Nature news runs some very readable articles on climate science, but is unfortunately behind a paywall. Which is a shame because they really should be widely read. Here’s a couple of recent beauties:

The Real Holes in Climate Science, (published 21 Jan 2010) points out that climate change denialists keep repeating long debunked myths about things they believe undermine the science. Meanwhile, in the serious scientific literature, there are some important open questions over real uncertainties in the science (h/t to AH). These are discussed openly in the IPCC reports (see for example, the 59 robust findings and 55 uncertainties listed in section 6 of the Technical Summary for WG1). None of these uncertainties pose a serious challenge to our basic understanding of climate change, but they do prevent absolute certainty about any particular projection. Not only that, many of these uncertainties suggest a strong application of the precautionary principle, because many of them suggest the potential for the IPCC to be underestimating the seriousness of climate change. The Nature News article identifies the following as particularly relevant:

  • Regional predictions. While the global models do a good job of simulating global trends in temperature, they often do poorly on fine-grained regional projections. Geographic features, such as mountain ridges, which mark the boundary of different climatic zones, occur at scales much smaller than the typical grids in GCMs, which means the GCMs get these zonal boundaries wrong, especially when coarse-grain predictions are downscaled.
  • Precipitation. As the IPCC report made clear, many of the models disagree even on the sign of the change in rainfall over much of the globe, especially for winter projections. The differences are due to uncertainties over convection processes. Worryingly, studies of recent trends (published after the IPCC report was compiled)  indicate the models are underestimating precipitation changes, such as the drying of the subtropics.
  • Aerosols. Estimates of the effect on climate from airborne particles (mainly from industrial pollution) vary by an order of magnitude. Some aerosols (e.g. suphates) induce a cooling effect by reflecting sunlight, while others (e.g. black carbon) produce a warming effect by absorbing sunlight. The extent to which these aerosols are masking the warming we’re already ‘owed’ from increased greenhouse gases is hard to determine.
  • Temperature reconstructions prior to the 20th century. The Nature News article discusses at length the issues in the tree ring data used as one of the proxies for reconstructing past temperature records, prior to the instrumental data from the last 150 years. The question of what causes the tree ring data to diverge from instrumental records in recent decades is obviously an interesting question, but to me it seems to be of marginal importance to climate science.

The Climate Machine, (published  24 Feb 2010) describes the Hadley Centre’s HadGEM-2 as an example of the current generation of earth system models, and discusses the challenges of capturing more and more earth systems into the models (h/t to JH). The article quotes many of the modelers I’ve been interviewing about their software development processes. Of particular interest is the discussion about the growing complexity of these models, once other earth systems processes are added: clouds, trees, tundra, land ice, and … pandas (the inclusion of pandas in the models is an in-joke in the modeling community) . There is likely to be a limit to the growth of this complexity, simply because the task of managing the contributions of a growing (and diversifying) group of experts gets harder and harder. The article also points out that one interesting result is likely to be an increase in some uncertainty ranges from these models in the next IPCC report, due to the additional variability introduced from these additional earth system processes.

I would post copies of the full articles, but I’m bound to get takedown emails from Macmillan publishing. But I guess they’re unlikely to object if I respond to emails requesting copies from me for research and education purposes…

What makes software engineering for climate models different?
A history of climate modeling

4 Comments

  1. Not only that, many of these uncertainties suggest a strong application of the precautionary principle, because many of them suggest the potential for the IPCC to be underestimating the seriousness of climate change.

    Too often in discussions about climate policy precautionary principle is used as an excuse to forgo rational cost benefit analysis, quickly leading to the triumph of fear over reason (and it provides a convenient foundation for spin in the popular press as well).

    In the context of global climate change, Hammitt (2000) argued that the precautionary principle adds little to benefit cost analysis, which can incorporate aversion to uncertainty as well as severe harms. Similarly, Montgomery and Smith (2000) suggested that an ‘act then learn’ approach based on a rich decision analysis that includes the value of information collected over time is consistent with a weak form of the precautionary principle, but that decision analysis provides more specific (and therefore more useful) guidance in choosing among policy options. Indeed, the vagueness of the precautionary principle is sometimes seen as its primary weakness (Bbodansky, 1991).
    Risk Based Decision Analysis in Support of Precautionary Policies

    Decision theory (some might call it applied probability theory as logic) provides all the tools we need, hand-wavy calls for action based on vague appeals to a precautionary principle are a step backwards for science-based decision support. That paper linked above also has some good discussion about decision when there are multiple stake-holders with differing values. I think plenty of stake-holders will reasonably answer “so what” to the robust findings of the IPCC.

    Yes, there is a robust scientific consensus that human activity is causing the atmosphere to warm up. But so what? Decision-makers need to know how climate change will affect specific political jurisdictions, and, more importantly, what types of interventions will make a difference, over what time scales, at what costs, and to whose benefit—and whose detriment.
    Op Ed by Daniel Sarewitz

    (Thank you for the links today and yesterday to the model history stuff, that’s always interesting.)

  2. Josh: Michael Tobis already posted a thorough counter-argument to Sarewitz:
    http://initforthegold.blogspot.com/2010/03/daniel-sarewitz-is-wrong.html

    I find it interesting that people who argue for “rational cost benefit analysis” feel compelled to add the word “rational” to the noun phrase. Reminds me of snide comments that any field that feels compelled to add the word “science” probably isn’t one :-)

    I don’t think cost benefit analysis is a sensible approach to dealing with global warming for several reasons. First, like most economic theory it explicitly ignores externalities. However the global climate is the biggest externality of all. (Stern calls climate change the biggest ever failure of market systems). Second it’s explicitly geared towards investment decisions, whereas here we’re talking about protecting an asset from a systemic threat (as an analogy, think about how few organizations have any idea how to compute the current value to them of their software infrastructure, and the necessary costs of maintaining it – cost benefit analysis doesn’t work for this question either). The costs of in terms of human life and human suffering from climate change are sufficiently large that no attempt to quantify them in economic terms will adequately capture what most people would agree is at stake. Even Stern does some pretty obnoxious things in quantifying the impact of human deaths in terms of lost economic utility. Finally, cost benefit analysis has never been successfully applied to problems of this scale.

    What we lack is an ethical framework for inter-generational responsibilities (such as “pass on a habitable planet to our children”). Cost-benefit analysis avoids these ethical questions, at a time when we desperately need to address them.

  3. Your arguments about in-applicability due to scale are unconvincing; did you read that paper? All of your objections are actually addressed in it (and your point about externalities being ignored is just incorrect). I only added the ‘rational’ to emphasize how irrational arguments from the precautionary principle tend to be.

    Focusing solely on a single feared outcome, believing only that it would be terrible or that there is some non-zero chance that it will occur, is not a sufficient basis for taking precautionary action.

    If you like, I’ll just say ‘decision theory’ (I prefer the Bayesian flavor) rather than ‘rational cost benefit’. I wouldn’t know anything about fields of science, I just measure stuff…

    You can still justify precautionary policies on a decision-theoretic basis, but you can’t just wave your hands about the possibility of grey goo or mini black holes and justify any policy you want.

    The costs of in terms of human life and human suffering from climate change are sufficiently large that no attempt to quantify them in economic terms will adequately capture what most people would agree is at stake.

    If that were the case to a high certainty then it actually wouldn’t be that hard to put a cost on it (the entire expected value of the world’s economy would be the cost, and any action we could take to prevent the loss would be rational).

    I follow MT’s blog so I’ve seen his response already; he’s not very convincing either and his criticisms of that op-ed are muddled and not to the point, but I’ve already said my piece about ethical / metaphysical confusion on a similar post of his, and you both seem content to avoid taking a rational view of consensus building (aw crap, I said it again), so I think on this one we’ll just have to agree to disagree.

    I am still very interested in what you have to say about the process of building scientific software though, you seem to be one of the few people actually talking about it in public, so please keep it up (and thanks for what you’ve done so far, you’ve got some good students working for you).

  4. > Nature news runs some very readable articles on climate science, but is unfortunately behind a paywall. Which is a shame because they really should be widely read.

    I couldn’t agree more.

    This from their website is interesting; perhaps there’s hope.

    “Nature’s mission statement

    First, to serve scientists through prompt publication of significant advances in any branch of science, and to provide a forum for the reporting and discussion of news and issues concerning science.
    Second, to ensure that the results of science are rapidly disseminated to the public throughout the world, in a fashion that conveys their significance for knowledge, culture and daily life.

    How do we encourage them to fulfill #2? (which I couldn’t have written better – except I’d add “the future”, in the “significance” dept.)

  5. Pingback: The Climate Change Debate Thread - Page 900

Join the discussion: