By Dr. Roy Spencer, PhD (reprinted from his blog with permission)
UPDATE (12:35 p.m. CDT 19 May 2011): revised corrections of CERES data for El Nino/La Nina effects.
While I have been skeptical of Svensmark’s cosmic ray theory up until now, it looks like the evidence is becoming too strong for me to ignore. The following results will surely be controversial, and the reader should remember that what follows is not peer reviewed, and is only a preliminary estimate.
I’ve made calculations based upon satellite observations of how the global radiative energy balance has varied over the last 10 years (between Solar Max and Solar Min) as a result of variations in cosmic ray activity. The results suggest that the total (direct + indirect) solar forcing is at least 3.5 times stronger than that due to changing solar irradiance alone.
If this is anywhere close to being correct, it supports the claim that the sun has a much larger potential role (and therefore humans a smaller role) in climate change than what the “scientific consensus” states.
BACKGROUND
The single most frequently asked question I get after I give my talks is, “Why didn’t you mention the sun?” I usually answer that I’m skeptical of the “cosmic ray gun” theory of cloud changes controlling climate. But I point out that Svensmark’s theory of natural cloud variations causing climate change is actually pretty close to what I preach — only the mechanism causing the cloud change is different.
Then, I found last year’s paper by Laken et al. which was especially interesting since it showed satellite-observed cloud changes following changes in cosmic ray activity. Even though the ISCCP satellite data they used are not exactly state of the art, the study was limited to the mid-latitudes, and the time scales involved were days rather than years, the results gave compelling quantitative evidence of a cosmic ray effect on cloud cover.
With the rapid-fire stream of publications and reports now coming out on the subject, I decided to go back and spend some time analyzing ground-based galactic cosmic ray (GCR) data to see whether there is a connection between GCR variations and variations in the global radiative energy balance between absorbed sunlight and emitted infrared energy, taken from the NASA CERES radiative budget instruments on the Terra satellite, available since March 2000.
After all, that is ultimately what we are interested in: How do various forcings affect the radiative energy budget of the Earth? The results, I must admit, are enough for me to now place at least one foot solidly in the cosmic ray theory camp.
THE DATA
The nice thing about using CERES Earth radiative budget data is that we can get a quantitative estimate in Watts per sq. meter for the radiative forcing due to cosmic ray changes. This is the language the climate modelers speak, since these radiative forcings (externally imposed global energy imbalances) can be used to help calculate global temperature changes in the ocean & atmosphere based upon simple energy conservation. They can then also be compared to the estimates of forcing from increasing carbon dioxide, currently the most fashionable cause of climate change.
From the global radiative budget measurements we also get to see if there is a change in high clouds (inferred from the outgoing infrared measurements) as well as low clouds (inferred from reflected shortwave [visible sunlight] measurements) associated with cosmic ray activity.
I will use only the ground-based cosmic ray data from Moscow, since it is the first station I found which includes a complete monthly archive for the same period we have global radiative energy budget data from CERES (March 2000 through June 2010). I’m sure there are other stations, too…all of this is preliminary anyway. Me sifting through the myriad solar-terrestrial datasets is just as confusing to me as most of you sifting through the various climate datasets that I’m reasonably comfortable with.
THE RESULTS
The following plot (black curve) shows the monthly GCR data from Moscow for this period, as well as a detrended version with 1-2-1 averaging (red curve) to match the smoothing I will use in the CERES measurements to reduce noise.
Detrending the data isolates the month-to-month and year-to-year variability as the signal to match, since trends (or a lack of trends) in the global radiative budget data can be caused by a combination of many things. (Linear trends are worthless for statistically inferring cause-and-effect; but getting a match between wiggles in two datasets is much less likely to be due to random chance.)
The monthly cosmic ray data at Moscow will be compared to global monthly anomalies the NASA Terra satellite CERES (SSF 2.5 dataset) radiative flux data,
which shows the variations in global average reflected sunlight (SW), emitted infrared (LW), and Net (which is the estimated imbalances in total absorbed energy by the climate system, after adjustment for variations in total solar irradiance, TSI). Note I have plotted the variations in the negative of Net, which is approximately equal to variations in (LW+SW)
Then, since the primary source of variability in the CERES data is associated with El Nino and La Nina (ENSO) activity, I subtracted out an estimate of the average ENSO influence using running regressions between running 5-month averages of the Multivariate ENSO Index (MEI) and the CERES fluxes. I used the MEI index along with those regression coefficients in each month to correct the CERES fluxes 4 months later, since that time lag had the strongest correlation.
Finally, I performed regressions at various leads and lags between the GCR time series and the LW, SW, and -Net radiative flux time series, the results of which are shown next.
The yearly average relationships noted in the previous plot come from this relationship in the reflected solar (SW) data,
while the -Net flux (Net is absorbed solar minus emitted infrared, corrected for the change in solar irradiance during the period) results look like this:
It is that last plot that gives us the final estimate of how a change in cosmic ray flux at Moscow is related to changes in Earth’s radiative energy balance.
SUMMARY
What the above three plots show is that for a 1,000 count increase in GCR activity as measured at Moscow (which is somewhat less than the increase between Solar Max and Solar Min), there appears to be:
(1) an increase in reflected sunlight (SW) of 0.64 Watts per sq. meter, probably mostly due to an increase in low cloud cover;
(2) virtually no change in emitted infrared (LW) of +0.02 Watts per sq. meter;
(3) a Net (reflected sunlight plus emitted infrared) effect of 0.55 Watts per sq. meter loss in radiant energy by the global climate system.
WHAT DOES THIS MEAN FOR CLIMATE CHANGE?
Assuming these signatures are anywhere close to being real, what do they mean quantitatively in terms of the potential effect of cosmic ray activity on climate?
Well, just like any other forcing, a resulting temperature change depends not only upon the size of the forcing, but also the sensitivity of the climate system to forcing. But we CAN compare the cosmic ray forcing to OTHER “known” forcings, which could have a huge influence on our understanding of the role of humans in climate change.
For example, if warming observed in the last century is (say) 50% natural and 50% anthropogenic, then this implies the climate system is only one-half as sensitive to our greenhouse gas emissions (or aerosol pollution) than if the warming was 100% anthropogenic in origin (which is pretty close to what we are told the supposed “scientific consensus” is).
First, let’s compare the cosmic ray forcing to the change in total solar irradiance (TSI) during 2000-2010. The orange curve in following plot is the change in direct solar (TSI) forcing between 2000 and 2010, which with the help of Danny Braswell’s analytical skills I backed out from the CERES Net, LW, and SW data. It is the only kind of solar forcing the IPCC (apparently) believes exists, and it is quite weak:
Also shown is the estimated cosmic ray forcing resulting from the month-to-month changes in the original Moscow cosmic ray time series, computed by multiplying those monthly changes by 0.55 Watts per sq. meter per 1,000 cosmic ray counts change.
Finally, I fitted the trend lines to get an estimate of the relative magnitudes of these two sources of forcing: the cosmic ray (indirect) forcing is about 2.8 times that of the solar irradiance (direct) forcing. This means the total (direct + indirect) solar forcing on climate associated with the solar cycle could be 3.8 times that most mainstream climate scientists believe.
One obvious question this begs is whether the lack of recent warming, since about 2004 for the 0-700 meter layer of the ocean, is due to the cosmic ray effect on cloud cover canceling out the warming from increasing carbon dioxide.
If the situation really was that simple (which I doubt it is), this would mean that with Solar Max rapidly approaching, warming should resume in the coming months. Of course, other natural cycles could be in play (my favorite is the Pacific Decadal oscillation), so predicting what will happen next is (in my view) more of an exercise in faith than in science.
In the bigger picture, this is just one more piece of evidence that the IPCC scientists should be investigating, one which suggests a much larger role for Mother Nature in climate change than the IPCC has been willing to admit. And, again I emphasize, the greater the role of Nature in causing past climate change, the smaller the role humans must have had, which could then have a profound impact on future projections of human-caused global warming.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
John Finn says:
May 21, 2011 at 1:12 am
It doesn’t look to be the dominant driver otherwise temperatures would much lower than they currently are. UAH temperatures during the recent La Nina have been as high as those during the 1986/87 El Nino.
Still banging the same old drum John?
You are not considering how the climatic effect of the Svensmark Hypothesis works. By reducing cloud cover during the high solar cycles of the later C20th, more sunlight gets absorbed into the ocean. This warms the ocean. A couple of decades later, the run of big el ninos starts, creating the step changes in surface temp so neatly demonstrated by Bob Tisdale.
Given the obviously large thermal inertia of the ocean, why would you expect the temps to drop as soon as the sun goes quiet? If it took a couple of decades after the high solar cycle of the late 1950’s for the warming to get going, it’ll probably take a couple of decades after the end of the run of high solar cycles before the cooling ocean finishes burping out accumulated heat and lowers surface temps too.
Stephen Wilde wrote (May 20, 2011 at 11:46 pm):
“Where those air masses interact we see more clouds and the solar effect seems to work by causing more (or less) meridional jets, more (or less) air mass mixing and therefore longer (or shorter) lines of air mass interaction across the globe resulting in more (or less) clouds.”
Can Bill Illis, Dr. Spencer, or someone else suggest a time series that quantifies this length-to-surface-area ratio or fractal dimension that Stephen often writes about?
tallbloke says:
May 21, 2011 at 6:55 am
So as far as I can tell, he is taking account of the absolute values, but modelling them by multiplying up the detrended data using the factor he determined.
This is where it is wrong. Why should the forcing be calculated using the detrended data?
Stephen Wilde wrote (May 20, 2011 at 11:46 pm):
“To get that change in meridionality/zonality we first need a change in the atmospheric heights and as far as I know Svensmark’s idea does not deal with that.”
I suggest that someone (colleague or journalist conducting interview to be publicized) ask Svensmark directly about this.
—
Where Stephen’s exposition can improve dramatically:
Emphasize that the variation is semi-annual. The decadal-scale changes are in the amplitude of the semi-annual wave. See Leroux (1993) for some helpful pictures and Sidorenkov (2005) for words that fit on a single page to go with Leroux’s pictures. North-south terrestrial asymmetry canNOT be dismissed as irrelevant.
Leif Svalgaard says:
May 21, 2011 at 7:41 am
Why should the forcing be calculated using the detrended data?
Because climate is complicated and the Earth is a noisy laboratory.
Roy Spencer says:
Detrending the data isolates the month-to-month and year-to-year variability as the signal to match, since trends (or a lack of trends) in the global radiative budget data can be caused by a combination of many things. (Linear trends are worthless for statistically inferring cause-and-effect; but getting a match between wiggles in two datasets is much less likely to be due to random chance.)
tallbloke wrote (May 21, 2011 at 7:18 am):
“Given the obviously large thermal inertia of the ocean, why would you expect the temps to drop as soon as the sun goes quiet? If it took a couple of decades after the high solar cycle of the late 1950′s for the warming to get going, it’ll probably take a couple of decades after the end of the run of high solar cycles before the cooling ocean finishes burping out accumulated heat and lowers surface temps too.”
The dynamics occur at the semi-annual timescale. There’s no multidecadal lag. The multidecadal variations come from north-south terrestrial asymmetry.
Tallbloke says:
May 21, 2011 at 8:07 am
Because climate is complicated and the Earth is a noisy laboratory.
Whole generally true, the claim is that the solar [GCR] influence is the MAJOR [SOLE – ONE AND ONLY] driver in which case the noise is the signal. Now if you concede that the solar [GCR] influence is minor, hard-to-detect, almost drowning in the noise, you might use the ‘noisy laboratory’ excuse.
In any event, one might expect that Spencer did a correct analysis. However, trying to reproduce his graph shows that the wiggles do not match, especially after 2006 [where the amplitudes of his wiggles are too large – leading to artificially enhanced forcing]. Here is following his recipe: http://www.leif.org/research/Moscow-2000-2011.png.
This is also directly visible by eye-balling his graphs.
The mechanisms of ice decline are fully explained each and every yearly cycle by atmospheric and oceanic circulation pattern variations. Therefore, a warmer’s only recourse is to then explain how the increasing anthropogenic CO2, and only just that portion, has had enough energy potential to change the atmospheric and oceanic circulation pattern variations that have led to the recent decline.
I have certainly not yet seen a mathematical proof of this anthropogenic CO2-atmospheric/oceanic pattern connection mechanism, nor have I seen a plausible mechanism without a maths proof.
Gates, you are stating nothing more than speculative conjecture.
Leif Svalgaard wrote (May 21, 2011 at 7:41 am)
“Why should the forcing be calculated using the detrended data?”
tallbloke replied (May 21, 2011 at 8:07 am)
“Because climate is complicated and the Earth is a noisy laboratory.
Roy Spencer says: Detrending the data isolates the month-to-month and year-to-year variability as the signal to match, since trends (or a lack of trends) in the global radiative budget data can be caused by a combination of many things. (Linear trends are worthless for statistically inferring cause-and-effect; but getting a match between wiggles in two datasets is much less likely to be due to random chance.)”
It’ll become more interesting when Dr. Spencer finds time to step beyond the linear approach (or maybe money to pay a grad student or postdoc to do it). Cross-correlation, while informative, is patently insufficient for unearthing the full nature of complex [as in complex numbers, not as in complicated] relations.
This one sentence;
“While I have been skeptical of Svensmark’s cosmic ray theory up until now, it looks like the evidence is becoming too strong for me to ignore.”
tells me more about Dr Spencer’s honesty and integrity than any other source. Is it not the statement of a true scientist ?
Leif Svalgaard says:
May 21, 2011 at 8:28 am
Tallbloke says:
May 21, 2011 at 8:07 am
Because climate is complicated and the Earth is a noisy laboratory.
Whole generally true, the claim is that the solar [GCR] influence is the MAJOR [SOLE – ONE AND ONLY] driver in which case the noise is the signal.
You are assuming wiggles are only caused by drivers. What about rebounding oscillations within the terrestrial system?
Paul Vaughan says:
May 21, 2011 at 8:30 am
It’ll become more interesting when Dr. Spencer finds time to step beyond the linear approach (or maybe money to pay a grad student or postdoc to do it). Cross-correlation, while informative, is patently insufficient for unearthing the full nature of complex [as in complex numbers, not as in complicated] relations.
Roy Spencer says:
May 20, 2011 at 5:25 AM
No one really knows what causes natural cycles in the climate system like the PDO. It could be natural oscillations in the thermohaline circulation of the ocean. Your question is a little like asking, “what causes chaos?”. There are nonlinearities in the climate system on a wide variety of timescales that are too complex for us to predict, let alone understand.
And as he drily noted in response to a critic on his blog:
Roy Spencer says:
May 13, 2011 at 5:26 AM
So, when IPCC-related studies assume linear feedbacks, it’s ok, but when I do, it’s not OK? Hmmm.
Paul Vaughan says:
May 21, 2011 at 8:30 am
Cross-correlation, while informative, is patently insufficient for unearthing the full nature of complex [as in complex numbers, not as in complicated] relations.
But is often patently sufficient for deluding statisticians into seeing things that aren’t there.
Leif Svalgaard wrote (May 20, 2011 at 10:28 pm):
“If I have some criticism then it would be that the cosmic ray data should not be detrended as Svensmark’s hypothesis works with the actual count of the particles. Twice as many, gives twice as many ions, etc.”
The atmosphere is FLUID. If something external DRUMS on it, it changes spatially in SHAPE (not in absolute size). If one thrusts a rod of semi-infinite length into a pool, the moment of greatest change for the pool is the initial contact of the blunt finite end.
It doesn’t matter if Svensmark is right or wrong or if it’s GCR or something confounded with GCR or if one correlated station is better than another; what matters is the coherence THAT IS OBSERVED. Suggestion: Physicists can best help out via efforts to explain what is observed.
Schwing, F.B.; Jiang, J.; & Mendelssohn, R. (2003). Coherency of multi-scale abrupt changes between the NAO, NPI, and PDO. Geophysical Research Letters 30(7), 1406. doi:10.1029/2002GL016535.
tallbloke says:
May 21, 2011 at 8:37 am
You are assuming wiggles are only caused by drivers. What about rebounding oscillations within the terrestrial system?
Spencer is assuming that [‘forcings’]. But, what about your oscillations? But a number on them, a theory, and a mechanism, and come back for a discussion about the science.
Paul Vaughan says:
May 21, 2011 at 8:10 am
The dynamics occur at the semi-annual timescale. There’s no multidecadal lag. The multidecadal variations come from north-south terrestrial asymmetry.
And that asymmetry (and LOD changes on the multi-decadal scale) causes the oceans to shift large amounts of energy around. And up and down. Leading to longish lags between solar input and climate response.
tallbloke addressing Svalgaard (May 21, 2011 at 8:37 am):
“You are assuming wiggles are only caused by drivers. What about rebounding oscillations within the terrestrial system?”
Excellent question. The level of discussion has elevated.
Paul Vaughan says:
May 21, 2011 at 8:53 am
It doesn’t matter if Svensmark is right or wrong or if it’s GCR or something confounded with GCR or if one correlated station is better than another; what matters is the coherence THAT IS OBSERVED.
Coherency of multi-scale abrupt changes between the NAO, NPI, and PDO.
Everybody claiming a correlation [no matter which] claims it is OBSERVED. In any event your examples are just climate vs. climate. No external forcings. So, not relevant for the topic at hand.
Paul Vaughan says:
May 21, 2011 at 9:01 am
“What about rebounding oscillations within the terrestrial system?”
Excellent question. The level of discussion has elevated.
The wiggles were taken as evidence for external direct driven forcing, so the level has dropped considerably.
tallbloke, there’s no lag looking at solar cycle deceleration (instead of TSI or sunspot numbers or whatever else someone might have in mind). Excessive reliance on “anomalies” has blinded some to higher frequency oscillations. The ocean can lose heat in 3 months. It doesn’t need decades. The highest variance is OVERWHELMINGLY at high frequencies; there is no escaping this observation.
You are getting there chaps.
Solar forcings from above (external) modulated over time by oceanic forcings (internal) from below.
The outcome of that struggle at any given moment being reflected in the surface pressure distribution (especially the position and behaviour of the midlatitude jets) which has an effect on the size and position of the various climate zones.
The regions that see most in the way of climate changes are those situated at or near a climate zone boundary so that as the surface pressure changes vary those regions cross from one side to another of those boundaries for variable lengths of time.
Paul Vaughan says:
May 21, 2011 at 9:08 am
tallbloke, there’s no lag
Paul, I think you missing an important element of the lags. Invoking unknown [even variable] lags of several decades are an efficient shield against falsification.
Leif Svalgaard says:
May 21, 2011 at 8:55 am
tallbloke says:
May 21, 2011 at 8:37 am
You are assuming wiggles are only caused by drivers. What about rebounding oscillations within the terrestrial system?
Spencer is assuming that [‘forcings’]. But, what about your oscillations? But a number on them, a theory, and a mechanism, and come back for a discussion about the science.
See the quote from Roy above about chaos. The fact that he has managed to tease out some signal from the noise is great, and note that he states it is a preliminary finding, and the accuracy is not well constrained due exactly to the noise problems we are discussing.
It’s still progress though, and at least he’s willing to work on the problem because he can no longer ignore the evidence. Unlike some who are still in denial.
tallbloke says:
May 21, 2011 at 9:16 am
and the accuracy is not well constrained due exactly to the noise problems we are discussing.
Even more to his deficient analysis. What he plots is not what he says he plotted as I just showed.
It’s still progress though, and at least he’s willing to work on the problem because he can no longer ignore the evidence. Unlike some who are still in denial.
Unlike some who have looked at this in detail [not preliminary and not with deficient data analysis] and found it wanting and the evidence unconvincing.
Paul Vaughan says:
May 21, 2011 at 9:08 am
tallbloke, there’s no lag looking at solar cycle deceleration (instead of TSI or sunspot numbers or whatever else someone might have in mind). Excessive reliance on “anomalies” has blinded some to higher frequency oscillations. The ocean can lose heat in 3 months. It doesn’t need decades. The highest variance is OVERWHELMINGLY at high frequencies; there is no escaping this observation.
I’m not sure what you mean by solar cycle deceleration. Cycle lengths getting longer? If so, Archibald posted about a scandinavian whose study showed an approx one cycle lag between solar activity change and surface temp change.
“The ocean can lose heat in 3 months. It doesn’t need decades. ”
The top 30m of the ocean can lose heat in 3 months, but that is a small proportion of the thermal mass of the ocean. The shifting of the courses of large undersea rivers of water mixes down energy due to the coriolis effect and tidal action. The stored energy can be held or released over long timescales, and the variation is much greater than the higher frequency annual variation. The shallow ocean was several degrees warmer all the way to the bottom some millions of years ago.