by Judith Curry
In recent weeks, we are seeing two very interesting debates, both of which involve Richard Lindzen. The first debate involves the recent WSJ op-ed No Need to Panic About Global Warming. The second involves Lindzen’s seminar at the House of Commons.
Lets take a look at the latest response and parries in these ongoing debates.
New York Review of Books
The original op-ed in the WSJ is No Need to Panic About Global Warming. Nordhaus replies in the New York Review of Books Why the global warming skeptics are wrong. The New York Review of Books has published a response by Cohen, Happer and Lindzen entitled In the Climate Casino: An Exchange, which also includes a response from Nordhaus. It is a lengthy exchange, here I focus on the Norhaus’ 6th point related to policy response.
Cohen, Happer and Lindzen:
In his sixth point, Professor Nordhaus says that we did not properly represent his results when we said, “Nearly the highest benefit-to-cost ratio is achieved for a policy that allows 50 more years of economic growth unimpeded by greenhouse gas controls.” The difference between Professor Nordhaus’s optimal carbon tax policy and a fifty-year delay policy is insignificant economically or climatologically in view of major uncertainties in (1) future economic growth (including reductions in carbon emissions intensity); (2) the physical science (e.g., the climate sensitivity); (3) future positive and negative environmental impacts (e.g., the economic “damage function”); (4) the evaluation of long-term economic costs and benefits (e.g., the discount rate); and (5) the international political process (e.g., the impact of less than full participation).
Professor Nordhaus computes a $0.94 trillion difference between the net benefits of the two policies, just 4 percent of the computed maximum $22.55 trillion in supposed environmental damage. Results are given to three or four numerical significant figures. Yet we do not know the underlying driver for all of this, the climate sensitivity, to even one significant figure.
This relatively small difference, indeed whether it is positive or negative, depends critically on factors such as the five listed above, in particular the value of the climate sensitivity. Professor Nordhaus chooses 3.0 degrees C for doubling of CO2,9 a value that empirical evidence suggests is greatly exaggerated.10 To illustrate the point, for a climate sensitivity of 1.0 degree, a value suggested by a number of empirical studies, Professor Nordhaus’s “DICE” model calculates that the optimum policy’s net benefits drop from about $3 trillion to a net cost of about $1 trillion, and the benefit-to-cost ratio plunges from 2.4 to 0.5. The fifty-year-delay policy is then greatly preferred.
We are asked to take the computed difference between the two policies seriously despite Professor Nordhaus’s finding11 that the optimal policy ultimately “saves” only about 0.1 degree C in global warming relative to the fifty-year delay. Putting this in perspective, 0.1 degree is only about 10 percent of the observed warming since 1850 and is a typical year-to-year fluctuation. This tiny difference is predicted by the DICE model to occur fifty years to two centuries in the future, and yet climate models have failed the test of prediction over twenty years. Furthermore, as outlined in our Op Eds, the strong negative environmental impacts assumed in the DICE model’s economic damage function are acknowledged to be extremely uncertain. There exist potential net benefits of increased atmospheric CO2, especially for a small climate sensitivity (e.g., in agricultural and timber productivity).12
Thus, when one considers the nature and magnitude of uncertainties in the climate sensitivity, the economic damage function, and the discount rate, Professor Nordhaus’s defense of a difference in policies that is tiny compared to these uncertainties is difficult to understand.
The larger point here is that uncertainties in the physical science and the economic science need to be properly considered. As suggested above, a key uncertainty in the economic analysis can be treated by examining the economic impact of realistic values of the climate sensitivity. We have seen that a likely small climate sensitivity turns the optimum policy economic values sharply negative. Mother Nature continues to tell us that the climate sensitivity is likely to be below the range considered by Professor Nordhaus.15 This is not surprising because his choices of its most likely value and its statistical “spread” were strongly influenced by a suite of climate models that have exaggerated past warming and that share common problems. These considerations make Professor Nordhaus’s option of a fifty-year delay the wisest policy choice.
Nordhaus replies (excerpts):
The final part of the response of CHL comes back to the economics of climate change and public policy. They make two major points: that the difference between acting now and doing nothing for fifty years is “insignificant economically or climatologically,” and that the policy questions are dominated by major uncertainties.
But the larger point is that climate-change economics and policies are haunted by vast uncertainties. They mention five: economic growth, physical science, the impacts of climate change, politics, and discounting.
The first is a set of threats from climate change to the “world’s cultural and natural treasures” (to cite the words of the UNESCO World Heritage Convention), among them major glaciers, marine and terrestrial biodiversity, archaeological sites, and historical cities and settlements.
A second and even more dangerous uncertainty is caused by “tipping points” in the earth system. Among the global-scale tipping points identified by earth scientists are the collapse of large ice sheets in Greenland and Antarctica, changes in ocean circulation, feedback processes by which warming triggers more warming, and the acidification of the ocean.h
The thrust of CHL’s argument is that the uncertainties are likely to resolve in favor of inaction rather than strong action to slow climate change policies, and in any case, they argue, policies are unimportant given the size of the uncertainties.
However, the major problem with the conclusions of CHL is that they ignore the perils of the climate-change uncertainties. To illustrate, think of the issues as if we are playing roulette in a Climate Casino. Each time the roulette wheel stops, we resolve one of the uncertainties. Our best guess is that CO2 doubling will increase temperatures by 3°C, but if the ball lands on black it will be 2°C while a ball on red will produce 4°C. Similarly, a ball in a black pocket will lead to minimal damages from a certain amount of warming, while a ball in a red pocket will lead to much larger warming than we anticipate. On the next spin, a ball in the black will produce low growth and slow growth in emissions, while a ball in the red will produce rapid growth in CO2 emissions. And so forth.
But, in the Climate Casino, the ball also might land on zero or double-zero. If it lands on zero, we find significant loss of species, ecosystems, and cultural landmarks like Venice. If it lands on double-zero, we find an unanticipated shift in the earth’s climate system, such as a rapid disintegration of the West Antarctic Ice Sheet.
CHL suggest in effect that the ball will always land in the black pocket. We might hope that all the balls land to our advantage on black, but the odds of that outcome on five spins of the wheel are only 1 in 50.i Moreover, when the different uncertainties interact, the outcomes are likely to be even more costly because of nonlinearities in the physical system. For example, assume that the climate uncertainties are larger than we thought and that the impacts were much more damaging than we projected. This would lead to disproportionately larger damages than in the “best-guess” case.
The point is that CHL have the impact of uncertainty exactly backward. A sensible policy would pay a premium to avoid the roulette wheel in a Climate Casino. This means that the economic model estimates of the cost of doing nothing for fifty years are understated because they cannot incorporate all the uncertainties—not just the obvious ones such as climate sensitivity but also the zero and double-zero uncertainties such as tipping points, including ones that are yet undiscovered.
It is possible that the world will not warm over the coming years. It is possible that the impacts will be small. It is possible that a miraculous technology will be invented that can suck CO2 out of the atmosphere at low cost. But in view of the evidence we now have, it would be foolish to bet on these outcomes just because they are possible.
Lindzen’s House of Commons presentation
Lindzen’s presentation to the House of Commons was discussed previously here and here. Several blogospheric critiques were pointed out in Part II. There is a new rebuttal out (h/t BishopHill), written by some of the biggest names in UK climate science: Hoskins, Mitchell, Palmer, Shine and Wolf entitled A critique of the scientific content of Richard Lindzen’s Seminar in London, 22 February 2012. Key excerpts:
We agree that scientific arguments should be based on physical reasoning and data, without exaggerating either the effects or our certainty (or uncertainty) about them. RSL is right to draw attention to uncertainties in climate change feedbacks e.g. associated with clouds. However, it is wrong to infer from this that we know nothing about these feedbacks. Contemporary science suggests unambiguously that there is a substantial risk that these feedbacks will lead to human- induced surface temperature change considerably larger than 1oC in global average this century and beyond.
Temperature and other data
We do agree with RSL that “obsessing” over the global-average temperature is not useful. However a global average is not exactly “an obscure statistical quantity”. It is certainly true that on time-scales of a decade or less it is usually a residual of positive and negative anomalies in different regions and shows considerable year-to-year variability, associated with, for example, natural variations of sea- surface temperatures in the tropical Pacific Ocean. We also agree that “the quantity is easy to abuse”. Unfortunately one of RSL’s slides (Slide 12) on this contains a major error. The slide purports to show that one of the research institutes that performs an ongoing analysis of the global temperature record since 1880 revised its 2008 analysis of past data in 2012 so as to give an increased warming at a rate of 0.14oC per decade. In fact RSL’s figure was obtained by looking at the difference between data from land regions only in the later analysis and land plus ocean in the earlier analysis. Since the land is warming quicker than the ocean, a spurious impression of “manipulation” resulted. RSL has since admitted this error.
At every stage models should be evaluated by exhaustive comparison with observations. The models encapsulate our understanding of the basic science of the climate system, including for example, Newton’s laws of motion, the laws of thermodynamics and the quantum theory of radiation. When deficiencies are found at one level then improvements are sought and the lessons learnt should cascade to models at other levels. This is, of course, the ideal: the actual development of the science is rather more irregular but very definitely in this direction. Even the models at the more complete and complex end contain many uncertainties and deficiencies, which are widely recognised within the modelling community, but they are the best guide we have as to how the climate system may change in the future. Their results are not to be accepted in an unquestioning manner; they should be analysed in detail, with the dominant processes behind any climate variability and change thoroughly investigated using observations and simpler models in the hierarchy.
It is interesting that, given his general scepticism over models, RSL is able to “know that the models are correct”, and hence “some of the recent temperature data must be wrong” – Slide 22, in giving a maximum in warming in the tropical upper troposphere – a hot spot. His view is based on the physical argument that the tropical atmosphere will have a temperature change with height consistent with it being neutral to tropical convection. Whatever the cause of surface warming in the tropics, our current understanding, in agreement with RSL’s, is that this warming will amplify into the upper troposphere. Whether this is consistent with observational data from radiosondes and satellites has been a continuing source of debate, and has been used by those who question the validity of models as strong evidence for their rejection. The current understanding is that models and data are probably consistent within the model uncertainty and observational error, but there is still no firm observational confirmation of the “hot spot”. However, RSL uses the possible conflict between models and data to question the accuracy of the temperature data. Surprisingly his focus is not on the data for the middle and upper troposphere but on the surface data, with the suggestion that the warming there is actually less than analyses have given.
Similarly in slides 34-38, RSL surprisingly invokes the lack of change in surface temperatures during summertime in the Arctic as evidence that “CO2 in not a major player”. It has been well-established for decades, using the same models invoked by RSL for his “tropical upper troposphere hot spot” argument, that Arctic summer temperatures are not expected to increase significantlly, in response to increasing CO2 levels, while sea-ice still exists in the Arctic. The physical reasoning is straightforward: once the Arctic sea-ice has been brought to melting point (as it is during summer), any additional energy goes into melting the ice, rather than raising its temperature.
Climate forcing and sensitivity
On Slide 3, RSL claims that the derived sensitivity of climate to a doubling of CO2 is less than 1C, based on the assumption that all the observed warming is due to atmospheric greenhouse gases. This claim would be wrong even without this assumption, because it confuses the transient warming as CO2 rises with the larger warming that would later be achieved as the oceans, with their large thermal capacity, come into equilibrium with the changed atmospheric state. The assumption itself is unjustifiable as it neglects other mechanisms that drive climate change. RSL notes that high sensitivities are possible only by “invoking unknown additional negative forcings from aerosols and solar variability as arbitrary adjustments”. It is indeed true, as is made clear in successive assessments of the Intergovernmental Panel on Climate Change, that there are considerable uncertainties in estimating the impact of aerosols on climate. However, to characterise these as “unknown” fails to recognise the considerable advances in understanding of the distribution and characteristics of aerosols over recent decades, from individual field campaigns, establishment of new observing networks, and observations of trends in solar radiation reaching the Earth’s surface. These present strong evidence that, in total, changes in aerosol concentrations will have cooled the climate system over recent decades. “Uncertain” does not imply “unknown” which in turn does not justify the assumption that their effects are, therefore, zero.
On the other hand, RSL’s assertion that the water vapour feedback may be negative goes against the body of observational, theoretical and modelling evidence which indicates that it is strong and positive. Modelling and observational studies do not rule out the possibility of a negative cloud feedback, though most models suggest a weak to moderate positive cloud feedback (there is not a strong positive feedback in models as RSL insinuates). In short, there is little credible evidence to support the low climate sensitivities that RSL proposes.
A pervasive aspect of RSL’s presentation was the conflation of uncertainty with ignorance; in his view, because we are uncertain about some aspect, we therefore know nothing about it and any estimate of it is mere guesswork. In this way we believe RSL does a disservice to the scientific method, which seeks to develop understanding in the face of inevitable uncertainties in our knowledge of the world in which we live. The scientific method has served society well for many hundreds of years, and we see no reason to doubt its validity for trying to quantify the risk of climate change and its impacts on society this century. On this basis we reassert that there is a substantial risk of human-induced climate change considerably larger than 1C in global average this century and beyond. There is nothing in RSL’s talk to cast doubt on the existence of this risk. It is up to policy makers, not scientists, to decide whether governments should take concerted mitigating action to try to reduce this risk. On this we do not comment.
Judge Judy’s verdict: I think that these exchanges have been terrific, getting to heart of the scientific and policy issues, and showing some genuine back-and-forth debate.
With regards to the exchange with Nordhaus, IMO the original 16 and the rebuttal written by Cohen, Happer and Lindzen have come out ahead of Nordhaus in this exchange. In the end, it seems to me that Nordhaus is justifying his argument based upon the possibility of truly catastrophic change on the timescale of a century.
With regards to the Hoskins et al. article. There were weaknesses in Lindzen’s argument, and even some bonafide errors. I agree with Hoskins et al. that Lindzen’s high level of certainty that climate sensitivity is 1C is unjustified. That said, I didn’t find the Hoskins et al. rebuttal to be all that effective. So points go to Hoskins et al. on this one, but far from a knockout.
In summary, both of these public discussions in the media and on the internet have been very very good, and valuable contributions to the debate and discussion on climate change.