New paper by Love et al suggests no prominent role for solar‐terrestrial interaction in global climate change. I’m providing it here for discussion.
We are not convinced that the combination of sunspot‐number,
geomagnetic‐activity, and global‐temperature data can, with
a purely phenomenological correlational analysis, be used to
identify an anthropogenic affect on climate.
Abstract
Recent studies have led to speculation that solar‐terrestrial interaction, measured by sunspot number and geomagnetic activity, has played an important role in global temperature change over the past century or so. We treat this possibility as an hypothesis for testing. We examine the statistical significance of cross‐correlations between sunspot number, geomagnetic activity, and global surface temperature for the years 1868–2008, solar cycles 11–23. The data contain substantial autocorrelation and non-stationarity, properties that are incompatible with standard measures of cross-correlational significance, but which can be largely removed by averaging over solar cycles and first‐difference detrending. Treated data show an expected statistically significant correlation between sunspot number and geomagnetic activity, Pearson ρ < 10^−4, but correlations between global temperature and sunspot number (geomagnetic activity) are not significant, ρ = 0.9954, (ρ = 0.8171). In other words, straightforward analysis does not support widely‐cited suggestions that these data record a prominent role for solar‐terrestrial interaction in global climate change.
With respect to the sunspot‐number, geomagnetic‐activity, and global‐temperature data, three alternative hypotheses remain difficult to reject: (1) the role of solar‐terrestrial interaction in recent climate change is contained wholly in long‐term trends and not in any shorter‐term secular variation, or, (2) an anthropogenic signal is hiding correlation between solar‐terrestrial variables and global temperature, or, (3) the null hypothesis, recent climate change has not been influenced by solar‐terrestrial interaction.
Citation: Love, J. J., K. Mursula, V. C. Tsai, and D. M. Perkins (2011), Are secular correlations between sunspots, geomagnetic activity, and global temperature significant?, Geophys. Res. Lett., 38, L21703, doi:10.1029/2011GL049380.
Conclusions
One of the merits of using three separate data sets in a correlational analysis is that intercomparisons can be made. After treatment for removal of autocorrelation and nonstationarity through simple averaging and differencing, we find statistically‐significant secular correlation between sunspot number and geomagnetic activity. This is expected,
and it serves as important support for our analysis method. On the other hand, after making the same treatment to the global surface temperature, correlations between temperature and either sunspot number or geomagnetic activity are not significant.
We have not, in this study, considered derived proxy metrics of relevance to climate change, such as reconstructed total‐solar irradiance [e.g., Fröhlich and Lean, 2004] or
interplanetary magnetic field [e.g., Lockwood et al., 1999]. Still, we believe that our methods are general, that they could be used for other data sets, even though our analysis, here, is tightly focused on specific data sets. [15] From analysis of sunspot‐number, geomagneticactivity, and global‐temperature data, three hypotheses remain difficult to reject; we list them.
(1) The role of solarterrestrial interaction in recent climate change is wholly contained in the long‐term trends we removed in order to reduce autocorrelation and nonstationarity. This possibility seems artificial, but we acknowledge that our method requires a nontrivial time‐dependence in the data that is different from a simple trend. Still needed is a method for measuring the significance of correlation between data sets with trends.
(2) An anthropogenic signal is hiding correlation between solar‐terrestrial variables and global temperature. A phenomenological correlational analysis, such as that used here, is not effective for testing hypotheses when the data record a superposition of different signals. Physics is required to separate their sum.
(3) Recent climate change has not been influenced by solar‐terrestrial interaction. If this null hypothesis is to be confidently rejected, it will require data and/or methods that are different from those used here.
Paper: http://www.leif.org/EOS/2011GL049380.pdf
h/t to Dr. Leif Svalgaard
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
M.A.Vukcevic says:
November 19, 2011 at 7:50 am
since the HMF data (green line) is Svalgaard-Cliver property (available also in their paper).
There should be no such ‘property’ claim. Everything we do is visible and up-front and belongs to the world. Sharing data and views with colleagues and critics is the best way to pursue science (and have them help to pinpoint flaws and errors before publication).
Re Paul Vaughan, or throwing cold water on flaming.
Simple cross-correlation analysis (including the multiscale time-integrated variety) has a role during preliminary exploration, but it should then be immediately & patently obvious to any sufficiently cognizant explorer that it alone canNOT finish the job because of the nonstationarity. Italics and caps in original, Paul Vaughan, 11/18/11, 6:54 am.
But Vaughan contradicts himself in the first sentence next:
Multiscale complex correlation (for example using adjacent derivative based complex empirical wavelet embeddings) can measure complex nonstationary relations where simple linear correlation fails catastrophically. Naïve investigators unknowingly encounter Simpson’s Paradox by falsely assuming independence and blindly running linear factor analyses (such as PCA, EOF, & SSA) without performing the right diagnostics. Bold added, Paul Vaughan, WUWT, 5/15/11, here.
In a couple of his guest posts and a big bunch of his comments here and on other blogs, Vaughan uses Simpson’s Paradox, (above) and especially the phrase the spatiotemporal version of Simpson’s Paradox, repeatedly. In not one case did he attempt to justify his application of this rudimentary statistical error of mixing things that shouldn’t be mixed. It’s cook-book statistics, and not worth calling a paradox.
Simpson’s Paradox has captured Vaughan’s imagination, so he uses it to salt through his writings as if it lent weight and merit to them. It does not. Everything is not a nail. He is a techno-name dropper. He thinks he is elevating himself by liberally accusing others of being naïve, not understanding his buzz words like spatiotemporal harmonics and integration across harmonics, or his wallpaper of pointless graphs. For a nice summary of Vaughan’s graphic arts, be sure to read the post by oldgamer56, 5/15/11, 8:18 pm, here.
What Vaughan knows is patently true, patently observable, or patently obvious. But what others write is patently false, patently untrue, patently unrelated, patently untenable, patently absurd. Others suffer from blind failures, blindly running, prejudices [that] blind, too blind to judge, and Blind innocents, including experienced & well educated ones, can easily fall into such traps unknowingly.
To the last, Vaughan he appends his only online credentials: Paul Vaughan, B.Sc. (biology/math-stats), M.Sc. (applied stats); Ecologist, Former Stats Instructor. 12/30/10, 12:51 pm, WUWT, here, and Perhaps I should clarify that my perspective is that of an ecologist. Paul Vaughan, 5/5/10, 3:40 pm, here.
Neither a statistician nor a physicist, he claims to be an ecologist, one with no training in the field, much less any kind of science. He’s an ecologist in the sense that everyone’s a philosopher, everyone in here with a crayon is a writer.
In his citation at 6:54 am, above, he twice exposes his lack of science literacy. He proclaims, italicized for emphasis, that cross correlation must come first in the scientific method. Even the bogus versions of the scientific method aren’t step-by-step recipes.
Secondly, in his contradiction about the application of correlation to nonstationary processes, he talks about the nonstationarity (bold added) as if it were a real world constraint instead of a matter of investigator’s choice in what he can fit to real world data. To be fair, he confesses his disconnect to the real world: As a contribution to this multidisciplinary discussion, I explore data, leaving physics to physicists & other qualified parties. Paul Vaughan, 4/11/11, 8:10 am, here.
And on this thread Mr. Vaughan appears in full flower. Imagine diagramming this sentence:
Multimoment multiscale spatiotemporal integration reveals nonrandom harmonic pattern-summary discontinuities, exposing the comedy tragically advocated by deceitful &/or naïve theoreticians who are in part constrained by a dominant culture that clings seemingly religiously to maladaptive traditions such as unjustifiable assumptions of randomness, independence, uniformity, linearity, etc. that are routinely misapplied (for example to conveniently render abstract conceptions mathematically tractable). Paul Vaughan, WUWT, 4/10/11, id.
On Tallbloke’s blog, Vaughan says,
The physics is known, but almost no one in the climate discussion has a clue about the topology of the sampling framework and its conceptual implications. The spatiotemporal version of Simpson’s Paradox is effectively holding the discussion hostage. As Judith Curry pointed out in one of her best pieces to date, such ignorance plays into the hands of those who need the perception & persuasion of so-called (but fundamentally mislabeled) “uncertainty”:
http://judithcurry.com/2011/08/22/can-we-make-good-decisions-under-ignorance/
With Absolute Sincerity,
Paul L. Vaughan, M.Sc.. Bold in original, 9/10/11, 7:47 pm, here.
Vaughan’s citation to Climate Etc. supports ignorance alright, but nothing about the spatiotemporal version of Simpson’s Paradox, or ignorance holding [anything] hostage.
Like Capt. (“just ask me specific questions”) Queeg, it’s all too much for Mr. Vaughan to explain:
In a nutshell, complex correlation involves matrices of correlations of multiscale adjacent derivatives of series (with each cell having both real & complex parts). Elaboration would be grossly impractical via this forum & medium. Most of the climate literature only subtly touches on the tip of the iceberg of what is possible — not entirely sure why, but it’s plain to see that the result has been catastrophic derailment by [the spatiotemporal version of] Simpson’s Paradox. Paul Vaughan, 3/13/11, 3:02 am, here.
I literally don’t have time to dig through the long, messy WUWT & Climate Audit threads on this subject… . Paul Vaughan, 9/11/11, 10:27 pm, here.
This is a misinterpretation. (Sorry, I haven’t time to elaborate … ). Paul Vaughan, 8/11/11, 2:00 pm, here.
Then there’s this:
Those of you talking about null hypotheses & randomness have to realize that statistical inference is MEANINGLESS prior to working out the nature of the complex conditioning (due to Simpson’s Paradox). Caps in original, Paul Vaughan, 2/5/11, 10:55 pm, here.
Does any of Paul L. Vaughan’s writings have substance?
Your comments about differencing, binning, & filtering are neither sensible nor well-founded. Your suggestion about PDFs is blindly naive (the kind of ignorance-based uncertainty promotion one expects from Climate Etc.). The climate pie is big enough that we can each have a piece of it. I can suggest that you focus on gradients rather than averages when brainstorming physical models. EOP (Earth Orientation Parameters) inform us clearly about SIMPLE climate asymmetries. Leave the data exploration to capable parties. Paul Vaughan, 11/11/18, 8:34 pm.
Paul, for failing to provide anything material to this vacant, self-aggrandizing diatribe, go to your room. Bad Paul; no pie.
Additional note:
It is highly unlikely that the temperature oscillations as shown here:
http://www.vukcevic.talktalk.net/HMF-T.htm
are caused by the change in the HMF-B, which I think is too weak to have such large effect.
Dr. Svalgaard may explain if he likes why HMF-B shows no Hale cycle, but it can be discerned from the
S-C paper.
That the particular temperature data set shows strong 22 year period, is not that odd if you are follower either of the Cosmic Rays, or even my hypothesis which has strong elements of the Solar- Earth magnetic field interactions since both have distinct magnetic polarity.
M.A.Vukcevic says:
November 19, 2011 at 8:09 am
Dr. Svalgaard may explain if he likes why HMF-B shows no Hale cycle
It does not, because the power spectrum says it does not: http://www.leif.org/FFT-HMF-B.png for the years where we have good data [1835-present]
but it can be discerned from the S-C paper.
The discerning eye sees what it want to see [e.g. canals on Mars, 22-yr cycle in HMF B] something being there or not.
follower either of the Cosmic Rays, or even my hypothesis which has strong elements of the Solar- Earth magnetic field interactions since both have distinct magnetic polarity
The cosmic ray 22-yr cycle is extremely weak and does not really show a 22-yr signal of intensity, but just of the shape of the 11-yr cycle. There is a weak 22-yr cycle in geomagnetic activity, that goes from sunspot maximum to maximum [every 2nd solar minimum has slightly higher geomagnetic activity], but this is also 2nd order effect.
Leif Svalgaard says: November 19, 2011 at 8:45 am
……..
You are not ready to let go, so here it is what one can discern from your paper
Leif Svalgaard, ETK, Inc., Houston, Texas, USA
&
Edward W. Cliver, Space Vehicles Directorate, Air Force Research Laboratory, Hanscom Air Force Base, Bedford, Massachusetts, USA
The IHV-index springs from the same source as the classical u-measure [Bartels, 1932], building on concepts by Broun [1861] and Moos [1910] who defined the interdiurnal variability U of the horizontal component at a given station as the difference between the mean values for that day and for the preceding day taken without regard to sign.
Your FFT analyser isn’t much good compared to the one I have, do compare resolutions of two (e.g. number of points between 20 and 40 years).
http://www.vukcevic.talktalk.net/Spectrum.gif
I have work to do now.
M.A.Vukcevic says:
November 19, 2011 at 9:36 am
The IHV-index springs from the same source as the classical u-measure
You are confusing the IHV index with the IDV index [the latter is the one that we use to infer HMF B]. As I pointed out, there is a weak 22-yr variation in IHV [and aa, ap, etc], because of the interplay between two effects [Russell-McPherron effect and Rosenberg-Coleman effect]. This have nothing to do with the Sun, but with the position of the Earth. The theory can be seen here: http://www.leif.org/research/suipr699.pdf section 9.
Your FFT analyser isn’t much good compared to the one I have, do compare resolutions of two (e.g. number of points between 20 and 40 years).
http://www.vukcevic.talktalk.net/Spectrum.gif
They look pretty much alike to me, including not showing any significant 22-year cycle.
I might read the article, there is always something new to learn or to use ‘in a way not known’.
You need to take a good close look at the two spectra.
Your shows 8 or 9 year component (due to the lack of resolution) while on mine (if you magnify x 2) there are spikes at 8, 9 and strong 10 &12, which is more in line with the various SSN lengths, while the rest is mostly noise. Also power at and above twice the fundamental is exaggerated in your spectrum, I assume a consequence of much wider bandwidth. Better accuracy does matter.
M.A.Vukcevic says:
November 19, 2011 at 11:50 am
I might read the article, there is always something new to learn or to use ‘in a way not known’.
Without reading the article you are sticking your head in the sand.
You need to take a good close look at the two spectra.
(due to the lack of resolution) while on mine)
You cannot improve the resolution beyond what the data provides.
Information is there but your program is too poor to capture it, and one questions the accuracy e.g. 14 &17 year periods.
http://www.vukcevic.talktalk.net/Spectrum2.gif
You need better and more accurate analyser.
M.A.Vukcevic says:
November 19, 2011 at 2:24 pm
and one questions the accuracy e.g. 14 &17 year periods.
They and almost all the peaks are noise anyway, so no need to worry about them.
You need better and more accurate analyser.
You cannot go beyond what the data has. Torturing the data more, does not lead to better insight. Only better data. If the data really has a significant signal it will be in the simple FFT.
Example:
Here I show the FFT of HMF measured by spacecraft [left] and inferred from IDV: http://www.leif.org/research/FFT-HMF-B-Monthly.png
There is a true peak at 11 year [black oval) in both plots. Almost all the other ones are ‘noise’ in the sense that there is that variation in the messy data, but they are not true periods that will repeat ‘forever’. There is, however, a true period hiding in the noise, at exactly 1 year (red oval). This one is stationary and is guaranteed to be there 1000 years from now. Why? because it is caused by the varying distance to the Sun that makes B 5% larger at perihelion. Note that the FFT is good enough to pick up this very narrow peak. On the right, you can also see the red and black oval peaks, and many of the ‘noise’ peaks are also present in both (After all, it is the same reality we are looking at). However, there is a strong and narrow peak (blue oval) that is only seen in the IDV-derived HMF. This is a ‘true’ peak and the FFT is perfectly suited for picking it up [and it will also be there a 1000 years from now]. Why? because it is caused by the varying angle between the Earth’s dipole axis and the solar wind direction [the equinoctial semiannual variation of geomagnetic activity [see e.g. http://www.leif.org/research/Semiannual-Comment.pdf ]. It is an artifact in the sense that HMF B will not have it as the HMF does not know what the Earth is doing [and sure enough, there is no blue oval peak on the left plot]. We publish yearly values for IDV partly to remove this artifact. But the point is that ‘my’ FFT program [DPLOT] can see it. So, real, true, stationary periods come through just fine, and those are the only ones of interest. Noise that comes and goes might peak here and there, but who cares?
M.A.Vukcevic says:
November 19, 2011 at 2:24 pm
You need better and more accurate analyser.
Here is a file for you to play with: http://www.leif.org/research/Vuk-B.xls
It is slightly different [on purpose to show that the result is not sensitive to the exact interval] from my monthly file. It gives B for 27-day Bartels rotations for the years 1883-2010 derived from IDV, So its power spectrum should have the 11-year peak, the 1-yr peak, and the semi-annual peak. See if you can find them.
@Jeff Glassman (November 19, 2011 at 8:07 am) [ http://wattsupwiththat.com/2011/11/15/are-secular-correlations-between-sunspots-geomagnetic-activity-and-global-temperature-significant/#comment-801822 ]
I suggest redirecting your focus & energy towards understanding nature.
Paul Vaughan says:
November 19, 2011 at 7:20 pm
@Jeff Glassman (November 19, 2011 at 8:07 am)
I suggest redirecting your focus & energy towards understanding nature.
Jeff is quite correct in his assessment. Your statement is not a valid or substantive response to Jeff’s comment.
Doc
Don’t know what your game is, but I may have good idea.
There is 6 months sharp peak as expected, and not much else until you aproach 11 years, except small bumb at 72 bartels.
http://www.vukcevic.talktalk.net/Spectrum3.gif
If your ‘machine’ can’t get to speed you need a better one.
M.A.Vukcevic says:
November 20, 2011 at 1:48 am
There is 6 months sharp peak as expected, and not much else until you aproach 11 years, except small bumb at 72 bartels.
Why start in 1900 and not 1883? Or is the graph just mislabeled?
Data file may be too long, prog got stuck.
If the above results are of interest (i.e. an improvement on what you got, else there is no point) I can try again or do 1835-1900, but if your data is Ok it shouldn’t matter, unless you expect that the sun has change rot rate.
M.A.Vukcevic says:
November 20, 2011 at 4:32 am
Data file may be too long, prog got stuck.
Better machine conked out 🙂
Its OK, ye olde FFT still works.
‘Not so fast my friend….!’
It was your ‘silly’ data file which had two sets of data on top of each other.
Of course it worked OK when I separated them.
http://www.vukcevic.talktalk.net/Spectrum4.gif
It barely shows 1 year period. It appears 1 year was prominent in the pre 1900 section, but disappeared into noise after 1900.
I am expecting an explanation, since I went into lot of trouble to show you what superior spectrum analyser comes up with.
M.A.Vukcevic says:
November 20, 2011 at 10:25 am
I am expecting an explanation, since I went into lot of trouble to show you what superior spectrum analyser comes up with.
Since there is only data on a 27-day resolution from 1883 on, and the data before 1883 was yearly means you cannot look for a 1-yr period during 1835-1883.
“”””” Paul Vaughan says:
November 18, 2011 at 11:22 am
@george E. Smith (November 18, 2011 at 10:46 am)
Aliased data is particularly informative if the nature of the aliasing is understood. “””””
I’m even prepared to take your word for that statement Paul. Now it remains for you to explain just HOW one is to understand the “nature of the aliassing”.
Given that ALL you have is the aliassed data; from whence comes the extra information; since it ISN’T in the data record ?
Leif Svalgaard says:
………………
So for all purposes 1900-2011 is OK, or do you want 1883-2011 tested?
George E. Smith; says:
………………
Since I spent many years working on TV studio productions, I can only say that ‘aliasing’ is an absolute pain in the backside.
M.A.Vukcevic says:
November 20, 2011 at 12:18 pm
So for all purposes 1900-2011 is OK, or do you want 1883-2011 tested?
I expect both to be OK. Although the noise level is expected to higher for the pre-1900 data [fewer stations contribute]
Leif Svalgaard
Scafetta’s New Paper Linking Mid-Latitude Aurora to the 60 Year Temperature Cycle.
“We argue that the aurora records reveal
a physical link between climate change and astronomical oscillations.”
“In particular, a quasi-60-year large cycle is quite evident since 1650 in all climate and
astronomical records herein studied, which also include a historical record
of meteorite fall in China from 619 to 1943. These findings support the
thesis that climate oscillations have an astronomical origin.”
“The existence of a natural 60-year cyclical modulation of the global surface temperature
induced by astronomical mechanisms, by alone, would imply that at least
60–70% of the warming observed since 1970 has been naturally induced.
Moreover, the climate may stay approximately stable during the next decades
because the 60-year cycle has entered in its cooling phase.”
Dr. Scafetta Says: “In the paper I argue that the record of this kind of aurora can be considered a proxy for the electric properties of the atmosphere which then influence the cloud cover and the albedo and, consequently, causes similar cycles in the surface temperature.”
http://landscheidt.wordpress.com/2011/11/10/scafettas-new-paper-linking-mid-latitude-aurora-to-the-60-year-temperature-cycle/