Disagreeing with the Copenhagen geoengineering consensus
September 7, 2009, 1:19 pm
Filed under: Geoengineering

At roughly the same time as the Royal Society was weighing in on geoengineering last week, so was Bjorn Lomborg’s Copenhagen Consensus project, considerably less convincingly. The context here was a process by which a panel of five economists was briefed on a range of investments that might be made to do something about climate change: for each intervention there was a briefing paper and a pair of discussion papers analysing that briefing. After reading all this and hearing oral arguments the panel voted on which of the interventions was considered the best investment.

The intervention portfolio on offer featured carbon taxes of various strengths, R&D into new energy technologies, carbon sequestration and direct-air carbon capture, planning for adaptation, measures to control methane and black carbon, geoengineering of various types, forest management and expansion and north-south technology transfer. The final ranking was cloud-whitening geoengineering first, energy R&D second, aerosol geoengineering third, then carbon storage, adaptation planning, air capture, forests, methane and black carbon measures and, at the bottom, the four tax options.

Copenhagen Consensus rankings

Copenhagen Consensus rankings

Roger Pielke Jr, who was involved as one of the geoengineering discussants, comments on some of the oddnesses and inadequacies of the process on his blog. As he pointed out in his written submission to the Copenhageners (you can get all the papers involved from Lomborg’s site), the Copenhagen Consensus briefing on the costs and benefits of geoengineering, by Eric Bickel of the University of Texas, Austin and Lee Lane of the American Enterprise Institute, presented a very positive result with a rather spurious aura of accuracy. Obviously Roger’s arguments did not sway the panel, but I think they are pretty good.

The apparent accuracy comes from using a numerical model to link the economy and climate, a model called DICE developed by William Nordhaus. Bickel and Lane looked at solar radiation management (stratospheric haze or cloud whitening) with this model in the context of three strategies: doing nothing else at all, following the “optimal” abatement strategy (a strategy in which DICE sets the amount of money spent on emissions reduction so as to minimise the sum of damage done by warming, expressed in dollars, and the costs of that emissions reduction), and limiting warming to 2ºC. In all those contexts, they look at the effects of adding 1, 2 or 3 W/m^2 of cooling to other measures, if any. Much of what follows is unsurprising; the no-control world gets warmer slower with solar radiation management, and the costs of  the “optimal” strategy are reduced, because there is less harm done by warming while emissions are reduced.

One result that’s worth noting is that left to itself, DICE doesn’t like the 2ºC limit at all — it is seen as more costly than doing nothing because, within the constraints of the model, the costs of emissions control are greater than the economic damage done by going over 2ºC. With just 1 W/m^2 of  geoengineering, though, the situation changes: a strategy aimed at limiting temperature change to 2ºC that makes use of this relatively small amount of solar radiation management works out cheaper than either following the “optimal” path or doing nothing. As the authors put it:

This result is obtained because SRM holds temperatures in check, avoiding climate damages, while society builds the capital and technology necessary to achieve emissions reductions at lower cost. The policy lesson, of course, is that SRM can lower the costs of pursuing non-optimal greenhouse gas control strategies [such as a 2ºC temperature cap], not that non-optimal strategies are harmless.

Since, outside the world of the model, there is a pretty much sure-fire guarantee that greenhouse gas control strategies will be non-optimal (and would be even if there were any real world way of defining that optimum) that seems to me an insight worth taking away.

But in general, the paper fails to convince. The degree to which the watts per square metre removed by solar radiation management cannot really be equated with the watts per square metre added by greenhouse gases is not considered in depth, and while the authors address some of the possible indirect costs of geoengineering — rainfall pattern changes, political strife, risk of intermittency, etc — in their discussion they do not do so, as far as I can see, in their explicit numerical modelling. So they basically end up saying that the up-front costs of geoengineering look likely to be very small in comparison to the costs of greenhouse warming, which is true, but insufficient. Reading the paper you feel that doing the modelling differently, and caring about the uncertainties more, might give you a very different answer . Roger very helpfully shows that this is indeed true by pointing to an analysis by Klaus Keller and colleagues at Penn State recently submitted to the journal Climate Change. This too uses the DICE model, but with somewhat different add-ons and assumptions — as a result it gets radically different and less favourable results (pdf). This makes it hard to see the Bickel and Lee results as robust.

But though to my eyes the paper oversimplifies the issues, fails to be explicit about costs or reasonable about uncertainties, and is constrained by assumptions, Lomborg — who I have frequently enjoyed talking to and with whom I’ve agreed on at least some things in the past — contrives to put an incredibly positive gloss on the results (pdf). This is in large part because, unlike almost everyone else I’ve talked to on the matter, he seems to be happy seeing geoengineering and emissions control as an either/or proposition, rather than, at best, a both/and.

Bickel and Lane offer compelling evidence that a tiny investment in climate engineering might be able to reduce as much of global warming’s effects as trillions of dollars spent on carbon emission reductions.

First, there’s the “might”; it might — it might not. That’s why the research is needed (see below). And then there’s the willingness to ignore the other effects of carbon dioxide. If CO2 levels continue to rise unabated, with solar radiation management counteracting their warming effects, they still result in ocean acidification, massive ecological shifts towards C3 plants which will disturb many tropical ecosystems, and changes in the hydrological cycle. Bickel and Lee make the point that one should compare the hydrological changes in a model with the hydrological effects of unabated warming, which look greater. But why? Why not compare them with the effects of mitigated warming?

Ken Caldeira made the “it’s not either/or” point in a piece on the Copenhagen results in the Washington Post. He amplified his position in email to Joe Romm, who put his comments up on Climate Progress.

If we keep emitting greenhouse gases with the intent of offsetting the global warming with ever increasing loadings of particles in the stratosphere, we will be heading to a planet with extremely high greenhouse gases and a thick stratospheric haze that we would need to main more-or-less indefinitely. This seems to be a dystopic world out of a science fiction story. First, we can assume the oceans have been heavily acidified with shellfish and corals largely a thing of the past. We can assume that ecosystems will be greatly affected by the high CO2 / low sunlight conditions — similar to what Earth experienced hundreds of millions years ago. The sunlight would likely be very diffuse — maybe good for portrait photography, but with unknown consequences for ecosystems.

We know also that CO2 and sunlight affect Earth’s climate system in different ways. For the same amount of change in rainfall, CO2 affects temperature more than sunlight, so if we are to try to correct for changes in precipitation patterns, we will be left with some residual warming that would grow with time.

On top of this Bjorn seems to me to underestimate the scientific and technological uncertainties quite markedly

Many of the risks of climate engineering have been overstated. The biggest challenge is public perception. Many environmental lobbyists oppose even researching climate engineering. This is startling given the manifold benefits. If we care most about avoiding warmer temperatures, it seems that we should be elated that this simple, cost-effective approach shows so much promise.

Public perception may well be a problem for geoengineering research. But to think that it is a greater problem than the science (what do these techniques actually do? how much do they effect ocean currents/precipitation/ozone etc?) the technology (how do you spread aerosols without coagulation at just the right particle size; how do you filter seqwater well enough to pump out ultrafine sprays?) and the governance (whose hand is on the thermostat?) seems daft to me.

It’s hard not to see this sort of over-enthusiasm as at least in part a stick with which to beat greens. There may be some need for such chastisement — the “don’t even think about geoengineering” stance, while understandable as both a statement of a worldview and as a piece of practical politics, is still one to avoid. But in getting so close to the “geoengineering means we don’t have to worry” position, it seems to me that Lomborg does the current debates in the field some disservice.

Incidentally, Roger strongly recommends the analysis paper on energy R&D; I haven’t read it, but look forward to doing so.

Update: I meant to mention Alan Robock’s critique of the Bickel and Lane paper on Real Climate, but forgot.

Further update: Lee Lane responds at length in the comments


2 Comments so far
Leave a comment

First Lomborg tells us climate change isn’t a problem cos it’s not really happening, then he tells us it’s not a problem cos we can spray particulates into the atmosphere. Maybe we should stop getting our knickers in a twist and take a nice warm bath in the vinegary oceans.

Comment by Gaia

Dear Mr. Morton,

Your blog post seems a bit confused. The paper that Eric Bickel and I wrote compares various SRM technologies and concludes that two are most promising. These techniques are marine cloud whitening and stratospheric aerosol injection. The paper recommends, therefore, that an R&D program should spend an average of about $750 million a year over ten years in order thoroughly to explore these two options.

The paper does not recommend deploying either of these technologies at this time. Indeed, it points out that the technologies could not be deployed without further development. Rather, it argues that these concepts offer very large potential benefits and that their direct costs appear to be modest. The paper assessed the potential benefits of SRM as part of an integrated strategy for coping with climate change. Our analysis is in this regard consistent with the Ken Caldeira comment that you cite. Climate policy does not have the option of entirely dispensing with GHG controls. (The real questions relate to how much and when.) Therefore, our paper considered the effects of SRM with three different GHG control regimes. As your blog post accurately states, we found that the size of the potential benefits of SRM depend, in part, on the nature of the accompanying GHG control regime.

The paper also stresses that large uncertainties exist about the indirect costs of SRM. In our review of the literature we found no serious effort to quantify these costs, and it was certainly beyond the scope of this project for us to attempt to quantify them. (My own sense is that this is a task likely to consume several years of intensive research and substantial sums of money.) This lack of analysis presents a problem for any attempt to evaluate SRM as the paper acknowledges time and again. On the one hand, some claims exist that the indirect costs could be large. On the other hand, no real evidence shows that they actually would be. Still less has anyone shown that no means can be devised to limit them.

The logical response to this quandary would seem to be the systematic research effort, and that is what our paper recommends. The very large potential benefits that might be available from SRM imply that a significant R&D investment would be a good bet – albeit an unavoidably uncertain one. The expert panel, after reading our paper, and the two critiques of it by Dr. Pielke and Dr. Smith, respectively, seemed to agree with our conclusion. The panel ranked research into marine cloud whitening as their first priority. It picked research into stratospheric aerosols as their third priority. It rated both as “very good” options. Perhaps you disagree, but that was their conclusion.

Your blog states that our paper “…presented a very positive result with a rather spurious aura of accuracy.” In fact the first sentence of the paper describes the effort as “a preliminary and exploratory assessment”. The final paragraph reads, “While our analysis is preliminary, we believe it makes a strong case that the potential net benefits of SRM are large; the question is whether or not the indirect costs will change the calculus. Only research can answer this question.” Many times between those two statements the paper points to issues that cannot be quantified and yet may significantly change the assessment. It certainly cannot derive from the paper’s failure to warn the reader of its preliminary nature.

So the source of your perception of a “spurious aura of accuracy” remains a bit of a puzzle. Your post does, though, provide a link to Alan Robock’s critique of our work. Robock’s assessment was largely ad hominem, and, at several points, it is patently inaccurate. Your post does not mention our reply. (It is posted on the same website.) Perhaps the aura that troubles you stems as much from one-sided reviews as it does from the actual contents of our paper, but that, of course, is for you to decide.

It may also be useful to point out that SRM is not the only response to climate change that may entail potentially large unintended consequences. Will GHG controls, for example, lead to a global trade war? If one follows the current debate in the U.S. Congress, the possibility seems quite real. The welfare impacts could be very large. Yet few proponents of stringent GHG controls seem inclined to pause to consider the potential harm. Then too, everyone can now see that actual GHG control regimes will be far less cost-effective than are hypothetical global carbon taxes. GHG control regimes will certainly lead to mandates and subsidies for the use of biofuels, nuclear power plants, and many other technologies with possibly nasty side effects. In the biofuels case, as we have already seen, the damage can be on a global scale.

These possible problems do not argue for ignoring the risks of deploying SRM. Still less do they make a blanket case against GHG controls. They do, however, underscore the importance of applying the same standards across the full spectrum of possible responses to climate change. Some economists refer to the “Nirvana fallacy”. In it, an idealized vision of how one strategy might work is compared to a vision of an alternative strategy that more fully reflects the imperfections that political and social reality always bring. Thankfully, your blog post did not make such misleading comparisons. I hope, though, that you, and others, will raise the same kind of questions that you are rightly asking about SRM with regard to the hidden, but real, costs of other more conventional policies. Only a level-playing-field analysis can guide the search for a more optimal mix of climate policies.

Best regards,

Lee Lane

Comment by Lee Lane

Leave a Reply

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 )

Google+ photo

You are commenting using your Google+ 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 )


Connecting to %s

%d bloggers like this: