Predicting Climate (updated with response)

The problem with James Lewis' argument (SCIENCE: Earth climate is too complex to predict) is that uncertainty (as deduced by Roe and Baker) scales with the predicted amount of warming. Moreover, the probability distribution is asymmetrical, with a long tail to the right - see their figures 2-4. In other words, with a predicted warming of say 2.5 degrees C, the likelihood of the actual warming being twice as large is substantially greater than that of no warming at all. Indeed, all of the distributions in their paper are bounded away from zero in the direction of positive warming.

I have long believed that climate modeling as practiced by the IPCC replaces poorly understood systems with poorly understood models and further that linear modeling and the omission of negative feedback as implemented in most climate models foreordains the observed results when one crunches the numbers. The fact that there is uncertainty in prediction is a technicality. If warming equals bad, the uncertainty discussed in the Science paper is about how bad things will get, not whether they will get bad.

Since not even the draconian measures favored by climate alarmists will do much to slow increasing CO2 emissions, it is fortunate that most climate models, their complexity not withstanding, omit much. Among the omissions are factors that could plausibly overwhelm the effects of increasing greenhouse gas concentrations - see, for example, the work of Svensmark and his associates on solar variability, cosmic rays and cloud cover. Essentially, these authors argue that during periods of solar quiescence, the earth is bombarded with greater numbers of cosmic rays which in turn, promote cloud formation and the reflection of greater amounts of the sun's radiation back into space before it warms the earth. Correspondingly, when the sun is more active, a larger fraction of cosmic radiation is deflected away from the earth, cloud formation is reduced and solar-induced warming, enhanced. Thus, variations in the sun's activity are amplified by the cosmic ray - cloud connection. In short, it's not the amount of radiation sent our way that's important, but what gets through.

Such amplification, so far as I am aware, is not considered in any of the IPCC models, which, if they consider solar variability at all, restrict said variation to changes in radiative output.  The sun's activity being variable over periods ranging from days to decades to centuries, Svensmark and his associates argue that historical climate fluctuations, including the Medieval Optimum and the Little Ice Age, were driven by changing activity of the sun. Whether or not this is correct, it certainly not unreasonable to demand that climate modelers be able to reproduce past fluctuations which perforce were not driven by human activity. And if Svensmark is right, the next period of cooling may be just around the corner - see, for example, Jerry Pournelle's prophetic Fallen Angel published some years back.

I agree that calls for a rush to action amount to a hysteria inducing campaign, in many cases calculated by their promoters to serve a variety of "laudable" ends - see the recent letter by Dan Botkin in Opinion Journal. Not the least of these ends, I believe, is enhancement of the promoters own influence and prestige. And I am well aware, by virtue of my employment as an academician, of efforts to silence dissenters from the so-called consensus view. History is replete with similar examples - none of them especially flattering.  What would be helpful in this regard is not an analysis of uncertainty inherent in models that build in run away warming at the outset, but of the historical accuracy of environmentalist doom and gloom. Without having undertaken such a survey, it is my sense that most such warnings proved false - the famous Ehrlich-Simon wager, for example. Moreover, and again without having done the analysis, it seems to me that when one prediction fails, it is simply replaced by another, usually more dire. At least that is my impression from having listened to this sort of thing for 35 years. Similar perceptions, I would imagine, motivate many who counsel moderation today. In this regard, unfortunately, the paper cited by Lewis offers little solace.

W. M. Schaffer, Ph.D.
Professor of Ecology and Evolutionary Biology

James Lewis responds:

Reply to Professor Schaffer: Uncertainty dooms climate models.
In response to my summary of two Science magazine articles and a New Scientist report based on those articles, Professor Schaffer argues that "The fact that there is uncertainty in prediction is a technicality. If warming equals bad, the uncertainty discussed in the Science paper is about how bad things will get, not whether they will get bad."

I appreciate Prof. Schaffer's qualifications in these matters, and particularly his sensible skepticism about highly debatable climate models that are constantly headlined by popular quacks to predict the end of civilization --- if we all don't immediately swallow some unbelievably expensive snake oil.

However, I can't agree with Professor Schaffer's interpretation of the Science paper by Roe and Baker (Science, 318, 5850, 629-632). As the authors write,
"...it is evident that the climate system is operating in a regime in which small uncertainties in feedbacks are highly amplified in the resulting climate sensitivity. We are constrained by the inevitable: the more likely a large warming is for a given forcing (i.e., the greater the positive feedbacks), the greater the uncertainty will be in the magnitude of that warming."
Since "small uncertainties in feedbacks" are inevitable in the climate system, this means plainly that climate sensitivity to inputs like CO2 are unpredictable, and the bigger they might be, the less we will be able to predict them. That is a limit on climate modeling as such. Indeed, that is the plain interpretation given by two invited public commentators writing a Perspective piece in the same issue of Science (318, 5850, 582 - 583. Myles R. Allen and David J. Frame simply call their piece: "ATMOSPHERE: Call Off the Quest. "  They conclude:
"An upper bound on the climate sensitivity has become the holy grail of climate research. As Roe and Baker point out, it is inherently hard to find. It promises lasting fame and happiness to the finder, but it may not exist and turns out not to be very useful if you do find it. Time to call off the quest." (italics added)
Thus for Allen and Frame, the degenerative uncertainty in climate models is not, as Professor Schaffer suggests, a "technicality." It goes to the very heart of the question whether climate modeling is worth doing. Why should we invest vast amounts of money and emotional turmoil in predictions of gloom that cannot in principle be verified?   Scientists sometimes like to believe that we have infinite amounts of money for science. We don't. Money spent on dubious climate models is unavailable for AIDS and breast cancer research, and for other important scientific and social needs.  We must allocate our science resources rationally to scientifically useful aims, and the Roe and Baker paper just demonstrated that this is not possible.

It is also not quite accurate to say that "the uncertainty discussed in the Science paper is about how bad things will get, not whether they will get bad." The Roe & Baker paper is not a fact-based prediction of doom. It is a theoretical paper about climate models. As Professor Schaffer understands very well, the current predictions of doom are dubious indeed.

If the Roe and Baker paper really predicted a world catastrophe, the question would be why the planet has not burned up many times before. After all, there are always positive feedback loops in the climate that have uncertainties built into them. If Professor Schaffer were right, those uncertainties must have led to multiple hell-fire periods in the history of the earth, using the Roe & Baker argument. But that is empirically false.

The key is that Roe & Baker make a mathematical argument about climate models --- not an empirical prediction about rises in earth temperatures. As Allen and Frame point out, their Science paper raises the most fundamental doubts about the climate modeling. Roe & Baker show that climate modeling is a degenerating effort, in which the kind of certainty constantly proclaimed by global doom peddlers is simply, mathematically  unachievable.

Money invested in climate modeling has rapidly diminishing returns. That spells doom, not for the earth, but for the prophets of doom themselves --- who deserve to be discredited for playing on popular fears and peddling dubious science, as Professor Schaffer correctly points out.


The problem with James Lewis' argument (SCIENCE: Earth climate is too complex to predict) is that uncertainty (as deduced by Roe and Baker) scales with the predicted amount of warming. Moreover, the probability distribution is asymmetrical, with a long tail to the right - see their figures 2-4. In other words, with a predicted warming of say 2.5 degrees C, the likelihood of the actual warming being twice as large is substantially greater than that of no warming at all. Indeed, all of the distributions in their paper are bounded away from zero in the direction of positive warming.

I have long believed that climate modeling as practiced by the IPCC replaces poorly understood systems with poorly understood models and further that linear modeling and the omission of negative feedback as implemented in most climate models foreordains the observed results when one crunches the numbers. The fact that there is uncertainty in prediction is a technicality. If warming equals bad, the uncertainty discussed in the Science paper is about how bad things will get, not whether they will get bad.

Since not even the draconian measures favored by climate alarmists will do much to slow increasing CO2 emissions, it is fortunate that most climate models, their complexity not withstanding, omit much. Among the omissions are factors that could plausibly overwhelm the effects of increasing greenhouse gas concentrations - see, for example, the work of Svensmark and his associates on solar variability, cosmic rays and cloud cover. Essentially, these authors argue that during periods of solar quiescence, the earth is bombarded with greater numbers of cosmic rays which in turn, promote cloud formation and the reflection of greater amounts of the sun's radiation back into space before it warms the earth. Correspondingly, when the sun is more active, a larger fraction of cosmic radiation is deflected away from the earth, cloud formation is reduced and solar-induced warming, enhanced. Thus, variations in the sun's activity are amplified by the cosmic ray - cloud connection. In short, it's not the amount of radiation sent our way that's important, but what gets through.

Such amplification, so far as I am aware, is not considered in any of the IPCC models, which, if they consider solar variability at all, restrict said variation to changes in radiative output.  The sun's activity being variable over periods ranging from days to decades to centuries, Svensmark and his associates argue that historical climate fluctuations, including the Medieval Optimum and the Little Ice Age, were driven by changing activity of the sun. Whether or not this is correct, it certainly not unreasonable to demand that climate modelers be able to reproduce past fluctuations which perforce were not driven by human activity. And if Svensmark is right, the next period of cooling may be just around the corner - see, for example, Jerry Pournelle's prophetic Fallen Angel published some years back.

I agree that calls for a rush to action amount to a hysteria inducing campaign, in many cases calculated by their promoters to serve a variety of "laudable" ends - see the recent letter by Dan Botkin in Opinion Journal. Not the least of these ends, I believe, is enhancement of the promoters own influence and prestige. And I am well aware, by virtue of my employment as an academician, of efforts to silence dissenters from the so-called consensus view. History is replete with similar examples - none of them especially flattering.  What would be helpful in this regard is not an analysis of uncertainty inherent in models that build in run away warming at the outset, but of the historical accuracy of environmentalist doom and gloom. Without having undertaken such a survey, it is my sense that most such warnings proved false - the famous Ehrlich-Simon wager, for example. Moreover, and again without having done the analysis, it seems to me that when one prediction fails, it is simply replaced by another, usually more dire. At least that is my impression from having listened to this sort of thing for 35 years. Similar perceptions, I would imagine, motivate many who counsel moderation today. In this regard, unfortunately, the paper cited by Lewis offers little solace.

W. M. Schaffer, Ph.D.
Professor of Ecology and Evolutionary Biology

James Lewis responds:

Reply to Professor Schaffer: Uncertainty dooms climate models.
In response to my summary of two Science magazine articles and a New Scientist report based on those articles, Professor Schaffer argues that "The fact that there is uncertainty in prediction is a technicality. If warming equals bad, the uncertainty discussed in the Science paper is about how bad things will get, not whether they will get bad."

I appreciate Prof. Schaffer's qualifications in these matters, and particularly his sensible skepticism about highly debatable climate models that are constantly headlined by popular quacks to predict the end of civilization --- if we all don't immediately swallow some unbelievably expensive snake oil.

However, I can't agree with Professor Schaffer's interpretation of the Science paper by Roe and Baker (Science, 318, 5850, 629-632). As the authors write,
"...it is evident that the climate system is operating in a regime in which small uncertainties in feedbacks are highly amplified in the resulting climate sensitivity. We are constrained by the inevitable: the more likely a large warming is for a given forcing (i.e., the greater the positive feedbacks), the greater the uncertainty will be in the magnitude of that warming."
Since "small uncertainties in feedbacks" are inevitable in the climate system, this means plainly that climate sensitivity to inputs like CO2 are unpredictable, and the bigger they might be, the less we will be able to predict them. That is a limit on climate modeling as such. Indeed, that is the plain interpretation given by two invited public commentators writing a Perspective piece in the same issue of Science (318, 5850, 582 - 583. Myles R. Allen and David J. Frame simply call their piece: "ATMOSPHERE: Call Off the Quest. "  They conclude:
"An upper bound on the climate sensitivity has become the holy grail of climate research. As Roe and Baker point out, it is inherently hard to find. It promises lasting fame and happiness to the finder, but it may not exist and turns out not to be very useful if you do find it. Time to call off the quest." (italics added)
Thus for Allen and Frame, the degenerative uncertainty in climate models is not, as Professor Schaffer suggests, a "technicality." It goes to the very heart of the question whether climate modeling is worth doing. Why should we invest vast amounts of money and emotional turmoil in predictions of gloom that cannot in principle be verified?   Scientists sometimes like to believe that we have infinite amounts of money for science. We don't. Money spent on dubious climate models is unavailable for AIDS and breast cancer research, and for other important scientific and social needs.  We must allocate our science resources rationally to scientifically useful aims, and the Roe and Baker paper just demonstrated that this is not possible.

It is also not quite accurate to say that "the uncertainty discussed in the Science paper is about how bad things will get, not whether they will get bad." The Roe & Baker paper is not a fact-based prediction of doom. It is a theoretical paper about climate models. As Professor Schaffer understands very well, the current predictions of doom are dubious indeed.

If the Roe and Baker paper really predicted a world catastrophe, the question would be why the planet has not burned up many times before. After all, there are always positive feedback loops in the climate that have uncertainties built into them. If Professor Schaffer were right, those uncertainties must have led to multiple hell-fire periods in the history of the earth, using the Roe & Baker argument. But that is empirically false.

The key is that Roe & Baker make a mathematical argument about climate models --- not an empirical prediction about rises in earth temperatures. As Allen and Frame point out, their Science paper raises the most fundamental doubts about the climate modeling. Roe & Baker show that climate modeling is a degenerating effort, in which the kind of certainty constantly proclaimed by global doom peddlers is simply, mathematically  unachievable.

Money invested in climate modeling has rapidly diminishing returns. That spells doom, not for the earth, but for the prophets of doom themselves --- who deserve to be discredited for playing on popular fears and peddling dubious science, as Professor Schaffer correctly points out.