Rog asked about the results of my peeking at the SST2 leading UAH TLT data where I replied it is unsafe but of interest. Well he did ask a leading question.
Figure 1
This needs some explanation.
I work with my own versions of time series extracted from published gridded datasets. With UAH this is a clone of published time series without the rounding. With some other datasets subject to “adjustments” instead of verbatim weighted average of gridded, with missing data handling there is sometimes difference, if not large. Hadley SST2 is subject to alternation by the publishers.
Figure 2
That is the monthly I am using. (ask if you want the data)
The first lesson is look.
Figure 3
Figure 3
Second lesson, passeth through the eye of a band. Not explaining about filtering here, I write my own software in C. In this case pretty brutal but could go far further. It will do.
Result is Figure 1.
Lagging UAH TLT by a couple of months, r2 = 0.9
Given this tightly filtered this is unlikely to be accepted as significant.
Doing fancy things with this produces very perplexing answers, such as the instantanous phase looks like the solar magnetic cycle but I cannot reasonably show it matches. Maybe with perspiration.
Any theories on what this might be if anything?
Additional comment, the longer period is the usual vagely 45 year wave. That’s in solar data too but not the usual data, long story.
Figure 4
Novel for solar data? Might post on this one day.
Please take the whole thing as maybe, any mistakes as is.
[Update, plot data, XLS (97/2000/XP) spreadsheet without plots sst2-uah-solar-2012-03-16 (2.8M) large because I have included solar data. ]
Posted by Tim Channon, co-moderator.
Tallbloke's Talkshop 
Thanks Tim, very interesting if even more cryptic than usual. 🙂
So Figure 4 is what exactly? A spectrum of what solar data?
Fig 1 still looks like there is an obvious lag, with tlt following along after sst.
Lots of other interest in fig 1 anyway. The ~3.7 year peak is obviously ENSO. It’s also 1/3 of the average solar cycle length. More on the ENSO – Solar cycle link in the next post.
First fig 3 has some interesting periodicities. Can you give us the periods of the peaks rather than us having to try to guess them from a log scale?
Good stuff, Tim, which corroborates Tallbloke’s graph on the ‘Arms’ post.
Tim says: “Any theories on what this might be if anything?”
Not theory, more wild speculation… 🙂
Sea water contains ~3.5% dissolved salts and is an electrolyte which can carry an electric current.
Click to access ocean_encycl.pdf
Electromagnetic ocean effects
“The oceans play a special role in electromagnetic induction, due to their relatively high conductivity and the dynamo effect of ocean currents. Typical crystalline rocks of the Earth’s crust have electrical conductivities in the range of 10^-6 to 10^-2 Siemens (1/Ω) per meter. In comparison, sea water, with 2.5-6 S/m, is a very good conductor. Electrical currents are induced in the oceans by two different effects: Induction by time varying external fields and induction by motion of the sea water through the Earth’s main field.”
As the sun is the main driver of Earth’s magnetic field via changing solar activity, perhaps changes in electrical ocean currents effect circulation rate/direction/overturning/strength of gyres?
Don’t think we have sufficient data about ocean currents to know what’s happening – still much to learn. Any thoughts?
So figure 4 shows that a wiggly blue thing varies my more than an order of magnitude compared to a set of other numbers?
Obvious, really.
Steve: fig 4 is what is known as ‘a teaser’ 🙂
Relax and get to know the style.
Cryptic, well yes it was done after dark, hence cryptonite.
Fig 4, That’s a woggle, unless it is female and they do good figures.
Where was I?
I’ve not shown numeric periods, partly it takes time and exact numbers don’t mean very much with what is likely to be noise or chaos.
A further problem is pulling out what are very long periods given the dataset lengths. I can do that better with the analyser software. With DFT the bin sizes are large and whilst interpolation can be done.
Easy answer for me, make up a spreadsheet (without plots) and you make of it what you will.
Doing that now. post it later.
Updated post with link to data in portable XLS format.
Thanks Tim, I’ll have a look.
It is clear that these two measurements are measuring near enough to the same thing. It is also clear that Solar cycles play very little part as the main period is 11.1 years with some side lobes, and that period is very weak in your spectra.
Try this.
‘It is also clear that [optical] Solar cycles play very little part as the main period is 11.1 years with some side lobes, and that period is very weak in your spectra.’
Figure 4 is solar data, not a lot of 11.1 year in there either. It is a measure of solar asymmetry which has more to do with magnetics.
Neither would I assume too much about sunspot cycle amplitude being dominant.
Tallbloke,
You are making the mistake that the statistic distribution of the frequencies present in the SOI are somehow fited by a broad gaussian with mean of 3.8. Nothing could be further from the truth. Please have a look at the second graph at:
http://astroclimateconnection.blogspot.com.au/2010/03/why-do-long-term-periodicities-in-enso.html
There two dominant peaks at 3.6 and 4.8 years giving an “average” 4.2 years, if that has any real world physical meaning.
I believe that you are seeing a 3.8 year signal not a 3.7. The 3.8 year signal that tchannon has plotted in his first figure is almost perfectly in phase with the number of days
spring tides (i.e. New/Full Moon) are from a given specific date e.g. Perihelion. In other words, what you are seeing in figure 1 is a Lunar tidal signal.
Thanks Ian, I’ll bow to your greater knowledge. I’ve been labouring under the misapprehension that the long term ‘average ENSO period’ is around 3.7 years.
Your comment is very interesting. In physical terms is this implying there is some sort of interaction between the annual variation in insolation due to orbital eccentricity and the Lunar tidal timing? It doesn’t seem obvious that such an interaction would have an influence on the occurrence of ENSO events. But then, it’s a complicated phenomenon so I’ve probably misunderstood something else too.
Thanks for your comment in ‘suggestions’ I’ll take a look tomorrow morning.
It is neither, it is a doublet (and a little more), hence the amplitude variation, beating.
This does not show up in LoD or at least not that I recall when I was looking very closely. If that is the case it points more to a solar origin or no connection to anything much.
If it is lunar, what is the modulation math?
Possibly some kind of harmonic train exists but I’ve only briefly looked, too noisy to say much.
Too late now to dig out some more figures.
The link Ninderthana gives, needs discussion.
Strictly there is no such thing as a sub-harmonic, a subject in it’s own right, oddities which if they exist go under other names.
Chandler wobble? No and I have a tale to tell. A key to that is near 6.4y, the relationship of 1, 1.19, 6.4, where unfortunately 6.4 does not match the lunar nor the dominant gravitational, wish it did so it is a loose end. (there are suspected dataset problems, unresolved issue)
Matches of that kind must be exact, about orbitals where noise is low.
I’m off the shut eye, I hope.
Tallbloke,
[N.B. the 3.8 year lunar spring tidal cycle is a sub-multiple of the 19 year Metonic cycle since 5 x 3.8 = 19.0 years]
I am also humbled by my ignorance on these matters. However, I think what it means is that alignment of the lunar spring tidal cycles with the seasonal (Annual) solar cycle every 3.8 years, somehow influences one of the following [remember that there is little seasonal change in the tropics/equator]: :
a) the retention or release of cool deep ocean water.
b) the rate of cloud formation (Is there some sort of seasonal cosmic-ray variation?)
c) the rate at which tropical/equatorial atmospheric heat is redistributed towards the poles.
d) the small variation in solar insolation due to ellipticity of the Earth’s orbit may play a role.
Some have argued that the Lunar tidal effects are only a modulating influence on some deeper Solar-Terrestrial connection – they might be correct.
Another more esoteric explanation is the fact that:
a) If a spring tide (New/Full Moon) occurs on a given day of the year,3.796 tropical years will pass before another spring tide occurs on the same day of the year.
b) If a lunar node aligns with the Sun on a given day of the year, 6.410 years will pass before another lunar node aligns with the Sun on roughly the same day of the year.
(0.5 draconic months)/(13.5 draconic months – 1.0 tropical year)
= (13.606110 days/2.122791 days) = 6.410 years.
c) In order to have a lunar node realign with the Sun on the same day of the year, AND for the Moon to return to the same phase (e.g. New/Full Moon = Spring tides) as well, it would take a period of time set by the beat period between 3.796 and 6.410 years i.e. 9.308 years.
The 6.410 year period for the Lunar nodes to align with the Sun on a given day of the year is remarkably close to the realignment time for the three largest terrestrial planets:
4 x SVE = 6.3946 years where SVE = synodic period of Venus and Earth
3 x SEM = 6.4059 years SEM = synodic period of Earth and Mars
7 x SVM = 6.3995 years and SVM = synodic period of Venus and Mars
The line of nodes of the lunar orbit appears to rotate around the Earth, with respect to
the Sun, once every Draconitic Year (TD = 346.620 075 883 days). This means that the
Earth experiences a transition from a maximum to a minimum in the meridional and
zonal components of the tidal stress (or vice versa), at times separated by:
¼ TD = 86.65002 days 1st tidal harmonic
5 x ¼ TD = 1 ¼ TD = 433.275095 days = 1.18622 years 2nd tidal harmonic
5 x 1 ¼ TD = 6 ¼ TD = 2166.375474 days = 5.93111 years 3rd tidal harmonic
The first point that needs to be made about this is that there appears to be an almost
perfect synchronization between the three tidal harmonic intervals and submultiples of
the sidereal orbital period of Jupiter (TJ = 4332.82 days = 11.8624 sidereal years):
1/50 TJ = 86.6564 days
1/10 TJ = 433.282 days
1/2 TJ = 5.93120 years
tchannon,
I think that you are first graph shows the 3.8 year period is modulated by the 4.43 yrs (= 1/2 x 8.86 yrs = 1/2 x the time for the lunar line-of-apse to realign with the Sun).
(4.43 x 3.8) / (4.43 – 3.8) = 26.72 years
which is close to the long-term modulation that you are seeing in the sea surface temperatures.
Please ignore the errors in my first post above
tchannon,
I think that you are first graph shows the 3.8 year beating with the 4.43 yrs period = 1/2 x 8.86 yrs = 1/2 x the time for the lunar line-of-apse to realign with the Sun.
(4.43 x 3.8) / (4.43 – 3.8) = 26.72 years
which is close to the long-term modulation that you are seeing in the sea surface temperatures.
Last night as I was attempting to stop nodding (nod off) it came to me there _is_ a curiosity which might show a commonality between the Chandler wobble and the shake here. (new craze there, Nod, Wobble and Shake by the Jellies), both have a node at 2005/6.
I now think it is worthwhole to do a long post on the Chandler wobble where some surprises are in store, with the intent of raising it’s profile in the mind of readers. I also like to pour scorn on NASA, who claim to know the cause.
Looking in more detail at the extracted waves the two datasets do not match as well as I had hoped.
The analyser here can trivially split doublets.
A centre period is 3.63 and 3.73 years but the modulation is too different.
Not that simple because there are other terms in there.
So broadly there is similarity but the match is not as good as it appears.
That said it would be truly remarkable if there was a close match.
tchannon,
Have you looked at the paper by N. Sidorenkov that connects lunar tides, ENS0 and the Chandler wobble?
http://www.springerlink.com/content/f043rtx683755172/
Chandler wobble of the poles as part of the nutation of the atmosphere-ocean-Earth system
N. S. Sidorenkov
Abstract
The paper presents calculated spectra of El Niño Southern oscillation (ENSO) indices. The ENSO spectra have components with periods that are multiples of the Earth’s free (1.2 years) and forced (18.6 years) nutation periods. Analysis of a 41-year series of exciting functions for the atmospheric angular momentum confirms the existence of such periodicity. Nutation waves responsible for the El Niño phenomena in the ocean, the Southern oscillation in the atmosphere, and the presence of subharmonics of the Chandler period (1.2 years) and superharmonics of the lunar period (18.6 years) in the ENSO spectra are described. A model for the nonlinear nutation of the Earth-ocean-atmosphere system is constructed. In this model, the ENSO, acting at frequencies of combinational resonances, excites the Chandler wobble of the Earth’s poles. At the same time, this wobble interacts with the nutation motions of the atmosphere and World Ocean.
__________
Translated from Astronomicheski Zhurnal, Vol. 77, No. 6, 2000, pp. 474–480.
Original Russian Text Copyright © 2000 by Sidorenkov.
tchannon,
The 6.40 year re-alignment period for the three largest terrestrial planets modulates the
1.0 year orbital period of the Earth (with respect to the stars) to give 1.1852 years.
1.1852 (sidereal) years = 432.9 days
This is very close to the normally accepted period of the Chandler wobble of 433 days.
Also as noted above:
If a lunar node aligns with the Sun on a given day of the year, 6.410 years will pass before another lunar node aligns with the Sun on roughly the same day of the year.
(0.5 draconic months)/(13.5 draconic months – 1.0 tropical year)
= (13.606110 days/2.122791 days) = 6.410 years.
Do you think that this is just a coincidence?
If there is interest I can post what I know so far and hope discussion can resolve the remaining problems.
It would be useful to rummage over any thoughts that you have on this data Tim.
Rummage, a lovely word, but will it come out ravished?
Had some thoughts so I am having another go at figuring out the “something” which is odd about the cw data. (maybe it ought to be ccw ;))
I have a feeling that like say sea ice data we see a unipolar effect from a bipolar stimulation, except this doesn’t have a straight baseline. (I suspect something similar with ssn but in that case is non-linear and falls into the noise)
That will make no sense.