Reader Marvin was kind enough to pass on Wednesday’s Pajama’s Media article What is—and what isn’t—evidence of global warming to a “highly educated” person (those are not scare quotes, but Marvin’s words).
And that person was sweet enough to respond, in writing. I post that person’s (who is unknown to me) comments below, in block quotes; the bolds are in the original. I answer, point by point.
My remote interlocutor discusses only one argument of my article, the skill of the climate models. I will have much more to say later about the main thrust of my article, which is why most people believe in man-made global warming.
Climate does change on times scales both long and short.
We now have a better appreciation of where and when climate has changed in the past and what its effects may have been on human history.
There has been global warming over the last 150 years concurrent with an increase in atmospheric CO2 unprecendented in the last few millenia.
That we are better equipped to detect and assess climate changes currently underway and their impacts, and to plan for future climate change.
There has not only been warming over the last 150 years. In some periods of time, the temperature has gone up, others down. Further, the certainty we have in the actual temperature measurements has increased in time. It has now reached the point where we can be roughly certain, but not in all areas—yet. Before 1900 or so, we are a lot less sure, and can only state changes to temperatures in a crude way, and only at a few locations. The global mean temperature before that date has a large plus/minus bound.
Given that, I will agree with my interlocutor that some sources would point to temperatures roughly increasing a small amount over the last 150 years. However, many sources (proxies) also point to other eras which were much warmer than it is presently.
CO2 has increased, true. Some of that increase is inarguably man-made. Some of it is natural and the result of the world growing warmer. I would shy away from the word “unprecedented”, especially as many studies point to CO2 and temperature roughly tracking each other. Before industrialization, it was likely the case that temperature increases preceded CO2 increases.
I agree that we can and should plan for climate change. Importantly, I have yet to see anybody attempt to define an “ideal” climate. Lastly, as I always say, it is a trivial fact that humans influence the climate; it is only a question of how much, etc.
What is at issue is whether the rapid increase in atmospheric CO2 and other greenhouse gases is capable, to first order, of inducing or accelerating global warming. To this end a number of climate research centers have been developing computer models that might show the response of climate to increasing CO2 would be; their consensus is that the increasing CO2 is responsible for most of the global warming over the last century, and they predict substantially more global warming over the next century based on various scenarios regarding future anthropogenic greenhouse gas emissions. The article that you sent, and other critiques of these models, have made several points:
The models do not predict periods of global temperature stability during the last century.
The models fail to incorporate some physical processes and feedbacks, on relevant spatial and temporal scales, that could noticeably affect their predictions.
Regarding the second of these critiques we can safely say that continued research will provide new and better ways to model climate, like any other scientific endeavor. In any case, we try to use what we think we know today — unless it can be refuted or demonstrated to have uncertainty that renders it not useful. Some of the earlier refutations centered around inadequate physics have been progressively addressed as the models become more sophisticated, for example with regard to coupling multi-layer ocean circulation in a dynamic way, effects of soil moisture, ice caps and snow cover, solar variation, volcanic eruptions, etc.
It is rational to believe that “continued research will provide new and better ways to model climate” and that model physics will improve. But I do not agree with the suggestion that we use to make decisions a partial, error-prone theory because it is all we have. We do not just have the AGW theory—there are rivals.
There are many competing theories for why the climate takes the temperature values it does. AGW is one of these. So is one we can call, the “Dude, whatever” theory (DWT), which predicts the climate will be just what it was last year with a little added plus/minus. AGW predicts that next year will be warmer than this year, and that the year after that will be warmer than next year, and so on, all with a little plus/minus.
Any climate observation is consistent with both our theories (and many others). DWT has the advantage of being very simple. A sophisticated theory (like AGW) should be able to beat it silly in any kind of forecasting contest.
Well, AGW predictions do not beat DWT predictions; and in fact DWT beats AGW. So what’s the better theory?
DWT also beats the Nothing Ever Changes (NEC) theory, which says (among other things) that the mean global temp will always be a constant (with no plus/minus). NEC theory is the simplest we can think of, DWT is next in complexity, and AGW is the most complex. There are other theories in between the complexity of DWT and AGW that, I think, would beat DWT in making predictions.
Therefore, I claim that it is DWT that should be used in decision making until such a point that the AGW models can spank it.
The article’s critique is more related to evaluation of model predictions. To quote: “But model fitting old data is not direct evidence that the theory behind the model is true. Many alternate models can fit that data equally well. It is a necessary requirement for any model, were it true, to fit the data, but because it happens to is not a proof that the model is valid. For a model to be believable it must make skillful predictions of independent data. It must, that is, make accurate forecasts of the future.” This is an eminently reasonable point but there are some nuances that seem to be ignored, for example that the predictive ability of models may be scale dependent. In particular, detection of climate change especially on regional spatial scales may require multi-decadal future observation to overcome the short-scale inter-annual variability. In other words, a truly post-model test might need to wait a few generations. Given the back-casting skill of the climate models, and their predictions of the effects of future CO2 emissions growth on climate, what is the prudent course of action?
Absolutely, the current climate models might make skillful forecasts for certain regions and scales and not for others. And I agree that a true “post-model test might need to wait a few generations.” I am willing to wait and believe it is a requirement that we do so.
This is because I do not agree that the “back-casting skill of the climate models” inspires us to trust them (my interlocutor uses the word “skill” in its normal, plain English sense, and I use it in its abnormal, technical one). Another way to state this is that we should trust the climate models because the fit they past data well.
But many models would fit that past data well, including DWT. And there are still other rival models which predict, via physical arguments of orbital changes and solar forcing, that we are headed back to our next ice age. Those models could also fit the past data well.
It is only true that the AGW model is the one that had been worked on most assiduously. Orders and orders of magnitude more money and man hours have been devoted to its study. Further, its predictions of temperature increase are only a small percent of the overall observed variability of temperatures—not modeled temps, but their actual, error-prone observations; error which increase our uncertainty in the AGW predictions. Really, we are talking about very small, even barely noticeable changes. So small, that it would be rational to devote energy to rival climate theories.
Note: I am talking about climate model predictions, and not forecasts of what might happen given the climate takes a certain state. Those predictions tend to be apocalyptic, or silly, or both; like the increase in prostitutes in the Philippines.
There still remains the question of the back-casting skill of the climate models. First, the models have been tuned to agree with some large-scale trends, although recent models rely less on tuning. Second, they do not reproduce some multi-year global-scale warming pauses. Third, models disagree on regional scale climate for even for spatial scales on the order of 1000 km. And there are other issues of model fit to data, for example with regard to the ice caps. But, by and large, they are pretty good at getting right the geographic variation of climate in the 20th century. The general spatial patterns of temperature and precipitation are reasonably well reproduced as well as their year to year variability, for example the major rainforests and deserts are where they should be. The models have also been used to see what the climate of the 20th century might looked like if there were no increases in atmospheric CO2. Comparisons of this kind are sometimes used to detect ‘signatures’ of CO2-induced climate change by comparison with observed climate changes during the same period, for example by a close examination of daily minimum and maximum temperatures and seasonal variations.
Even though models circa 2009 are tuned less than those circa 1984, they all still rely heavily on tuning, which includes the process of “analysis.” This is where the raw observations are statistically manipulated so that they fit into model space. This is not an unusual process—meteorological models make use of these tricks—but it is a form of tuning (I use the word “trick” like we normally do, as a technique).
I’ll agree that the models can be used in the sense of making sure the deserts are where they should be and so forth. I would not suggest that our models can’t say anything useful about the physics of the atmosphere; clearly they can. The signatures of CO2 are the point in question, however.
Let’s recall that these models were not built in isolation: they were designed with CO2 in mind, and tuned to those physics. Would it not be interesting to develop a suite of GCMs without this explicit call to CO2 forcing? To, that is, attempt to discover other physics that could account for the observed temperatures?
The models will not reproduce the details of the historical record because ‘natural’ variability will obscure trends on short time scales. But the failure of the climate models to reproduce multi-year pauses in global warming is still perplexing to some modelers. I am not so concerned because I feel that the unmodeled short-scale natural variability would not preclude such patterns even over a decade or two — but I haven’t done any analysis to support this belief. In short, my belief is that the models provide a good enough representation of the consequences of increasing atmospheric greenhouse gases.
It is true that the physical processes that the models were unable to represent, and that caused those “trends on short time scales” (something caused them), might be unimportant in the long run. AGW theory is consistent with this. In fact, any observation of the climate is consistent with AGW. No matter how much cooling we see, it still might get hotter in the future.
Meanwhile, another interpretation simply states that the models have—so far—failed to make correct independent predictions because those physical processes that the models were unable to represent, and that caused those “trends on short time scales”, are extremely important in the long run.
Really, the only point at which I and my interlocutor disagree is in how much to trust the models. I am not yet ready to trust them to the extent that they be used to create rules and regulations that are mandated by force of law. I am willing, as I assume my questioner is, to pay, in the form of taxes, for more research on climate—to, that is, continue to do what we have been doing. I would not support new taxes.