which he co-authored): The changes observed over the last several decades are likely mostly due to human activities, but we cannot rule out that some significant part of these changes is also a reflection of natural variability., he states that “To be sure, this statement is leaning over backwards to encourage the alarmists”. To my mind, this statement is actually a fair assessment of both the NRC report, and IPCC report to which it was referring. To claim that this is ‘alarmist’ is such a gross overuse of the term as to make it useless except as a rhetorical device.
Lindzen has frequently claimed that within the scientific community “alarm is felt to be essential to the maintenance of funding”. I have yet to see any empirical evidence of this and a brief perusal of active NSF grants related to climate change reveals a lot of interesting projects but none that jump out as being ‘alarmist’. Having sat on panels that decide on funding allocations and as a reviewer of proposals for both US and international agencies, my experience has been that these panels actually do a very good job at deciding which proposals are interesting, tractable and achievable. I have not seen even one example of where the degree of ‘alarmism’ was ever a criteria in whether funding was given. (NB. I don’t regard my own grants (viewable here) as remotely ‘alarmist’ and I don’t have too much trouble getting funding (fingers crossed!)).
In some ways Lindzen’s thinking on the climate change issue has not changed much since 1999, as can be seen in an older rebuttal of his position by Jim Hansen (scroll down to Table 1). However, he does seem to have become convinced that the 20th Century warming is real. What is interesting about the comparison between then and now, is that Hansen made two appeals to the data gathering community to test a) whether water vapour feedbacks can be observed, and b) whether the ocean heat content is increasing in line with the model predictions. It is quite telling that both of these data analyses have since been made and they confirm Hansen’s contentions, not Lindzen’s.