Tuesday, January 20, 2009

Properly Framing Skepticism: A Reply to Knappenberger

Update: Please see the comments for further discussion.

A few weeks ago on the MasterResource blog, Paul Knappenberger noted that, in light of generally stagnant global average temperatures over the last decade (and falling temperatures over the last half dozen years), we might justifiably begin to doubt the reliability of the models on which much of the concern over global climate change is based. On the basis of this concern and a few allusions to other lines of evidence which seem to cast doubt on the case for alarm, Knappenberger concludes that:
While coming years may or may not continue the cooling trend of the past several, they will almost assuredly continue to add to the growing evidence that our coming climate will likely be far less detrimental than the popular projections of it to which we are often exposed.

In writing this response, I don't want to create the impression that I think that Knappenberger does a bad job in his post. I think he built a very reasonable case for concern over the reliability of climate models, and to a large extent I agree with him. About a year ago, I wrote a post on this blog discussing some of my reservations about climate modeling, noting that models raise uncertainty in at least five important ways: There may be problems resulting from "tuning" models; the IPCC uses averages from multiple models, potentially distracting us from serious and different flaws in each model; the models cannot effectively model small-scale phenomena which could be important in determining the future state of the climate system; they make use of proscribed variables; and they cannot effectively capture unprecedented, game-changing possibilities without opening themselves up to radical uncertainty. Knappenberger discusses a number of these issues in his post, and I think he does a good job.

But I do have one objection to his handling of the issue with which he concerned himself in his post: his conclusion was not at all supported by his argument. I believe that Knappenberger has mistaken the absence of evidence to be the evidence of absence. He coherently argued that climate models had made a vague prediction about what would happen in the future, and that since the prediction has not obtained as expected, we have reason to doubt that the climate models are reliable. However, he then appears to jump from the claim that climate models (which predict warming) are unreliable to the claim that warming is unlikely. But this jump cannot be sustained without further evidence.

To illustrate the problem, we might imagine that we are on our way to the mall with a particularly superstitious friend when she suddenly exclaims, "I bet we find a parking spot right away; we've gotten green lights the whole way so far, and that usually means I'm going to find a parking spot!" We might be skeptical of this claim, and regard her thesis as entirely unsupported. But what does that mean about the prediction that we'll find a parking spot? It simply doesn't tell us anything. We'd need to throw out our friend's claim entirely and appeal to entirely different lines of evidence to discuss the truth of the matter. In the same way, showing that climate models' predictions of future warming are unreliable does not suggest that there will not be warming. It simply suggests that we should place less emphasis on their predictions in forming our outlooks, or discard them completely.

The reason I belabor this point is that I think there's a broader point to be made here. There is a fundamental and important difference between an argument that says, "The state of scientific knowledge is not advanced enough that we can make a reliable prediction that distressing or catastrophic global climate change will occur in the relatively near future as a result of human activities," and an argument that says, "The state of scientific knowledge is advanced enough that we can make a reliable prediction that distressing or catastrophic global climate change will not occur in the relatively near future as a result of human activities." In his post, Knappenberger supports (effectively!) the first kind of argument. But in his conclusion, he advances the second kind, and without justification. I think it's critical that in thinking about climate change, we make sure to keep in mind this distinction and frame our skepticism in a way that we can defend.

14 comments:

Chip Knappenberger said...

Danny,

(I have left these same comments at MasterResource.org)

Thanks for your thoughtful comments. In a single, rather limited piece, there is only so much I can go into and thus could not discuss the full nature of climate change. However, despite the limited content, I do not believe that I made the error of logic that you claim I did (and in any case, I didn’t mean to).

I quote you as saying “I believe that Knappenberger has mistaken the absence of evidence to be the evidence of absence” and yet in my article, as you quote. I state “they will almost assuredly continue to add to the growing evidence that our coming climate will likely be far less detrimental than the popular projections.” In other words, there is growing evidence that future climate change will not be as detrimental as often portrayed—so, in fact, I have no absence of evidence—it is just that I did not discuss it all in this particular article.

If you would like to see more about the large and growing amount of evidence that global warming is not proceeding along alarmist lines, I invite you to visit the website World Climate Report, where we have been discussing climate change issues for many years and where you will find myriad evidence that a close look at the science reveals a picture that is not nearly as scary as much of the press makes it out to be.

Thanks again for your interest in the topic and your readership!

-Chip

Danny said...

Thanks for dropping by! (I'll cross-post these comments to your blog as well.)

I think perhaps I misinterpreted the statement you made in your post. When you said "they will almost assuredly add to the growing evidence," I read that as saying, "they will strengthen the case made by the growing evidence." If so, then I think my criticism stands; the issues you cited don't weigh in either direction with regard to the strength of the case made by the growing evidence.

But it now occurs to me that you may have meant that phrase in the sense of "they will add another, separate dimension to the case made by the growing evidence." That is, that the case made by the growing evidence can be paired effectively with the notion that competing hypotheses are unreliable. If that's what you meant, then I apologize for the misunderstanding and withdraw my criticism.

Chip Knappenberger said...

Danny,

Thanks again.

I meant that a relative flat temperature history in recent years is another piece of evidence that global warming is not looking to be catastrophic. Add this the modest sea level trends, most up-to-date research on hurricanes that doesn't finger global warming, increasing trends in agricultural yields, declining human sensitivity to heat waves, etc., etc., (all documented at World Climate Report) and you start to build a pretty solid case that while there is an anthropogenic warming pressure on global climate, that the manifest changes won't be, in net, terrible--especially if you are able to adapt—an ability for which we have shown a certain propensity.

-Chip

Danny said...

I'm honestly not sure how you're coming to that conclusion. It seems like in order to defend that claim, you'd need to show something like "The inadequacy of these models should lead us to reject the hypothesis of dangerous climate change in our future," or "The recent absence of a warming trend is captured more accurately in an alternative model of the climate system which does not predict dangerous climate change in our future." But as far as I can tell, you haven't made any such argument in your post.

That's not to say that you couldn't make such an argument, or that there is no support for that argument. It's just that we both know that it's against the rules to draw conclusions about the future state of the climate system on the basis of a decade- or half-decade-long data series. That's particularly true when that data series is incongruent with the general trend. And the scientific basis for concern is not solely based on the idea that the models are reliable, nor is it based on the idea that global temperatures will rise consistently in the forseeable future. So it just seems like using this as evidence against the general hypothesis, when it's actually accommodated by that hypothesis, is a little premature based on what you wrote in your post.

All that being said, I'll definitely take a look through some of the material over at World Climate Report. There's sort of a lot there, though; is there something in particular you had in mind -- maybe a summary or primer?

Thanks so much for carrying on this discussion with me; it's really nice to hash some of this out with someone who knows more about it than I do!

Chip Knappenberger said...

Danny,

We seem to be going around and around.

The slowing of the rise of global temperatures in recent years is but one piece of evidence that future climate change does not appear to be on a course leading to a general disaster.

Another is that the fears of active and intense future hurricane seasons fuelled by rising temperatures seems to be much toned down by recent scientific studies.

Another aspect of the issue is that much of the increase in losses from extreme weather events can be directly explained by changes in population demographics--not climate change. In fact projections of climate change and demographic changes show that changing demographics drive future weather losses to a far greater degree than future climate changes.

Other studies show that adaptations generally trump climate changes...agriculture yields continue to rise, human sensitivity to extreme heat events has been declining, low level ozone concentrations have been declining (across the U.S.).

Still other evidence from Greenland's glaciers suggest that a rapid sea level rise in the coming century is unlikely.

I can go on with other examples.

The point I am trying to make is that there is evidence besides the recent temperature behavior that suggests, to me at least, that the impact of future climate change will be towards the low end of the popular projections.

-Chip

Danny said...

Sorry if I'm a little slow on the uptake here, and thanks so much for having the patience to bear with me!

I was not arguing that there is no good evidence against the predictions of disaster. My point was that the discrepancy between the recent short-term temperature record and the predictions of global climate models is not good evidence towards that conclusion.

To illustrate it another way, imagine if I told you that "My grandmother heard somewhere that the Earth circles the Sun," and used that to support the conclusion that the Earth does in fact circle the Sun. If you responded, "Your grandmother has nothing to do with anything," it would not be an appropriate response for me to say, "Yes, but there is other evidence as well; I just didn't happen to discuss it!" The point would be that the particular argument in question was flawed, not that the conclusion was false.

I guess I should ask you, though: Do you believe that the recent temperature record is itself evidence that global climate change will not become severe? I mean, it's my understanding that the mainstream hypothesis is very much built on the supposition that global average temperatures may fall for significant periods of time. As far as they're concerned, the particular observations you cited in your post are entirely accommodated by their theory. So while the records are problematic for the models which did not predict them, they are not the sort of evidence which would create problems for the general hypothesis of problematic future warming. Is that not the right interpretation?

*** Just as a side note, I'd also want to point out that the IPCC model runs you displayed on your blog used a three-point smoothing, in case you hadn't noticed that (IPCC WG1 Ch10 p.803). I don't know if that really affects anything, though it would be significant if 2009 turns out to be warmer.

Chip Knappenberger said...

Danny,

Ok, I see where you are coming from.

Here is a link to an article that I wrote that looks into the recent temperature behavior in greater detail.

All the recent behavior does is to more strongly establish the longer-term trend (last three decades or so)at a value that lies towards the low end if the IPCC projections.

If the warming continues to proceed (in the long term) near the low end of the projected range, the impacts (which generally scale with warming) will also be less.

So, it is the longer-term temperature behavior, which is further ingrained by recent temperatures, that lead me to think that global warming will be less than commonly projected.

So, add this to the other list of items of evidence.

Does that seem reasonable?

(Yes, I know there is a smoother in the IPCC Figure, and more than that, some of the individual lines are model ensembles--so it is not the perfect figure--but I do have other analyses that I have not gone into that indicate that the recent slowdown lies at or beneath the lowest model projections for the length of the period in question...but this is a topic for another day...and maybe a pubication in the literature).

-Chip

Danny said...

That's a fascinating article; thanks for sharing it with me! Let me see if I understand:

You do show how natural phenomena have superimposed their influence onto what you appear to believe is a secular warming trend of 0.12-0.13 degrees C per year. If that trend continued over the next century, we'd indeed find ourselves at the low end of the IPCC's predictions.

But if I'm not mistaken, the IPCC predicts that warming will accelerate over the course of the coming century, no? Is there something I'm missing in your explanation that would help to explain why this is unlikely?

Thanks!

Chip Knappenberger said...

Danny,

You ask "But if I'm not mistaken, the IPCC predicts that warming will accelerate over the course of the coming century, no?"

Check out Figure 2 from my original article (taken from the IPCC). It depicts the temperature outcomes from their middle-of-the-road emissions scenario. You'll notice that there is not much temperature acceleration.

The climate forcing response to increasing CO2 levels is logarithmic—that is, you need to put in ever more to have the same climate effect. Thus the models tend to show a rather linear response to increasing emissions.

I contend that the models are just a but overly sensitive—and thus the slope of their long-term linear trend is just a bit high.

-Chip

Danny said...

Regarding the acceleration, I guess I was looking at the scenario projections at WG1, Ch10, p. 803. But that's probably besides the point.

Looking at the figures for radiative forcing in the same place, though, you can see that what was a choppy, variability-filled twentieth century is (implausibly) smoothed out into the twenty-first century, with only a tiny margin of error. If the mean temperature projections are built on the obviously false (but likely necessary) premise of a smooth path for radiative forcings, then wouldn't it be likely that the models would be inherently unable to capture the true variability of the climate system? Could it be that in light of the convergence of natural phenomena seemingly contributing to recent low temperatures, we are simply in the middle of one of the downward spikes in forcing not captured in the models' assumptions?

Chip Knappenberger said...

Danny,

Well, the climate models do have internal variability and generally replicate such phenomena as ENSO (which is primarily behind the recent low temperature), so despite smooth prescribed increases in anthropogenic forcing, the models do move jerkily from year to year. Obviously, the more forcing that is prescribed, the more it dominates “natural” variability, and the shorter the periods during which the model output doesn’t climb.

Our current low warming period is pushing the limits (of time and magnitude) of model projections of the impacts (on the global temperature record) of natural variability for the early 21st century. This is what leads me to start to wonder whether something is amiss. Perhaps the models do not capture the true degree of natural variability which means that the size of the impact that natural variability may have on the course of future climate is not fully captured, or perhaps, something else is going on involving feedbacks such that the recent low warming is spawned by increasing greenhouse gases to some degree—this would be something else that the models don’t quite capture well.

At this point in time, it is just speculation, but the longer this period of low warming continues, the less it seems to “fit the models.”

-Chip

Danny said...

Okay, that makes sense; thanks! But linking it back to the original point, it's not clear to me why you would take this as evidence in favor of a less severe warming trend in the future instead of as evidence for more variability than the models are able to capture. Sorry if that should be obvious; this is a lot of information all at once!

Chip Knappenberger said...

Danny,

Well, I take it as evidence (be it perhaps preliminary) of both—since the recent trend is on the low side of things (and adds to a 30 years history of a relatively low trend). It is proof of neither, but simply evidence (which perhaps I am misinterpreting) at this point. Only time will tell for sure.

-Chip

Danny said...

Fair enough. Thanks so much for taking the time to discuss this with me; it's been a pleasure!

Philosophy Blogs - BlogCatalog Blog Directory Libertarian Blogs Add to Technorati Favorites Back to the Drawing Board - Blogged
"Rational philosophy is on the march. It will f--- up all of your sh-- and leave you without any teeth."