Monday, August 18, 2008
Hathaway and the IPCC: both simply ignore contrary evidence
Notice anything strange about the solar cycle predictions advanced by NASA solar physicist David Hathaway? Current levels of solar activity lie well below the confidence interval put forward in Hathaway's early 2006 projection of solar activity, but this failure of his prediction scheme has no effect on the confidence with which he continues to predict an imminent and robust take off of solar cycle 24:
Hathaway's March 2006 sunspot prediction overlaid with his June 2008 sunspot prediction.
The solid middle line in this graphic represents Hathaway's maximum likelihood prediction for the smoothed average sunspot number. The dotted lines demark the 95% confidence interval given by Hathaway's prediction scheme.
If the smoothed average sunspot number goes outside of the predicted confidence interval, we should have 95% confidence that Hathaway's prediction scheme is WRONG. (If the scheme was right, the smoothed average number would only split off from the confidence interval 5% of the time.)
Hathaway's 2006 95% confidence interval ranged about 23 sunspots per year above and below his maximum likelihood prediction. The actual sunspot rate is now about 75 sunspots/year below his March 2006 prediction. Thus we can have probably about 99.9% confidence at this point that Hathaway's prediction scheme is wrong.
This evidence of an errant prediction scheme ought to prompt revision of the scheme, but it is evident from the comparison above that instead of revising his prediction scheme, Hathaway is doubling down. Notice how Hathaway responds to what was for him an unexpected year and a half lull in solar activity. He actually shortens the time horizon within which he expects the next solar cycle to erupt:
In March 2006, Hathaway's maximum likelihood prediction had solar cycle 24 turning upwards 5 or 6 months later. Two months ago Hathaway was predicting up-turn in two months.
How can this possibly be in response to the latest solar behavior (a year and a half of unexpected quiet)? Simple answer: it isn't.
Prediction schemes are SUPPOSED to be weighted by their success
According to NASA's June 2008 explanation of Hathaway's prediction scheme, Hathaway and his colleagues base their prediction of how future solar cycles will progress on a composite of different prediction schemes. Some of the prediction schemes have a theoretical basis. Some are simply statistical projections of past patterns. All are weighted commensurate with past performance. To the extent that theoretical considerations seem to improve predictive ability, they are weighted accordingly in the composite prediction.
But all of that goes out the window when it comes to predicting the length of solar minimums. Here there is no weighting of different schemes according to their performance. Rather:
That leaves no avenue for new information to enter the Hathaway/NASA prediction scheme. As NASA explains:
That makes no sense. We know that extended minimums do occur, and should be included as a possibility in any prediction scheme. In the past, when a solar cycle tailed off for longer than the sharpness and amplitude of its rise would have predicted, how long did it tend to stretch out, and on what pattern? The recent failures of Hathaway's prediction scheme should prompt a shift in attention to the past behavior of such statistical exceptions. Hathaway's latest prediction may well be right. Solar cycle 24 could fire up any day now. But it can't be the maximum likelihood projection at this point.
Hathaway failure to address Maunder Minimum type solar behavior is systematic
It isn't that Hathaway is trying to give precedence to theory-based prediction schemes over mere statistical projection. (Not that that would be valid in any case. Prediction schemes should be weighted by performance, period.) Hathaway is simply giving priority to his own purely statistical model. As he wrote in his 1999 paper:
Hathaway certainly knows that there are statistical patterns (past events) that don't fit the cycle shape he is projecting. William Briggs notes that, when Hathaway back-tests his own predictive scheme, he only looks at the sunspot record AFTER the Maunder Minimum (during which sunspot activity went into a 50 year lull). (For the date range Hathaway uses, see figure 1 here.)
Hathaway continues today in the same vein, specifically avoiding consideration of Maunder Minimum type solar behavior. According to NASA, Hathaway simply "believes" that "the quiet of 2008 is not the second coming of the Maunder Minimum."
Hathaway's predictive scheme should be weighed in NASA's composite prediction scheme according to its performance, as NASA does with the other parts of its composite. Cycle 23 is not matching the usual cycle shape, at which point a rational predictive scheme would start paying more attention to the statistical properties of past such anomalies. But Hathaway still gives 100% weight to his so-far non-performing prediction scheme.
The IPCC does the same thing
Hathaway's persistence in predicting a strong solar cycle 24 does not in itself suggest any agenda. If the sun resurges, it will warm the earth, allowing the global warming alarmists to continue to pretend that warming is being caused by fossil fuel burning, but a PREDICTION about what the sun will do does not expose or hide anything.
With one exception. By refusing to even consider the possibility of an extended solar lull, Hathaway himself can avoid having to address the impact that such a lull would have on global temperature. If he did address that question, he would obviously have to note that the last really long such lull seems to have caused the Little Ice Age, which would place him on the side of the "deniers" in the debate about human-caused global warming. If low solar activity caused the Little Ice Age, then the "grand maximum" levels of solar activity during the 20th century would be the cause of 20th century warming, and the hoax of human-caused warming would be exposed.
Looking at the science alone, however, Hathaway cannot be lumped with the IPCC. Most importantly, there actually is a solid body of evidence behind his predictive scheme. Past solar cycles HAVE tended to follow a particular pattern, where the shape of the up-phase predicts the shape of the down-phase.
In contrast, there is no evidence whatsoever for the theory of human-caused global warming. In theory, additional CO2 should have SOME heat trapping effect, but there is no sign of it in the geological record. What warming effect CO2 has is evidently too small to measure, and this is just what we would expect.
CO2 mostly traps the same wavelengths of infrared that the vastly more abundant water vapor, leaving very little heat trapping work for marginal increases in CO2 to perform. Thus theory itself says that the marginal effect of CO2 on temperature should be miniscule.
Where Hathaway doesn't look so good is in what he leaves out. This man's job is to study sunspot activity, and he is ignoring the Maunder Minimum???? But even here, he is not nearly as bad as the IPCC. Of course Hathaway should be highly focused at this point on the phenomenon of extended solar minima, but the truth is that we know almost nothing about these minima beyond that fact that there was a fifty year lull a couple hundred years ago.
In the case of the IPCC, the alternative prediction scheme that they consciously ignore is based on a very well defined and certain cause of global warming. The geologic record proves beyond any doubt whatsoever that global temperature has historically been driven almost entirely by the level of solar activity.
High levels of solar-magnetic activity shield the earth from galactic cosmic radiation (GCR). The level of GCR reaching earth can be measured from isotope signatures deposited in the geologic record, which also contains contemporaneous temperature signatures. Study after geologic study has found that the level of GCR "explains" statistically about 90% of global temperature variation on every time scale. (Fred Singer's book Unstoppable Global Warming: Every 1500 Years compiles the scientific evidence for laymen.)
We also have a good theory of the mechanism by which the level of GCR (and the sun's blocking of it) affects global temperature. (See Henrik Svensmark's book The Chilling Stars.) But having a working theory is just gravy. We don't need to know the mechanism in order to prove causality. A 90% correlation on all time scales over hundreds of thousands of years cannot be coincidence, and the causality can only go one way. It is not the temperature of the earth that is determining the level of solar activity, or the influx of GCR.
The IPCC's excuse for not taking solar magnetic effects into account is that the mechanism is not understood, but that is not a legitimate reason. The mechanism does not have to be understood for us to know that some such mechanism is at work, and how it will affect global temperature as solar activity changes. Any prediction that does not exploit the predictive value of that known physical relationship is unscientific.
Scientific fraud
The IPCC ought to be using the composite prediction scheme that NASA is nominally using for predicting sunspot activity (but not actually using when it comes to predicting how long a solar cycle will last). They should be exploiting the predictive value of ALL of the different available predictive mechanisms, whether they model the mechanisms behind known physical relationships, or just model the statistical relationships themselves.
In the case of global warming, MOST of the weight would go to the very well established causal relationship between GCR and global temperature, and NONE of it would go to CO2 effects for which there is NO evidence in the geologic record. (CO2 follows global temperature, as warming oceans release CO2, but there is no evidence that it has or can drive temperature in any significant way.)
Hathaway is tainted with a little of the same stain, because he too is fixating on what the evidence suggests is a failed prediction scheme to the exclusion of alternate possibilities that he obviously should be giving some weight to. He just doesn't go nearly as far in this direction as the IPCC, which gives 100% of their predictive weight to a theory for which there is NO evidence, while completely refusing to account or even acknowledge what we KNOW is the predominant driver of global temperature.
"Grand maximum" levels of solar activity during the 20th century easily explain the small amount of late 20th century warming, leaving little or nothing to be explained by alternative mechanisms like human burning of fossil fuels. The IPCC's particular fraud, for which there is no analog in Hathaway's work, springs from the coincidental correlation between solar activity and the human burning of fossil fuels over the 20th century.
When an explanatory variable is omitted from a statistical estimate, the explanatory power of the omitted variable will be misattributed to any correlated variables (however causally unrelated). The IPCC intentionally omits the solar-magnetic variable, knowing full well that its overwhelming explanatory power will be misattributed to the coincidentally correlated increase in atmospheric CO2. They then knowingly project this misattributed explanatory power forward in order to create a dishonest political attack on fossil fuel burning.
(I traced this ruse through the 4th IPCC draft report in the comments that I submitted for draft review.)
This scientific fraud by the IPCC is evil. In this context, it is disappointing that Hathaway is not eager to discuss the Maunder Minimum, and explain how every solar scientist knows about the well established correlation (and the implied causal relationship), between sunspot activity and global temperature. That he dodges this question by unscientifically failing to address Maunder Minimum type episodes in his analysis of our present solar lull does not speak well for his integrity.
UPDATE: Tim Ball also slams IPCC for omitting solar-magnetic mechanism, despite overwhelming evidence for it
The IPCC's excuse for omitting solar-magnetic effects from their prediction scheme is that the mechanims is not understood. Ball rejects this on the grounds that we actually have from Svensmark a very good theory:
The IPCC's position is like saying that, prior to Newton's and Einstein's theories of gravitation, the best prediction for what a rock would do when released into the air is that it would waft on the breeze. After all, we understand the mechanism of the air pushing on the rock, but we don't understand the mechanism of this invisible force that pulls a rock towards the earth, so it ought to be omitted from our predictions.
I discussed this last month in my post about the pair of frauds from the American Physical Society who claim that hundreds of thousands of years of close correlation between solar activity and global temperature does not imply causality. The hell it doesn't!
Other related posts
This post shows how the expected environmental impacts of CO2 burning are actually positive, and by a wide margin. Thus the proper micro-economic tax on fossil fuel burning, if there is to be one, would be negative (a subsidy), not the perverse high taxes and cap-and-trade systems that are expanding today.
My posts on the basic science of the global warming fraud here and here.
Hathaway's March 2006 sunspot prediction overlaid with his June 2008 sunspot prediction.
The solid middle line in this graphic represents Hathaway's maximum likelihood prediction for the smoothed average sunspot number. The dotted lines demark the 95% confidence interval given by Hathaway's prediction scheme.
If the smoothed average sunspot number goes outside of the predicted confidence interval, we should have 95% confidence that Hathaway's prediction scheme is WRONG. (If the scheme was right, the smoothed average number would only split off from the confidence interval 5% of the time.)
Hathaway's 2006 95% confidence interval ranged about 23 sunspots per year above and below his maximum likelihood prediction. The actual sunspot rate is now about 75 sunspots/year below his March 2006 prediction. Thus we can have probably about 99.9% confidence at this point that Hathaway's prediction scheme is wrong.
This evidence of an errant prediction scheme ought to prompt revision of the scheme, but it is evident from the comparison above that instead of revising his prediction scheme, Hathaway is doubling down. Notice how Hathaway responds to what was for him an unexpected year and a half lull in solar activity. He actually shortens the time horizon within which he expects the next solar cycle to erupt:
In March 2006, Hathaway's maximum likelihood prediction had solar cycle 24 turning upwards 5 or 6 months later. Two months ago Hathaway was predicting up-turn in two months.
How can this possibly be in response to the latest solar behavior (a year and a half of unexpected quiet)? Simple answer: it isn't.
Prediction schemes are SUPPOSED to be weighted by their success
According to NASA's June 2008 explanation of Hathaway's prediction scheme, Hathaway and his colleagues base their prediction of how future solar cycles will progress on a composite of different prediction schemes. Some of the prediction schemes have a theoretical basis. Some are simply statistical projections of past patterns. All are weighted commensurate with past performance. To the extent that theoretical considerations seem to improve predictive ability, they are weighted accordingly in the composite prediction.
But all of that goes out the window when it comes to predicting the length of solar minimums. Here there is no weighting of different schemes according to their performance. Rather:
At this phase of cycle 23 [the solar minimum] we now give full weight to the curve-fitting technique of Hathaway, Wilson, and Reichmann Solar Physics 151, 177 (1994).That technique says that the shape of the down-phase of the now tailing off solar cycle 23 can be predicted by the shape of its upswing and peak (which were established by the year 2000).
That leaves no avenue for new information to enter the Hathaway/NASA prediction scheme. As NASA explains:
The two parameters for this fit (cycle amplitude and cycle starting time) have remained unchanged since early 1999.The failure of Hathaway's prediction scheme to predict the delayed onset of solar cycle 24 ought to undermine confidence in Hathaway's scheme and reduce the weight given to it in NASA's composite scheme, but this mechanism is not operative because Hathaway's prediction scheme is the ONLY scheme NASA uses for predicting the end of a solar cycle. No matter how poorly it predicts, it is still weighted 100%.
That makes no sense. We know that extended minimums do occur, and should be included as a possibility in any prediction scheme. In the past, when a solar cycle tailed off for longer than the sharpness and amplitude of its rise would have predicted, how long did it tend to stretch out, and on what pattern? The recent failures of Hathaway's prediction scheme should prompt a shift in attention to the past behavior of such statistical exceptions. Hathaway's latest prediction may well be right. Solar cycle 24 could fire up any day now. But it can't be the maximum likelihood projection at this point.
Hathaway failure to address Maunder Minimum type solar behavior is systematic
It isn't that Hathaway is trying to give precedence to theory-based prediction schemes over mere statistical projection. (Not that that would be valid in any case. Prediction schemes should be weighted by performance, period.) Hathaway is simply giving priority to his own purely statistical model. As he wrote in his 1999 paper:
Ideally, we would like to predict solar activity using a model of the sun's magnetic dynamo along with current and past observations to initialize that model. Unfortunately, both the model and many of the important observations do not exist at present. We recognize that solar magnetism is the key to understanding the processes involved. We believe that the sun's differential rotation, meridional circulation, and large-scale convective motions all play important roles in producing the cyclic magnetic behavior that we observe. We have not yet, however, produced a theory that fully incorporates these mechanisms in a model that provides any predictive power.This applies a fortiori to Hathaway's 1994 prediction scheme. It is all statistical projection. So why isn't Hathaway looking at the full range of statistical evidence? Why is he fixing instead on his own predictive scheme to the exclusion of any consideration of the exceptions to it?
Hathaway certainly knows that there are statistical patterns (past events) that don't fit the cycle shape he is projecting. William Briggs notes that, when Hathaway back-tests his own predictive scheme, he only looks at the sunspot record AFTER the Maunder Minimum (during which sunspot activity went into a 50 year lull). (For the date range Hathaway uses, see figure 1 here.)
Hathaway continues today in the same vein, specifically avoiding consideration of Maunder Minimum type solar behavior. According to NASA, Hathaway simply "believes" that "the quiet of 2008 is not the second coming of the Maunder Minimum."
Hathaway's predictive scheme should be weighed in NASA's composite prediction scheme according to its performance, as NASA does with the other parts of its composite. Cycle 23 is not matching the usual cycle shape, at which point a rational predictive scheme would start paying more attention to the statistical properties of past such anomalies. But Hathaway still gives 100% weight to his so-far non-performing prediction scheme.
The IPCC does the same thing
Hathaway's persistence in predicting a strong solar cycle 24 does not in itself suggest any agenda. If the sun resurges, it will warm the earth, allowing the global warming alarmists to continue to pretend that warming is being caused by fossil fuel burning, but a PREDICTION about what the sun will do does not expose or hide anything.
With one exception. By refusing to even consider the possibility of an extended solar lull, Hathaway himself can avoid having to address the impact that such a lull would have on global temperature. If he did address that question, he would obviously have to note that the last really long such lull seems to have caused the Little Ice Age, which would place him on the side of the "deniers" in the debate about human-caused global warming. If low solar activity caused the Little Ice Age, then the "grand maximum" levels of solar activity during the 20th century would be the cause of 20th century warming, and the hoax of human-caused warming would be exposed.
Looking at the science alone, however, Hathaway cannot be lumped with the IPCC. Most importantly, there actually is a solid body of evidence behind his predictive scheme. Past solar cycles HAVE tended to follow a particular pattern, where the shape of the up-phase predicts the shape of the down-phase.
In contrast, there is no evidence whatsoever for the theory of human-caused global warming. In theory, additional CO2 should have SOME heat trapping effect, but there is no sign of it in the geological record. What warming effect CO2 has is evidently too small to measure, and this is just what we would expect.
CO2 mostly traps the same wavelengths of infrared that the vastly more abundant water vapor, leaving very little heat trapping work for marginal increases in CO2 to perform. Thus theory itself says that the marginal effect of CO2 on temperature should be miniscule.
Where Hathaway doesn't look so good is in what he leaves out. This man's job is to study sunspot activity, and he is ignoring the Maunder Minimum???? But even here, he is not nearly as bad as the IPCC. Of course Hathaway should be highly focused at this point on the phenomenon of extended solar minima, but the truth is that we know almost nothing about these minima beyond that fact that there was a fifty year lull a couple hundred years ago.
In the case of the IPCC, the alternative prediction scheme that they consciously ignore is based on a very well defined and certain cause of global warming. The geologic record proves beyond any doubt whatsoever that global temperature has historically been driven almost entirely by the level of solar activity.
High levels of solar-magnetic activity shield the earth from galactic cosmic radiation (GCR). The level of GCR reaching earth can be measured from isotope signatures deposited in the geologic record, which also contains contemporaneous temperature signatures. Study after geologic study has found that the level of GCR "explains" statistically about 90% of global temperature variation on every time scale. (Fred Singer's book Unstoppable Global Warming: Every 1500 Years compiles the scientific evidence for laymen.)
We also have a good theory of the mechanism by which the level of GCR (and the sun's blocking of it) affects global temperature. (See Henrik Svensmark's book The Chilling Stars.) But having a working theory is just gravy. We don't need to know the mechanism in order to prove causality. A 90% correlation on all time scales over hundreds of thousands of years cannot be coincidence, and the causality can only go one way. It is not the temperature of the earth that is determining the level of solar activity, or the influx of GCR.
The IPCC's excuse for not taking solar magnetic effects into account is that the mechanism is not understood, but that is not a legitimate reason. The mechanism does not have to be understood for us to know that some such mechanism is at work, and how it will affect global temperature as solar activity changes. Any prediction that does not exploit the predictive value of that known physical relationship is unscientific.
Scientific fraud
The IPCC ought to be using the composite prediction scheme that NASA is nominally using for predicting sunspot activity (but not actually using when it comes to predicting how long a solar cycle will last). They should be exploiting the predictive value of ALL of the different available predictive mechanisms, whether they model the mechanisms behind known physical relationships, or just model the statistical relationships themselves.
In the case of global warming, MOST of the weight would go to the very well established causal relationship between GCR and global temperature, and NONE of it would go to CO2 effects for which there is NO evidence in the geologic record. (CO2 follows global temperature, as warming oceans release CO2, but there is no evidence that it has or can drive temperature in any significant way.)
Hathaway is tainted with a little of the same stain, because he too is fixating on what the evidence suggests is a failed prediction scheme to the exclusion of alternate possibilities that he obviously should be giving some weight to. He just doesn't go nearly as far in this direction as the IPCC, which gives 100% of their predictive weight to a theory for which there is NO evidence, while completely refusing to account or even acknowledge what we KNOW is the predominant driver of global temperature.
"Grand maximum" levels of solar activity during the 20th century easily explain the small amount of late 20th century warming, leaving little or nothing to be explained by alternative mechanisms like human burning of fossil fuels. The IPCC's particular fraud, for which there is no analog in Hathaway's work, springs from the coincidental correlation between solar activity and the human burning of fossil fuels over the 20th century.
When an explanatory variable is omitted from a statistical estimate, the explanatory power of the omitted variable will be misattributed to any correlated variables (however causally unrelated). The IPCC intentionally omits the solar-magnetic variable, knowing full well that its overwhelming explanatory power will be misattributed to the coincidentally correlated increase in atmospheric CO2. They then knowingly project this misattributed explanatory power forward in order to create a dishonest political attack on fossil fuel burning.
(I traced this ruse through the 4th IPCC draft report in the comments that I submitted for draft review.)
This scientific fraud by the IPCC is evil. In this context, it is disappointing that Hathaway is not eager to discuss the Maunder Minimum, and explain how every solar scientist knows about the well established correlation (and the implied causal relationship), between sunspot activity and global temperature. That he dodges this question by unscientifically failing to address Maunder Minimum type episodes in his analysis of our present solar lull does not speak well for his integrity.
UPDATE: Tim Ball also slams IPCC for omitting solar-magnetic mechanism, despite overwhelming evidence for it
The IPCC's excuse for omitting solar-magnetic effects from their prediction scheme is that the mechanims is not understood. Ball rejects this on the grounds that we actually have from Svensmark a very good theory:
...they studiously avoided any discussion of the clear relationship between sunspot activity and temperature. They claimed there was no mechanism to explain the correlation so it could not be included, but that is incorrect. A very valid mechanism known as the Cosmic Theory (Svensmark and Calder, “The Chilling Stars”) has been in the literature with increasing detail since 1991. The date is important because IPCC claimed it was excluded because it was not published in time to meet their cut off date for consideration.This is a very important deception to expose, but the scientific failure of the IPCC at this point actually goes much deeper. At the risk of sounding like a broken record, even if we did not understand the mechanism, the causal relationship between solar-magnetic activity and global temperature is fully proven by the close correlation between the two over hundreds of thousands of years.
The IPCC's position is like saying that, prior to Newton's and Einstein's theories of gravitation, the best prediction for what a rock would do when released into the air is that it would waft on the breeze. After all, we understand the mechanism of the air pushing on the rock, but we don't understand the mechanism of this invisible force that pulls a rock towards the earth, so it ought to be omitted from our predictions.
I discussed this last month in my post about the pair of frauds from the American Physical Society who claim that hundreds of thousands of years of close correlation between solar activity and global temperature does not imply causality. The hell it doesn't!
Other related posts
This post shows how the expected environmental impacts of CO2 burning are actually positive, and by a wide margin. Thus the proper micro-economic tax on fossil fuel burning, if there is to be one, would be negative (a subsidy), not the perverse high taxes and cap-and-trade systems that are expanding today.
My posts on the basic science of the global warming fraud here and here.
