Gravity solar z-axis, a casual look

Posted: November 19, 2012 by tchannon in Analysis, Astrophysics, Cycles, Gravity, solar system dynamics

Figure 1

[UPDATE] I have made a mistake although I had included caution, something looked wrong. On looking closely at the old file which is the source for this analysis it is for Earth, not the Sun. The basics will be similar. See comments, we are going to carry on because what has appeared in intriguing. In the process a rework on other versions of data will be much quicker and easier : Tim]

Recently on the Talkshop a discussion has started about Sol and gravity forces. I stepped up in case I can help. I think this is where the recent discussions commenced.

Story goes that some time ago I went down the route of attempting a computation but the result didn’t seem interesting in relation to whatever I was doing so I dropped the work.

Somewhat boring plot. Devil in the detail?


We don’t know whether z-axis is a key to solar activity patterns where it is suggested that analysis ought to be aligned to the solar rotation axis which is not the standard reference frame.

Whilst that is being resolved, here is work done aligned to the standard plane. There is not a large difference in alignment so it ought to be similar.

Revisiting the GB of files in the solar archive I came across what looks like the final result, so large it brings my computer to it’s knees, knobbly they are. (runs out of RAM, more about inefficient programs than actual need) Sorry folks, not putting up this data, too large.

This has 10 day data running apparently from a Thursday, tomorrow is TGIF, yet what did folks do 1st January 1400?

Here is the start date Julian 2232407.5
Have fun at

Dataset ends 2030.

I don’t remember the details but it looks as though I computed the gravitational force on the sun in XYZ format based on the standard astronomical plane. What the current discussions want is data on the solar plane. Never mind lets look at what is.

What follows might follow cockup theory so please not too hard, doing my best.

Figure 1 has been decimated to 1 year for plotting. (for pedants, low pass at just over 2 years and is end corrected by my own method)

Let’s take a look at the periodic structure. What follows is using the full data as input, 23,000 samples of floats.


Figure 2

My quick look (defaults) and this looks wrong. Is an octave z-chirp transform from 0.5y through 500y, windowed kaiser-bessel beta=3, logarithmic Y-axis

Assuming few readers are particularly familiar windowing is important, stops fictitious results which is to do with the input data having a finite length and has to be blocked off as best we can, which is a compromise.

The width of the lines (peaks) varies because unavoidable the time and hence period resolution approaches zero at the right hand side, limited by data set length. It becomes vague. I have a way of working around this, more on that later.

We can see the main planets Jupiter, Saturn, Uranus and Neptune. (Pluto is tiny and omitted from the data) The problem is the large amount of fast hash, <strong>that ought not to be there.</strong>

So far as I can tell I have done nothing wrong, everything is done to considerable precision and at good time resolution.

The solar system ought to be a linear system and therefore without significant interaction between planets, yet this implies there is enormous other things going on, or more likely something is wrong with the data or my processing.

I recognise clues in the spectra that there is more hidden and so


Figure 3

Same as figure 2 but using a Blackman Harris -92dB window (probably a nice demonstration of the difference) which has good vertical resolution but worse frequency resolution. (previous is the best at separating close spaced frequencies)

We can now see the previously missing ~19.8y dominant gravitational peak, at least it is in the normal orbital plane.

I still don’t believe this is real.

What I can do is extract items using other software. A slow by hand process. This gives exact phase and amplitude. Of course these might be random terms.

Any suggestions anyone? Might this be noise or artefacts from the data originating in integrated polynomial approximations, ephemeris? Am I looking too deeply into a junk layer?

I’m going to publish now and might update later with some details.


I’ve successfully run the analyser software on the 10 day data, simple enough but needing various tricks to get useful results. This is not about cheating.

A direct attack on the longer period primary problem would be twarted by the fast noise which is not of particular interest at the moment. I could force things by hand but an easier route is play tricks: low pass the data at 2 years, removing the troublesome content and use that as input data. This is undone later.

I chose 13 terms as reasonable in this case but is probably more than is wise. Apart from a few nudges to speed things up (this is a large dataset) it was just a matter of waiting. (a human can see slow progress to an obvious answer, give it a kick to get on with it, next stage, self corrects anyway)

I then took a deep breath and reverted to the original data as input, hoping it would simply accept instead of spotting the change and jumping to the faster data. This worked and minor refinement took place, probably so tiny it doesn’t matter.

Warning on what follows: there is an irresolvable technical problem: I cannot plot the full bandwidth data without showing ridiculously long plots, otherwise would alias, therefore I have to use @2y low pass filtered data.


Figure 4

Figure 4 visually quantifies the dominance of the first three terms of factors in a Fourier decomposition.

period 11.86 29.45 5.93
frequency 0.08 0.03 0.17
phase 3.04 4.6 6.14
amplitude 863 137.4 62.43

I don’t know what units came out of the original work so I can’t give them.

The plot uses a common y-axis scale and shows the original data filtered @2y for display (in blue) overlaying the 3 term model (in red). In addition one subtracted from the other is plotted (in green), the residual.

A detail mentioned here for clarity but also an aside of interest, at the very ends of the plot you can see the error caused by the 2 year low pass filter end correction being imperfect. This error is in the display version of the data, not the final result. (can’t meaningfully plot 22,000 points) Whilst an aside this is visual proof of the nonsense widely believed in science under everyone-knows about traversal filter length and used as an excuse for using moving average (shortest possible filter) yet at the same time ignoring the major artefacts added to the data by such a poor filter. Filter here was 585 long, about 16 years of data. It is the characteristic which actually matters and has no relationship to length.


Figure 5

Figure 5 is the magnified residual shown in figure 4 with 2 years knocked off each end.


Figure 6

Figure 6 is figure 3 with the result of removing the slower dominant terms. Cancellation is actually much greater than it appears, set more by the shortcomings of the DFT producing this plot than actual. Note it is 20dB (10x) per scale division.

It might be the case there are more factors within each discrete window but I have only cancelled one. Proving this would take a lot of detailed work, a waste of effort.



Figure 7

Figure 7 is the residual (same scaling as fig 6) with the entire model subtracted. Keep in mind that fast data below 2 years is excluded.

In match terms R-squared is not particularly apt, RMSD is more useful but not meaningful outside of a specific context. For the whole model excluding @2y r2=0.9999 and for the full data with high frequency r2=0.96 (size 20k, contains two unexpanded spreadsheets, ods and xls, plus instructions as .txt, should be portable)

Anything else needed?

  1. Sparks says:

    I recently plotted some planetary orbits manually and it has taken 3 days to plot 3 planets over 400 years, I finished Neptune and Uranus and I’m only half way through Jupiter. It’s along the same lines of trying to understand planetary influence on the sun, what I can see from the plot so-far is that Geoff Sharp seems to be on the right track.

  2. tchannon says:

    Do you feel able to tell us more? (looks across at lots of paper, pencil)

  3. Ray Tomes says:

    The z-axis discussions go back much further than the reference that you gave. Rog Tallbloke had an earlier article showing similarity of solar motion in z-axis direction to Earth LOD (length of day) and temperature. He communicated with me about this because I had suggested a mechanism whereby the solar z-axis motion should drive solar cycles.

    I first put forward this idea about 20 years ago. My argument goes as follows:

    1. Under GR the acceleration of horizontal light by gravity is twice as much as that of matter. This was predicted by Einstein and detected by Eddington and everyone since during eclipses of Sun. Vertical light has no doubling factor. We can’t even talk about acceleration of vertical light so we need a new definition which looks at rate of change of momentum.

    2. The centre of the Sun has much energy in e/m wave form in it. Therefore to that extent it also experiences twice the normal acceleration. So if 1 part in 10^5 of solar core is e/m waves then it experiences 5/3 (average of 2 and 2 and 3 for 1 spacial dimensions factor for light acceleration) of 10^-5 of normal acceleration.

    3. Initially I thought that this would move solar core in plane of planets but that rotation would mean only spectral components near solar rotation would build up. But I had ignored the tilt of the Sun. It is about 7 degrees so that over 10% of the effect is in the z direction and it does not get cancelled by solar rotation.

    4. The effect of the planets on the sun is to induce a toroidal circulation, even in the core, which reverses with planets motions from time to time. The major periods relevant are the 4 gas giant periods as well as the sums and differences of those periods and half these also.. If you make a list of these, you find a close match to many sunspot cycle periods.

    5. When I compared to the z-axis motion with the ssn periods I found that only periods near to 10.5 years are strong, with ones far from there very weak. The profile of the ratios indicates a classic resonance situation. The Sun has a natural resonance of 10.5 years (as Rog mentioned the accurate 10.49 year period now resolved in full sunspot record is right on this value) and so only z-component periods near that have much effect.

    6. Using only that assumption (10.5 year resonance) I was able to predict solar cycles over the full period of records with r=0.66 from planetary forces.

    7. I believe that this will explain things like maunder minimum, when planets hit resonance nose on and bring cycle to a pause.

    Further evidence for this hypothesis is given by the case of the earth. I know that outer planets have a 1.11 million year energy exchange (J and N). This same period is found in earth magnetic reversals quite clearly.

  4. tchannon says:

    Yes the subject goes back much further, within the thread on an article about a different subjects, starts about there. I guess the above post is not significant so it is not worth trying to link in the historic threads, maybe on a later more serious article.

    I have now completed an analysis which provides phase and amplitude of the more dominant longer components. Updating the article with this will take some time.

    One of the interesting features is the ~6y and finding this for real when I know that 6y and 3y are fairly distinctive in some terrestrial data is progress, but without a direct mechanism. In this case phase is no help.

    Dominant in the above, no surprise, is Jupiter and Saturn.

    The “noise” peak just above 1 year is novel and I suppose exciting if it is real. ~1.18y is the Chandler wobble. In the past I have looked hard for an extraterrestrial stimulation always drawing a blank. Since 1y and ~6y intermodulate to ~1.18y I think I need to do more on this but geocentric. I don’t buy the NASA handwave about ocean floor flapping, not given what else has come to light.

    As usual the whole exercise is fraught with Nyquist risks.

  5. tallbloke says:

    Tim, thanks for your effort here. I agree with Ray that to tease out the z-axis effect you need to calculate the planetary motion relative to the solar equatorial plane, rather than to the plane of invariance (which is what I think you’ve done?)

    If you look at Ray’s spectral analysis of the Z-xis components, there are some big differences due to the different referents.

    If the Chandler wobble is an interaction of J/2 and Earth’s 1 yr orbit, then there is something else going on apart from gravity. Resonant frequencies? E/M effect? Paul Vaughan may want to weigh in here too.

  6. tallbloke says:

    Here’s what I get using the JPL Horizons system

  7. tallbloke says:

    If you then smooth the z-axis data over two jupiter orbits to smooth out the J-S and J swings (possibly emulating the damped response of the motion due to the action of planetary gravity on the solar core relative to solar surface as per Ray’s theory), and smooth the sunspot record over the length of the solar cycle, and offset 45 years (inner planet’s return period), you get an interesting match, apart from an anomaly at the Dalton Minimum, and at ~1970 and now. These epochs coincide with the ‘hiccups’ in the radial distance of the Sun from the barycentre, emphasized with black ticks near the bottom axis in the plot below. These also coincide with Geoff Sharp’s ‘AM disturbances’.

    However, this is apparently unphysical, because the 45 year offset is putting the sunspot activity before the planetary motion. I think what were getting is an indication that we’re on the right track, but there’s something else involved. Maybe Ray could tell us more about his method for adding in a solar dydnamo oscillation at 10.5 years to get his r=0.66 correlation to SSN.

  8. Greg Goodman says:

    TC: “We don’t know whether z-axis is a key to solar activity patterns where it is suggested that analysis ought to be aligned to the solar rotation axis which is not the standard reference frame.”

    I think the key point is whether there is any precession of the solar axis over the period is question. If not, I agree the form should be similar.

  9. tallbloke says:

    Greg, so far as I’m aware, there is no rapid precession of the solar axis. Thinking it over, I came to the conclusion that an axial tilt cycle would be expected over the periodicity of the solar system’s progress above and below the galactic plane as it orbits the galactic centre. precession might be tied to the precession of Jupiter’s orbital eccentricity, but this would be a small component of the overall tilt of the solar axis.

  10. Greg Goodman says:

    TC: “We can see the main planets Jupiter, Saturn, Uranus and Neptune. (Pluto is tiny and omitted from the data) The problem is the large amount of fast hash, [strong]that ought not to be there.[/strong]”

    I did some quick calculations on mass/orbit^2 some time back and seem to recall Venus became a significant player, like 1/5 of Jupiter from memory. Odd you don’t see it.

    I have not used z-chirp practically, so I don’t know really what they should look like, but this spectrum seems more like real, noisy data than a pure model. Maybe a lot of that greck is calculation rounding errors. Looking at the appendix in the Solex 11 manual the errors are noisy and also have large orderly swings. I could imagine that throwing quite a bit of noise into the spectrum.

    Also, what version was this run on and what options were used ? There seem to be quite a few options that trade off speed/accuracy. Solex 11 apparently allows higher levels of calculation to improve accuracy.

    Since you do not seem too sure exactly how you created this data or what frame it was run on, it may be does not merit much head scratching about why it is the way it is.

    The plot seems good otherwise, perhaps it’s worth doing a shorter test run will all options for best accuracy and noting exactly what is done.

    sum( mass/sqrt(x^2+y^2+z^2)) is easy enough to do if everything is in a heliocentric inertial frame, which seems to be possible with Solex.

  11. Greg Goodman says:

    TB, the two major deviations in the plots you posted are separated by about 155year (by eye).

    The longest peak in the two plots Tim started this with looks very close to that.

    Variations in AM will relate to gravitation accelerations. I see these two observations as confirmations of my initial thought, the need to look at gravity more closely.

  12. Greg Goodman says:

    PS , if model or the calc errors cause shifts over time shorter series will have more consistent freq spectra.

  13. tallbloke says:

    Greg, more like 179 years (on average) IMO.

    Gravity is proportional to mass, radius and orbital velocity, so I would expect a regular relationship between a CoG calculation and CoM calculation. Still interesting to see though, so I look forward to the numbers being crunched. I still think this needs to be done relative to the solar equatorial plane, as that is the real world situation.

  14. Ray Tomes say :

    “4. The effect of the planets on the sun is to induce a toroidal circulation, even in the core, which reverses with planets motions from time to time. ”

    I remind you .
    Malaga (Tim Cullen) interesting comments

    We don’t comparate only axis “z” with “SSN” but “x,y,z” contemporary.
    Can we analyze mathematically and graphically ?
    I think … It’s very difficult !

  15. tallbloke says:

    Salut Michele:
    My comment at 9.00am shows a plot which has both the z-axis displacement and the x,y,z radial distance. But I agree it is difficult to represent this motion graphically.

  16. tchannon says:

    About to update the article.

    Uploaded separately is an archive containing minimal model (unexpanded), two different format spreadsheets and instructions are inside] grand size of 20k

  17. tchannon says:

    Article has been updated.

  18. Greg Goodman says:

    TB, you’re using the term “proportional to ” _very_ loosely. I think you mean “related to”.

    Since the proportional contribution of each planet will be substantially different, it may resemble at some times and not at all at others. It certainly won’t be the same overall shape, which seems to be what you are suggesting.

    If one was just going to be a scaled version of the other, it would be a pointless exercise.

    Yes, ideally this wants to be seen projected onto the solar z-axis not the normal to the equatorial plane. But if solar axis is not in precession on the the time-scales considered , 7 degrees will not change it much.

    The first thing it to see whether Solex will produce data without the “noise” or decide that all that HF signal is a proper, expected part of the gravitational variations.

    Bearing in mind rotational speed of V and E that is not impossible.

  19. Greg Goodman says:

    quick calc of relative contributions : m/r^2 , in Earth mass and AU


    Eccentricity and inclination should stop it being too boringly dominated by the jovian year.

    So don’t expect a trivial ” regular relationship” to previous plots.

  20. tallbloke says:

    Greg, I hope you are right. I’m looking forward to seeing the results.

  21. Greg Goodman says:

    TC: Anything else needed?

    Hi, ran up the spreadsheet model and got something looking like fig 1 on the face of it.

    What is fig 5? This actually looks like climate / SSN type variaitons that I have been looking at.

    Residual of what?


  22. tchannon says:

    “as well as a true wobble of the solar rotational axis”

    Seen 7.23 degrees mentioned before but this seems to mention the second rotation needed.

  23. tchannon says:

    Greg, wondered if anyone would pick that up.

    We have the data output by Solex.

    Now take just the Jupiter, Saturn and 5.9y model data, everything else switched off.

    Point by point subtract the latter from the former.

    You don’t have the filtered time series used but this will give a very similar result since the SS is working.

    Leave the start date as is (midday 1st Jan 1400)

    Set fs to 1, annual.

    Set cells C19, C20 and C21 C19, D19, E19 to 0 [grrr…. ]

    Plot cols A and B

    Upper is what you would see,

    Darn WordPress filetypes… put a single column of the filtered data used inside a zip
    Runs 1st Jan 1400 annually.

    if you reverse the enables and subtract the result from the file in the zip it will recreate Figure 5… I hope.

    If that is okay, other dates and output rate will do. Given the nature of this data extrapolating a little is pretty safe.

  24. J Martin says:

    Ray Tomes Said;

    6. Using only that assumption (10.5 year resonance) I was able to predict solar cycles over the full period of records with r=0.66 from planetary forces.

    7. I believe that this will explain things like maunder minimum, when planets hit resonance nose on and bring cycle to a pause.

    Further evidence for this hypothesis is given by the case of the earth. I know that outer planets have a 1.11 million year energy exchange (J and N). This same period is found in earth magnetic reversals quite clearly.

    6. Can you dig out a graph for that ?
    7. 1.11 million years. Where are we now on that cycle ? Another graph would be great.

  25. tchannon says:

    Not having had time to look myself, surprise, too close for chance,

    29.45/14.73 = 2

    Wondered what on earth the 14.73 might be, almost exact 2nd harmonic of Saturn orbital, 99.93% which is in line with typical accuracy on this kind of thing. Realised that the second I ploted the two together, classic 2nd.

    Even harmonics are asymmetry, very common in natural systems. Odd harmonics less so, perhaps more human related. (flattening of both peaks)

  26. tchannon says:


    11.83/5.93 = 100.03% of 2nd harmonic of Jupiter but is a smaller effect.

    They are the major ones which I was not expecting, thought it would all be linear. Once you have non-linearity the whole gamut of interaction is present.

    I have good reason to suppose there are doublets present, would be, but little will be showing at this amplitude in an analysis. (doublet is two closely spaced frequencies which are related, eg. in Lunar data the doublets are related by the 18.6y lunar cycle, maths agrees with measurement)

    Why is there non-linearity?

  27. Greg Goodman says:

    TC: Once you have non-linearity the whole gamut of interaction is present. …Why is there non-linearity?

    1/r^2 is non-linear ! Once you have that many bodies mutually interfering with each other in a non linear way I’m not surprised the there a lot of apparently “spurious” peaks at the higher frequency end.

    It is interesting how significant the S components are, in view of it’s net effect at solar centre. This must be due to it’s proximity to J which is the dominant individual force by a long way. S makes itself felt by perturbing the orbit of J.

  28. Greg Goodman says:

    TC: Figure 4 visually quantifies the dominance of the first three terms of factors in a Fourier decomposition.

    No, the third peak is a very well defined one around 1.25 . It is at least as big a the 29.45 peak. Don’t be fooled into thinking it is not important because it’s very “thin” . This simply means it is very well defined.

    Any ideas about where that one comes from?

    I can see four clear harmonics of the J period in figure 2. Also a slightly twisted peak (likely an asymmetric doublet) just under 10y. This will be reflected symmetrically the other side of the main J peak and is getting obscured by the 14.73y peak you noted.

    This is probably slightly distorting the apparent max of the 14.73 peak.

  29. Greg Goodman says:

    One final point, which is probably important in looking at overall effects of gravity, is the bunching just above 1.1 years.

    Here the DFT gives good resolution but this should not fool anyone into ignoring the ammount if energy in this part of the spectrum. It is significant in relation to the broader, less well defined peaks at longer periods.

    This region is probably dominated by the minor planets and some harmonics and other resonances from higher up.

    Don’t forget Mercury. It’s only small but it’s in real close that makes gives it more pull than Neptune !!

    period 0.240 846 a

    m/r2=0.055/ 0.387/ 0.387

    Run the DFT plot down to 0.2 years, I think you’ll find a surprise. The 1.25 peak I noted above may be an interaction of Earth and Mercury.

    Running the same DFT on Svalgaards TSI recon may be interesting.

  30. tallbloke says:

    Hi Greg,
    “I can see four clear harmonics of the J period in figure 2. Also a slightly twisted peak (likely an asymmetric doublet) just under 10y”

    This is the half period of the J-S synodic period: 9.93 years.

    It represent a big swing in the changing barycentric radius as Jupiter and Saturn move from conjunction to opposition. It also shows up as one of the three main peaks near the solar cycle length in a Max Entropy Method spectrographic analysis of the sunspot numbers.

  31. tallbloke says:

    Tim and Greg. Fig 5 is indeed interesting! The dominant term is around 13.8 years, and this is related to Jupiter and Uranus’ synodic period. (As you’d expect given that it’s the next biggest planet after J and S have been accounted for by Tim’s model)


    If Tim could include this term in his model, the residual will probably then show a dominant period of ~12.8 years which is Jupiter and Neptune’s synodic period, or 19.86 years, the J-S synodic. Looking at the amplitudes in fig6, I’ll go with Neptune.



    In general: the reciprocal of the reciprocal of the faster moving body minus the reciprocal of the slower moving body is equal to the synodic period of the bodies (the time between conjunctions) – Thank you Nicolai Copernicus.

    It’s also worth noting that the other main harmonic of Jupiter and Uranus gives the average of one of the two common solar cycle length ranges:

    84.01*11.86/(84.01+11.86)=10.39 years (Which also coincides with the JEV syzygy cycle).

    The other being the Jupiter orbital period of 11.86 years

    This observation is thanks to the late Timo Niroma

  32. Greg Goodman says:

    Before we all spend too long speculating about the bumps, perhaps we need to clarify what it is that we’re looking.

    Unless I’ve missed something, the series being analysed is something Tim did quite a while ago using Solex but it not quite sure what settings created it.

    Perhaps Tim could clarify his best knowledge of what this is.

    RMS (x,y,z) ? In which case we’ve lost the sign, which basically rectifies the signal.
    equatorial z-axis component of gravity ? Signed or unsigned?

    I agree there does seem to be a notable c. 164y in the residual plot.
    U: m/r^2=0.0393

    With a gravitational pull on the sun an order of magnitude smaller than that of Mercury, I’m wonder how this is even visible.

    Again, raises the question of what we a looking at.

  33. tallbloke says:

    Help is at hand Greg, I’ve found Tim’s earlier post on the subject, which details what he did with Solex:

  34. tallbloke says:

    I’ve annotated fig 6 for ease of reference

  35. tallbloke says:

    Just for fun: Tims residual vs detrended HADcruV3GL


  36. Greg Goodman says:

    Ok, so as I was pointing out above this does not correspond to gravitation as I was suspecting by the presence Uranus period and lack Venus.

    So we’re back to square one on that question.

    The rough correlation that I spotted in Tim’s residual is interesting. It would suggest that shorter term periods , despite being smaller are the ones that matter.

    This may not be as paradoxical as it first seems. Changes on time-scales like 30y will create a lot less turbulence than sub-decadal changes.

    To put it another way, it is certain that a plot of rate of change would weight things differently and it may be rate of change of (whatever it is) that determines turbulence and hence short-term changes in solar activity at the photosphere.

    d/dt(sin(2*pi*t/P)) = cos(2*pi*t/P) *1/P

    ie amplitude of longer periods are reduced in proportion to the period in the differential.

    I’m still unclear what “location of the solar Z axis” actually means.

    As far I could tell from the manual, DE406 in Solex is a non inertial frame in the sense that it is non rotational. That’s a good start. It seems Tim used barycentre as the coord x,y,z reference (“observer”) point. So I’m guessing that this data is the position of the centre of the sun in that frame of reference. The problem is that the barycentre is moving as well. So this data does not reflect net gravitational force, nor the true position of the sun in a frame with a fixed centre.

    So my best guess is that this is the (signed) z component of the position of the sun relative to the barycentre (ie centre of mass of SS).

    Now, despite a rough correlation, there are still some significant and problematic deviations from the SST record in what we see above. Notably 1900-1910 and 1975-1985 which were LOD minima. Another is the 1945 drop that Hadley inserted into the climate record, which I questioned the soundness of in an article on Judith Curry’s site. I won’t go into that any further here, other than to note that this deviation may be due to data manipulations and not to take it too seriously in evaluating correlations here.

    So I see this as pretty strong evidence that there is something to be explored but we are still not looking at the right data.

    Whether it’s momentum, acceleration or gravitational force, it must be transformed to a static frame, not a point spinning around some where in and out of the equatorial plane.

  37. tallbloke says:

    “Ok, so as I was pointing out above this does not correspond to gravitation as I was suspecting by the presence Uranus period and lack Venus.

    So we’re back to square one on that question. ”

    Greg, you have to remember that the slower moving planets spend much longer above and below the solar equatorial plane. This means their cumulative gravitational effect will be larger than the distant gravitational component would lead you to believe. I appreciate what you are saying about shorter term ‘jerks’, but we have yet to determine the damping factor for the solar oscillation. If it is heavily damped, you won’t get much effect from the shorter term components.

    See Ray’s theory outline at comment #3 and his table of displacements in the z axis from the previous thread:

    Planet   Mass  Distance  Period  Inclination  Acceler.  Displacement
        M      D         P        I
    Mercury  0.056   0.387  0.2408522   3.18       0.021      0.0012
    Venus    0.826   0.723  0.6152078   3.75       0.10       0.039
    Earth    1.012   1.000  1.0000417   7.14       0.13       0.13
    Mars     0.108   1.524  1.880885    5.51       0.0045     0.016
    Jupiter  318.4   5.203   11.86233   6.00       1.228    172.9
    Saturn    95.2   9.538   29.4568    5.45       0.099     86.2
    Uranus    14.6  19.182   84.016     6.36       0.0044    31.1
    Neptune   17.3  30.06   164.802     6.36       0.0021    57.6

    Regarding your centre of gravity idea: Have you downloaded Tim’s planetary data so you can perform the calcs yourself, or do you need us to do them?

  38. tallbloke says:

    “So I’m guessing that this data is the position of the centre of the sun in that frame of reference. The problem is that the barycentre is moving as well.”

    I think the centre of the Sun will always be (0,0,0) in this data.

  39. Greg Goodman says:

    “but we have yet to determine the damping factor for the solar oscillation.”

    I would not waste time trying to guess things like that. Just look for the correlations and let those who are paid to work solar modelling try to explain the mechanism.

    The low pass filter that Tim applied to avoid aliasing in the plotting means the residual is effectively a high pass filtered version of the data. It is the HP data that shows correlation, the big LP components.

    The differential is a form of high pass filter. The d/dt of position is velocity. So this would tie in with attempts at correlating momentum.

    My inclination is still to look at gravity , which has obvious physical means of causing disruption. (I’m not sure why anyone would be linking solar activity to momentum, though I may have missed that bit).

    Gravity and acceleration are related to d/dt of momentum, effectively another HP filter, but you need the full vector , not just the magnitude that has already been looked at, and again relative to a static inertial frame, not SSB.

    The point of maximum displacement in an oscillation is also the point of maximum acceleration. The 90 degree phase shift of one diff becomes 180, so the form is the same just negated, plus the amplitudes now reduces by the square of the period.

    This should actually not be too different from Tim’s positional data with high pass filter, other than the inversion. However, the differences may fix the bits that don’t line up, which as I pointed out above tie in with LOD extremes.

    Since Tim was apparently mistaken in thinking what he had on disk was gravity data, we’re back to the need for positional information on each body plus the sun as time series and working out the mag and direction of each gravity vector at each time step; summing for all planets and projecting the resultant vector onto any direction of interest, eg solar z-axis if anyone has that.

    The maths ain’t that hard. The results of the ‘not quite right’ data would seem to suggest there is a noteworthy correlation underlying all this.

  40. Greg Goodman says:

    TB: “I think the centre of the Sun will always be (0,0,0) in this data.”

    From Tim’s description it sounded like he fed in 0,0,0 as the reference view point and that this meant SSB. What he got was solar position w.r.t. to SSB, ie not zero but what he plotted.

  41. Greg Goodman says:

    TB: “Regarding your centre of gravity idea: Have you downloaded Tim’s planetary data so you can perform the calcs yourself, or do you need us to do them?”

    I downloaded but I don’t see any point in working with data unless someone is capable of telling me what it is. As I commented at the end of the other thread:

    Solex 11 doc says: When the DE406 option is selected, the primary output resulting from the integration gives heliocentric equatorial rectangular coordinates referred to the DE406 reference frame of J2000.

    If someone can explain what that means in English, I’ve just wasted half an hour I didn’t intend to spend and I’m no wiser.

    Now, either Tim’s description of what he did is inaccurate, I’m misunderstanding his description or the manual has an error on this point.

    For now this remains a mess.

  42. Greg Goodman says:

    Maybe I’m confusing the two. What TC posted here is sun position relative to SSB, the data in the other thread was planetary data w.r.t to sun centre in x,y,z but fixed stars in rotational orientation.

    If he included sol in that data it may be enough.

  43. Greg Goodman says:

    unzip -t orb*
    testing: EARTH.OUT OK
    testing: JUPITER.OUT OK
    testing: NEPTUNE.OUT OK
    testing: MARS.OUT OK
    testing: MERCURY.OUT OK
    testing: SATURN.OUT OK
    testing: VENUS.OUT OK
    testing: URANUS.OUT OK
    testing: MOON.OUT OK
    No errors detected in compressed data of
    [/source code]

    Tim. Looks like you forgot one detail, sol 😉

    If you have the exact same settings could you post matching SUN.OUT, o/w please re-run to ensure all files are compatible data.


  44. tchannon says:

    You’ve been busy.

    The existing file this came from was one of those obscene mistakes, a 20M openoffice file is a warning because it is compressed. If I export to xls, ouch. This is beyond sensible for a spreadsheet but it’s done it’s interactive exploration job. At that point I might switch to coding up a script. (or even do it in C)

    Incidentally, I hate spreadsheets. A matter of what actually does the job, is accessible and portable.

    The faster data.
    Sure, can do something with that except this is about practicality. 700 years of data runs off the end of simple display.

    Lets step back.
    What we have now is proof of concept. Reproducing most steps is boredom.

    As Greg points out and what is a worry is the ‘What exactly is the underlying data?’

    Got it open. Runs across to column BB

    Julian date, 10 day samples.
    First width is planet data from Solex, include XYZR for each.

    Next width up is force calculation, planet mass ratio applied to each of XYZ

    And that tells me this is not what I thought it was. This is working on Earth and that explains the large higher frequency component. Never mind, the results are useful just not for what we thought. Sorry about that.

    Finally the X, Y and Z of the planet forces are summed and there is an RMS of that.

    We want Z so the data used is the Z not the RMS

    Extracted first few rows to a new file, see if you can figure it out (19k)

  45. tchannon says:

    Given a result can be produced but is wanted for heliocentric it’s a matter of agreeing the math, I can clone the rest for the parts of interest.

  46. Greg Goodman says:

    TC: “Incidentally, I hate spreadsheets. ”

    Incidentally, so do I. Spreadsheets are for business not science.

    “I export to xls, ouch. ” If we have to use spreadsheet format OO if fine by me.

    Content of the xls file seems to make sense but cols BA BB are showing ### , probably num overflow. (suggest working AU and Earth masses , FFT will be the same. )

    “planet mass ratio …”

    Could I suggest a bit more precision. You seem to be rounding to 2 sig.figs.
    S 5.6846×10^26 kg cf 570

    No sense in degrading the time series before we start.

    “And that tells me this is not what I thought it was. This is working on Earth …”

    So what has been plotted and discussed here is gravitational force experienced by Earth due to the 5 major planets , without sun and moon. Is that correct?

    Or the z cmpt ?

  47. Greg Goodman says:

    Suggest adding S & M in since both E and M will experience small variations due to planets hence S-E linkage will not be just the average annual cycle.

    That probably explains the deviations I noted above.

    Let’s stick with these data for the moment. Looks promising. Can you provide same thing with Sun, Moon and all true planets (prob ignore Pluto) but with “everything on” in Solex except the recent tweaks that may be less accurate over this time scale?

  48. tchannon says:


    The underlying float in memory is invariant.

    This passes through two layers before you see it.

    It passes through display format then a window with a fixed width which is the column width.

    ### means it cannot display in the format set for that cell at that display width.

    Widen the column. Such as drag with the mouse or select column, right click, adjust optimum

    Yes it is for earth, not the sun. Broadly it will look the same but this is why the higher frequencies are so prominent. (I was nervous about this, hence the warning in the original article)

    From what I dug out yesterday it appears the fundamental frame of reference is rooted at the barycentre. Solex can reference to a fixed point, a specific one is 0,0,0 which is the barycentre.
    With any other reference it omits the barycentre as a specific output. This suggests the most flexible output is barycentric, you get a full set of planets and the sun.

    Cloning the computation wouldn’t take long so doing it clean seems a good bet and I can leave the earth stuff sitting there for more work later.

    What I am not clear about is the wanted gravity calc. What I was doing, reason forgotten, is compute the gravitational attraction based on the mass of the two bodies but I have to assume I did some work to figure out the math and might have simplified. (I often do)

    Product of the two masses.

    Then it is a function of the X, Y or Z, and the actual distance and the above.

    This looks like scaling the effect according to distance and scaling X Y or Z accordingly.

    Final being used is simply sum the Z. Signs will cancel accordingly.

    This does not give direction.

  49. tchannon says:

    “Can you provide same thing with Sun, Moon and all true planets (prob ignore Pluto) but with “everything on””

    Already is except the sun, which oddly enough would give a 1 year. I expect I omitted that as just a nuisance but it would give a phase reference.

    Can’t see the original files, no sun data. Could try and figure it, do another solex run.

    Would probably be better now that I know what is wanted to script (or hack C since doubles are plenty) external processing for handling solex to primary time series. (would collapse the ss size too)

    Centring on the earth ought to use geocentric?

  50. tallbloke says:

    The documentation for JPL Horizons seems more comprehensive. I think I’ll have a go at using it to get data for the individual planets and send it down to Tim to compare with his Solex series. We need to get this right so using both and cross checking seems worthwhile.

    So we should set barycentre to (0,0,0) and refer to the solar equatorial plane. Agreed?

  51. Greg Goodman says:

    TC: “Already is except the sun, which oddly enough would give a 1 year. I expect I omitted that as just a nuisance but it would give a phase reference.”

    You missed my point. Earth orbit will see perturbations (not talking about 120ka precession ) like anything else. Sun is not static either. Could be trivial but let’s not assume. There will be a dominant 1y year but orbit is not a circle and not constant.

    I know moon is in the model of SS unless you select the simplified option that lumps E-M together. But what I meant is what is needed here is the moon’s position as time series, like the planets. Moon has a major effect on Earth and moon will be perturbed by planets, implying a modulation of lunar effect.

    “Centring on the earth ought to use geocentric?”

    Really doesn’t matter as long as it is known and consistent. Though, if for Earth, geo is easiest since you don’t need to do trig to get the distance. Idem for solar and heliocentric (rather than bary).

    “Final being used is simply sum the Z. Signs will cancel accordingly. This does not give direction.”

    The spread sheet snip you posted just added scalar magnitudes. That is not enough. I don’t see any reason to look at just z-axis in the case of Earth. but can easily be fixed given the relative position as x,y,z.

    First thing is to plot magnitude of total net gravitational pull.

    To see whether LOD is caused by planets, you probably want to look at the cross product of net planetary gravitational attraction, unit vector of earth axis and unit vector of instantaneous of the E-Sun direction. ie how planets affect the time the wet side of the Earth faces the sun.

    In view of the fact that just the z component looks relevant , hopefully the full calculation should produce something worthwhile.

  52. Greg Goodman says:

    TB, yes cross checking is an excellent idea. Always a good strategy.

    TB: “So we should set barycentre to (0,0,0) and refer to the solar equatorial plane. Agreed?”

    Depends what you want to look at. Be careful not to start confusing the two things we have here that we have just managed to unmix.

    could I suggest we restrict this discussion to follow gravitation effects on _the Earth_ for now, since that is what all the graphs here relate to (despite initial statements that this was solar).

    Your last overlay plot looked interesting, so let’s follow through on that for now.

    What I’d like to see at this stage is MARS.OUT , VENUS.OUT, etc for 9 planets , sol and moon. All from one run with as many SS objects as the program can carry.

    If you want to create a couple of cross-check files include J because it’s big and Mars because it’s an odd-ball. Or while you’re in there do all 9 + sol + moon, if you like.

  53. Greg Goodman says:

    To answer your question: for Earth , geocentric. Centre of earth if possible else geographic N pole. Actual earth obs point will probably mainly affect Moon results but anything off the earth’s axis will drift throughout the year with respect to a fixed frame since time of day moves with realative posn. of sun.

    For solar, heliocentric , equatorial plane (worry about tilt later).

  54. Greg Goodman says:

    for reference.

    planet	number	distance/km	   mass/kg	  m/d^2
    mercury		57909227	3.30E+023	98435100.43
    venus		108209475	4.87E+024	415678474.15
    earth 		149598262	5.97E+024	266858626.94
    mars		227943824	6.42E+023	12350075.48
    jupiter		778340821	1.90E+027	3133137631.95
    saturn		1426666422	5.68E+026	279220988.42
    uranus		2870658186	8.68E+025	10534326.86
    neptune		4498396441	1.02E+026	5060890.17
  55. tallbloke says:

    WordPress formatting is one of life’s mysteries 🙂

    OK, I’ll see what I can work out. I won’t be downloading such big datasets at home though, so it’ll wait until tomorrow.

  56. tchannon says:

    You want to do this on distance, not z-axis?

    For earth restricting the time length would help, can be easily extended later. 1700..2030 ought to be enough and knocks 10k lines off.

    10 day is marginal for earth, caution is needed over Nyquist. Ploy, use that for now and if the result are interesting do a shorter daily to check it matches. I actually expect most of the fast stuff to be irrelevant.

    Numeric precision is a slight problem because I am not set up for extended, which would take some time to do. I’m inclined to use what is. Input precision is limited anyway.

  57. tchannon says:

    Bit of luck for a change. I’ve successfully scripted turning a list of Solex output files into a unified file of the parts needed for computation.

    As a second move rather than simply operating on the numeric text I parsed into floats and wrote back out using defaults. It clones less any trailing zeros.

    A third move was parse an X Y Z set of columns and compute R, write that out. Same plus one more output 4.5628252897542

    With care it looks as though this can be script processed for the gravity calc, producing a time series for analysis.

    That will speed things up for further investigations.

    Now I need to know what computation is needed.

  58. tchannon says:

    I think I’ve cloned the existing computation in script, hopefully fairly flexibly. (should adjust automatically to planets included, knows from names in files)

    Uses geocentric data with sun and moon. (Solex accepted negative surface altitude, ’tis complaining of the heat)

    As expected this is overwhelmed by sun and moon. Push those out of the way and what is underneath looks as though it is like seen previously.

    Have more sanity checks to figure out.

    This might not be the calc you want to see. Starting point.

  59. Greg Goodman says:

    Tim, Sun and moon needs some thought. Sure all the rest is just a small perturbation on this and perturbation theory shows that such small signals can usually be treated additively without substantial error. My reason for wanting them in was that they will also contain perturbations and I did not want to throw that out as well.

    eg. Earth’s orbit is affected by major planets in conjunction/opposition , affects LOD. Correlation of LOD to climate looks non-coincidental and may indicate something that needs to be taken into account. Equally since moon is so important , any deviation of moon by planets may be noticeable.

    Inclusion of S & M does not matter for FFT, so doing FFT on series with and without these two and subtracting (or superimposing the plots) should show any smaller effects aside the big expected peaks.

    This would be worthwhile is assessing how much indirect influence the planets have as well in addition to the obvious direct effect of net gravitational pull.

  60. Greg Goodman says:

    TC: “I think I’ve cloned the existing computation in script,”

    Good result. Does that enable you to say what we have been looking at so far in all the graphs?

    z or R ? How many planets were summed ?


  61. Greg Goodman says:

    Now you have a good, known dataset out of Solex, could you post an archive of .OUT files as before including the S & M files, plus sufficient information for the dataset to be reproducible? ie Solex version and notes on relevant program settings.

    That way, if something useful comes out of this we know where it came from and results will be replicable.

  62. Sparks says:

    tchannon says:
    November 20, 2012 at 1:43 am

    Do you feel able to tell us more? (looks across at lots of paper, pencil)

    Not so much paper and pencil, than pushing buttons for a long time. (ha ha) I wanted to understand where the planetary orbits of Neptune, Uranus and Jupiter were during the progression of the sunspot cycles, but based from the orbit of Neptune. I measured the distance of Uranus from Neptune for 400 years and then I measured the distance of Jupiter from Neptune and plotted the distances in Astronomical Units to the dates on the sunspot record. (Even Svalgaards calibration).

    I had to configure the program to a point on Neptune and get the distances of Uranus and Jupiter and type the positions manually into a spread sheet and step it back a year and repeat, as the software I have that calculates orbital positions does not give a readout, and I couldn’t find the orbital data I was interested in online.

    It may have been a laborious task and for not much of a result, but the chart I made looks very interesting and has furthered my own understanding with the added bonus, that if anyone were to ask me if I knew the orbital distance between Neptune and Jupiter or Neptune and Uranus, I can say, “I do… What date would you like to know the distance for? Would you also like to know what the suns activity was like for that date too?”.

    I can tidy a version of the chart up so that it’s somewhat readable and post it for you to look at in about an hour or so, You can also give me some feedback on it if you like.

    Just out of interest, I’m also currently plotting other planetary data to different sunspot data, that will be added when ready, and I have been writing a program that displays Heliographic maps: 195, 171, 284, 304 as a 3d model that I’ve just began and that I have lots of ideas for, I did a quick post on it here

  63. Sparks says:

    Orbital distance of Jupiter and Uranus from Neptune vs Monthly Sunspot number

  64. tchannon says:

    Reckon Tallbloke will be all over you.

    It’s a pain when you have to fight tools, if only they did what you want directly.

    What you are doing looks useful avenue for exploration. I’ve wondered about relationships too and whilst there are various snippets around it looks like more could be done.

    Let me see. Oh yes. Pay great attention to equatorial zero crossings, seem to be dragons there.
    If there is data, see anything bizarre 1951? I have a specific reason for asking, came from the Wolf and Patrone paper which by accident highlighted what I was looking at.

    I mustn’t follow this up at the moment or I will lose the current focus. If I can help, of course.

  65. Sparks says:

    tchannon says:
    November 22, 2012 at 2:56 pm

    “see anything bizarre 1951?”

    I’m not 100% sure what you are referring to?, in 1951, it maybe unrelated but, all the planets were on the opposite side of the solar system from Jupiter twice that year, in a way the configuration would have had a noticeable effect if you are going by this??

    here’s a snapshot of 1951 planetary configuration

  66. Greg Goodman says:

    Orbital distance of Jupiter and Uranus from Neptune vs Monthly Sunspot number

    Thanks Sparks, I think you have safely ruled out any correlation between the those two distances and sun spot number. There are 9 planets, if you count Pluto. That gives you 72-2 to go. For brevity, it would probably be best if you flagged us when you get a positive result, rather than interrupting this thread every time you get a negative. 😉

  67. Sparks says:

    Greg Goodman says:
    November 22, 2012 at 5:22 pm

    That’s pretty harsh!

  68. Greg Goodman says:

    A bit direct perhaps plus a touch British humour, but don’t take offence.

  69. tchannon says:

    There is an abnormal looking sunspot event, sequence of three during the falling part of a sunspot cycle. This is time coherent with a barycentre closest approach.
    I spent some time trying dig out reports back then, lots of almost but records are bitty.

    Next similar event is unfortunately in the peak of cycle err… 22? 1990 or 1991 and whilst this was the original event I had remarked about nothing definite has turned up marking it as more than odd.

    Prior to that 1851, another looks like a case but data, not a lot.

    And so on. Not supposed to be an immediate solar reaction yet we know little.
    Here we are

    Zero crossings are another set of oddities.

  70. Greg Goodman says:

    Tim, any chance of you zipping up the OUT files? Thx.

  71. tchannon says:

    Slight cross-understanding. I have the same process automated by a computer program as was done partially by hand and partially in a spreadsheet, but I hadn’t tried to discover the Solex settings which produced the Solex data used originally above.

    Took all of 2 minutes to figure out the Solex settings. Know the date and have most of the basics, header gives clues. Compare Solex display with SS, same apart from last digits/rounding.
    DE406, 1400,1,1, orthogonal, 10 day, aberration and light speed off, otherwise defaults, comes up geocentric. Output would be for 23,000 samples. Position on earth? Doubt it matters, with a good reason being the 10 day step, to worry about that, use hours.

    This means I can cross check the program result with the existing SS result.

    Which I have now done.
    One finger trouble in the script and finger trouble in the spreadsheet, fortunately with very minor effect. I’ve done corrected the SS, done the same computation via script, clone excepting last digits.

    That is data already in the SS against newly created Solex data via a program so we are go when I’ve completed some crossing and dotting.

    I’m scaling mass. If we really want accurate units, AU will have to be sorted.

    At the moment I am in exploring mode where it is not worth recording much, takes too long, quick look first. Can usually reverse what I did. In this case after a year or so, not so easy but now done.
    I think it is sufficient to have a few Solex created headers and the data. Not a lot else matters.

    Building tools points at a more serious stage.

  72. tchannon says:
    Don’t want to do this too often, takes some time. [to upload]

    More in there than you expected. but is still ink drying, literally altering stuff.

    OUT files are the recreated output from way back.

    solex-comb,txt is the name for now of the used set of .out files imported as a tab separated list, drop into any SS.

    solex_cmb-m.txt is the computed time series tab delimited, above work uses Z except this is corrected version produced by the program. files.txt is the list of files to tell the program which to use. All fast hack stuff.

  73. Greg Goodman says:

    Tim, thanks for the file. However, I do have some difficulty understanding your rather terse writing style. I read what you’ve written three times and then have a guess at what I think you meant. I’m sure you’re busy and don’t want to waste much time on this but being a little more explicit would be helpful at times.

    TC: “Uses geocentric data with sun and moon. (Solex accepted negative surface altitude, ’tis complaining of the heat)”

    I understood this to mean that you had done a new run, following my suggestion for centre of earth obs point, yet you now say “OUT files are the recreated output from way back….solex_cmb-m.txt is the computed time series tab delimited, above work uses Z except this is corrected version produced by the program.”

    So what you’ve posted here is the original Solex run from years back but parsed by your lua script instead of by a spreadsheet.

    Is that correct ?

  74. Greg Goodman says:

    TC: “I’m scaling mass. ”

    Are you now using planetary masses with accuracy comparable to the rest of the data, not just 2 s.f. ?


    PS. geocentric does not mean centre of the earth (IMO) but _an_ earth based obs. point. usually a real observatory.

  75. Greg Goodman says:

    The point on earth will probably only make a difference for MOON.OUT , but here it will matter if we are trying to evaluate any possible perturbation of the Moon-Earth coupling by the planets.

  76. tallbloke says:

    “geocentric does not mean centre of the earth (IMO) but _an_ earth based obs. point. usually a real observatory.”

    The JPL ephemeris allows you to choose which observatory. 🙂

    I didn’t get time to download today at work, back there Monday.

    Parking this URL here so I don’t lose it

  77. tchannon says:

    The network was crawling perhaps explained by a major storm approaching, about to take out power for the whole area, still phases out all over the place. Unscheduled candle-lit dinner plus some other bother including having to recover some files. Life.

    Yes you are right, I uploaded the existing work, corrected version and created by a program. I want to make sure this makes sense before changing too much.

    The corrected data looks more sensible, no surprises. I’ll leave it at that.

    Mass is 10^24 which is then squared. Experience tells me it is better to keep exponents within a sensible range. If we need to know scale, take that into account.

    Planetary masses are not accurately known, at best are estimates. maybe to 2 digits. This tends to be about who you believe, all appearing serious yet disagreement is common. To 12+ digits, err, you are kidding?

    I’m going to finish off what I was doing to the software then I should be ready to produce the version you wanted.

    I’d prefer to shorten the data to speed things up, can always be redone later.
    Would say 1700 through 2030 @10 days be useful? This would include the sun and moon?
    Centre of earth not that I expect this to make a blind bit of difference.

  78. tchannon says:

    I’m all in with coding, need a break so I’ve taken a peek at the data with the sun and moon included.

    As I guessed would happen the real factor is a constant acting on the earth… is in orbit. This is fine except I wonder whether we are biting off more than can be handled.

    I don’t understand what is appearing, perhaps chasing maths errors because the residual is tiny and looks incoherent.

    So far the moon hasn’t figured much. This might be because it is too close to the Nyquist limit, implying I need to try at a higher sample rate, even larger data.

  79. Greg Goodman says:

    TC: “Planetary masses are not accurately known, at best are estimates. maybe to 2 digits. ”

    Oh, I was not aware of that. All of this is empirical guess work anyway since the system is too complicated to be solved analytically. Mass ratios are more accurately calculated than absolute mass in kg, so let’s go with that.

    Absolute units don’t really matter for what is being looked at here: correlation rather than mechanism.
    Similarly your extraction of 10^24 is fine. You could alternatively convert all the distances in AU during your parsing.

    A best guess would probably be IAU working group figures:
    Mercury MS/MMe 6.0236 x 106 3 x 102
    Venus MS/MVe 4.085 237 19 x 105 8 x 10-3
    Mars MS/MMa 3.098 703 59 x 106 2 x 10-2
    Jupiter MS/MJ 1.047 348 644 x 103 1.7 x 10-5
    Saturn MS/MSa 3.497 9018 x 103 1 x 10-4
    Uranus MS/MU 2.290 298 x 104 3 x 10-2
    Neptune MS/MN 1.941 226 x 104 3 x 10-2
    (134340) Pluto MS/MP 1.365 66 x 108 2.8 x 104
    (136199) Eris MS/MEris 1.191 x 108 1.4 x 106

    Note this is effectively the reciprocal of the masses we want.

  80. Greg Goodman says:

    TC: “So far the moon hasn’t figured much. This might be because it is too close to the Nyquist limit, implying I need to try at a higher sample rate, even larger data.”

    Yes, far too close. Probably 3 day is max interval that will characterise the various lunar cycles circa 29 days with a useful resolution.

    A shorter run would be fine to explore. If something significant comes out, then a full detailed run could be done later. Suggest 1750-2050 , that catches a some notable climate and sunspot variations, allowing to see whether there is any interesting linkage to be examined further.

    As you say S & M will swamp the time-series. What is needed is to examine the FFT of these two in isolation to determine whether they can be put aside. In reality, it expect both will manifest a small planetary signal as well as the obvious cycles.

    I don’t want to jump the gun and start assuming mechanism at this stage but an FFT of SUN.OUT should be worth isolating. This will show perturbations of the most important distance. Since both ends of this line are affected by other bodies, the nett effect should be informative.

    Equally, an FFT of 3 day data of MOON.OUT will show how the lunar gravitational pull is affected. I suspect planetary influence here will not be negligible. Again, the expected cycles will dominate but FFT will separate them out and clearly show the also rans.

    As I said earlier, this 3 day run for lunar position definitely wants to be earth centre, not an earth surface coordinate. We don’t want to pollute this with the (grossly undersampled) earth’s rotation.

    Since it does not imply any extra effort or run-time, I would suggest adopted earth centre for all further work.

    My earlier observation that the 2 year HP filtered z axis data bore some resemblance to climate and Tallbloke’s SST overlay, suggest there are some missing elements. This sun and moon data seems the obvious first place to look.

    If you could make such data available, I have some software for FFT.

  81. Greg Goodman says:

    Greg: “Similarly your extraction of 10^24 is fine. You could alternatively convert all the distances in AU during your parsing.”

    Configure Solex to output in AU 😉

    here are the relative masses the “right” way up.
    Their uncertainty estimates indicate uncertainties in the last (5th s.f.) digit, so suggest using at least 6 s.f. to avoid further degradation.


    Mercury 1.660137e-07
    Venus 2.447838e-06
    Mars 3.227156e-07

    Jupiter 9.547919e-04
    Saturn 2.858857e-04
    Uranus 4.366244e-05
    Neptune 5.151383e-05
    Pluto 7.32246e-09

    Constant of gravitation G 6.674 28 x 10^−11 m^3/kg/s^2 ; unc: 6.7 x 10^−15
    Astronomical unit [d] au 1.495 978 707 00 x 10^11 m ; unc: 3 m

    Geocentric gravitational constant
    GME [TT-compatible] 3.986 004 415 x 10^14 m^3/s^2 ; uncertainty: 8 x 10^5

    Earth/Sun mass ratio is not given directly on this page, so use GMe product directly.

  82. Greg Goodman says:

    Using the above mass ratios and au, will probably leave numbers circa 1e-09 , though more reasonable, it’s a factor you may want to drop before doing FFT, for reasons you already stated.

  83. Sparks says:

    Greg, No offense taken, I was in a rush.

    “it would probably be best if you flagged us when you get a positive result, rather than interrupting this thread every time you get a negative”

    I was showing the planetary data I produced, in graph form to point out that it looks good, it isn’t a negative result, and it’s not meant to show a correlation between the three planets and the Sun spot number. The data is actuate enough to work with using other factors such as planetary mass etc… I have all the other planets completed excluding Pluto for now, it actually went a lot quicker once I got a good system going (thanks for asking lol).
    As for interrupting the thread? I picked up on something tchannon said at the time and the conversation evolved from there. we’re exploring the same subject and discussing it, although I usually just read it’s still on topic.

  84. tallbloke says:

    Sparks: you are (more or less) on topic, and welcome. Everyone explores different data relating to the same solar system. Someone is going to hit the jackpot once in a while, and because we are all in the early stages of this stuff, anything goes.

  85. tallbloke says:

    Are the strengths of solar cycles determined by converging flows towards the activity belts?

    R. H. Cameron and M. Schüssler
    Max-Planck-Institut für Sonnensystemforschung, Max-Planck-Str. 2, 37191 Katlenburg-Lindau, Germany
    Received: 28 June 2012
    Accepted: 19 October 2012
    It is proposed that the observed near-surface inflows towards the active regions and sunspot zones provide a nonlinear feedback mechanism that limits the amplitude of a Babcock-Leighton-type solar dynamo and determines the variation of the cycle strength. This hypothesis is tested with surface flux transport simulations including converging latitudinal flows that depend on the surface distribution of magnetic flux. The inflows modulate the build-up of polar fields (represented by the axial dipole) by reducing the tilt angles of bipolar magnetic regions and by affecting the cross-equator transport of leading-polarity magnetic flux. With flux input derived from the observed record of sunspot groups, the simulations cover the period between 1874 and 1980 (corresponding to solar cycles 11 to 20). The inclusion of the inflows leads to a strong correlation of the simulated axial dipole strength during activity minimum with the observed amplitude of the subsequent cycle. This in agreement with empirical correlations and in line with what is expected from a Babcock-Leighton-type dynamo. The results provide evidence that the latitudinal inflows are a key ingredient in determining the amplitude of solar cycles.

    The simulated flows in the plots are worth a look bearing Ray Tomes theory in mind.

    Register for free at A&A to access the full article

  86. tchannon says:

    I’m going to the constants from a document showing Author: Royal Greenwich Observatory.

    Click to access Astronomical_Constants_2009.pdf

    Reason: Specifically deals with DE405 upon which DE406 is based. IAU figures are for DE421

    From that I’ve worked out this table.

    Constants for DE405
    Relative Mass kg Mass exatonne
    Sun 1.00000000000000E+00 1.98840000000000E+30 1.98840000000000E+06
    Earth 3.00348959632312E-06 5.97213871332889E+24 5.97213871332889E+00
    Moon 3.69430370015055E-08 7.34575347737935E+22 7.34575347737936E-02
    Mercury 1.66013679527193E-07 3.30101600371871E+23 3.30101600371871E-01
    Venus 2.44783833966454E-06 4.86728175458898E+24 4.86728175458898E+00
    Mars 3.22715144505387E-07 6.41686793334512E+23 6.41686793334512E-01
    Jupiter 9.54791938424322E-04 1.89850829036292E+27 1.89850829036292E+03
    Saturn 2.85885980666103E-04 5.68455683956479E+26 5.68455683956479E+02
    Uranus 4.36624404335156E-05 8.68183965580025E+25 8.68183965580025E+01
    Neptune 5.15138902053550E-05 1.02430219284328E+26 1.02430219284328E+02
    Pluto 7.39644970414201E-09 1.47071005917160E+22 1.47071005917160E-02
  87. tchannon says:

    I’m seeing spectra trains but with essentially no lunar data. Don’t know what this means but it is not useful, assuming it is real. (it does look sane)

  88. Greg Goodman says:

    TC: “I’m going to the constants from a document showing Author: Royal Greenwich Observatory.”

    Those figures are exactly the same as what I posted from the other source, within +/-2 in the 5th sf. , they agree to within the stated uncertainty. So it seems as long as we use relative masses there is no divergence in opinion on what these constants should be.

    Good news.

  89. tchannon says:

    I’ve computed from the ratios by using the given Solar mass. Absolute values are as you say unimportant.

    Right now I am getting a feeling for whether things are right.

  90. Greg Goodman says:

    TC: “I’m seeing spectra trains but with essentially no lunar data. Don’t know what this means but it is not useful, assuming it is real. (it does look sane)”

    What do you mean here about no lunar data? Sane? Again, please be a little more explicit, I don’t know what you are saying here.

    If you are still using MOON.OUT based on 10d sampling you will likely not see any clear monthly patterns, which you are presumably expecting to see.

    The anomalistic month (closest approach) is 27.55d the synodic month (S-M-E alignment) is 29.53d.

    Both these may be expected to give a signal and may be expected to resonate. If you sample that at 10d it will alias all over the place and just add noise to the TS and spurious peaks into the spectrum.

    To see whether there are any long term cycles in the lunar gravitational pull (due to large planets deviating its path, for example) you will need to sample at no more than 3d , gaussian filter at say 60d then decimate down to 10d interval matching the other data.

    As I have already suggested, it would be worth looking at short interval (non filtered) moon G on its own initially, to see what frequencies are present. Then you will know what is present apart from the obvious cycles and will be able to recognise anything that may show up in combined data.

    If you do such a high-res run of MOON.OUT, please post TT,x,y,z zipped. I have some other data I’d like to compare it to.


  91. tchannon says:

    Any idea what precisely the moon being geocentric means?

    There is some kind of lunar data but the amplitude is low, there are going to be dynamic range problems if ordinary D(F)FFT is used. Primary lunar term is at -54dB below earth, period 29d 12h 44s which is about right.

    I’m not set up to handle high dynamic range on DFFT but I can workaround these things except it takes a lot of time and effort. The analyser software on the other hand doesn’t care, not windowed either nor needs to be. Takes a long time to run.
    Experimentally went for lunar 7 day where I know there is a doublet+ in LoD data but is rather small. Locked immediately to something ~107dB below earth annual. Then added another, nice Lissajous from that pair but whether they are real, can’t tell, might be junk. It could have used other junk but no hunting. Maths doesn’t make real sense. With LoD many of these calc to 18.6y and maybe harmonic too. This one ~1.44y and ~4d, hence I suspect it isn’t real. Didn’t take time on looking further.

    In this kind of situation it’s better go find a specific thing, usually visible as something vague in DFFT. Remove and look again. If something is still there go get that too. Automating this is not feasible. Sometime more advanced stuff suggests itself but this is not a tutorial for an unreleased tool.

    If you want a dataset I suggest 5d. integer of 73 per year. Daily can be done, is size tradeoff. From other work I know the moon has factors out to at least 4 days. The general Nyquist problem really has nothing to do with what we use so much as the mis-sampling on creation. The only proper fix is with the sampling and that would mean writing our own software.

    Alternatively the more practical answer is decimate fast sampling to a more practical rate.

    There again what is the objective here?
    Rog wants to look at solar z-axis. Already have had a glimpse.

  92. Greg Goodman says:

    “There again what is the objective here?
    Rog wants to look at solar z-axis. Already have had a glimpse.”

    The initial thing was to use correct gravity on Sol instead of the barycentre mistakes and using distance instead of net gravity vector. I think that is the correct way to asses any possible planet SSN interaction and is worth doing.

    The Earth based idea really sprang out of your false recollection of what this data was when you first posted. It was an accident but an interesting one. When you did a low pass to get it to plot, the residual (ie HP filtered) seemed to have some vague correlation to climate indices. This was direct planets only, and had some significant deviations. This IMO begs that it be looked at more thoroughly. The result should either solve the deviations or trash the apparent correlation entirely. Either would be informative.

    As TB says, the initial plot was “tantalising” .

  93. Greg Goodman says:

    “With LoD many of these calc to 18.6y and maybe harmonic too. This one ~1.44y and ~4d, hence I suspect it isn’t real. Didn’t take time on looking further.”
    Don’t understand where you 1.44 came from but 18.6/1.44=12.9 : lunar cycles/earth year?

  94. tallbloke says:

    Greg Goodman says:
    November 25, 2012 at 7:03 am
    “There again what is the objective here?
    Rog wants to look at solar z-axis. Already have had a glimpse.”

    The initial thing was to use correct gravity on Sol instead of the barycentre mistakes and using distance instead of net gravity vector. I think that is the correct way to asses any possible planet SSN interaction and is worth doing.

    Sorry Greg, I’ve lost track a bit. Are you going forward with this? Do you have a dataset to work with now?

  95. Greg Goodman says:

    TC: “Alternatively the more practical answer is decimate fast sampling to a more practical rate.”

    Which is exactly what I suggested for MOON.OUT above.

    TC: “Any idea what precisely the moon being geocentric means?”

    If you are asking: why do geocentric moon only, the moon is the strongest tidal influence on Earth. Differences in geocentric distance to moon will modulate it. You suggested excluding S&M from the calculations because of the dynamic range problem, but before doing that you need to assess what you are going to ignore.

    The planets will affect the moon just as much as Earth and any deviation of the moon’s orbit will affect the Earth. These will not be equal and parallel, they will affect precession of lunar nodes etc.
    The effects could look very different from the direct planetary effects.

    Removing moon.out before evaluating this invites the risk of distorting the result and degrading correlation.

    If you use a 60d LP gaussian as I suggested (run twice if necessary) the dominant cycles will be sufficiently attenuated that the dynamic range probably won’t be so problematic. A 2d moon only run could then be resampled at every 5th point (NO averaging or interpolation) to coincide with the other OUT files.

    TC: “There is some kind of lunar data but the amplitude is low, ”

    In composite *.OUT ? We’ve already agreed that the lunar signal will have been totally mangled by being under sampled. Most of it will be aliased elsewhere. I’m not surprised any remaining amplitudes are tiny. But they shouldn’t be. That in itself is proof that MOON.OUT needs resampling.

    I understand if you don’t have time but I think this whole line of enquiry will be compromised unless that is looked at.

  96. Greg Goodman says:

    TB” Sorry Greg, I’ve lost track a bit. Are you going forward with this? Do you have a dataset to work with now?”

    I have planets but I don’t think we have a useful MOON.OUT. I hope my lastest comments explain what I think is needed and why.

  97. tallbloke says:

    So far as effects on the Sun by the planets are concerned, wouldn’t it be sufficient to treat the Earth-Moon system as a single entity?

    I think JPL does that in the data I used.

  98. Greg Goodman says:

    For effects on sun , yes. I think that is a separate though similar investigation. I have some ideas on that as well, but I’ll avoid further confusion by discussing the two here.

  99. Greg Goodman says:

    Tim, will the free version of Solex produce this output to the same accuracy?

  100. Greg Goodman says:

    Orbital forcing of tidal cycles is only a small portion of the spectrum (Fig. 67) of astronomically-driven periods which exert gravitational effects on Earth and the affairs of humankind (Rampino et al. 1987). The same periodic behaviour governs changes over time in the [b] energy distribution reaching Earth’s atmosphere [/b], and must have done so throughout geologic time. We are concerned here only with tidal phenomena within a small portion of the calendar and solar frequency bands.

    The orthodoxy says planets influence is minimal. However, in the planets-only plot Tim produced above I already pointed out the significant energy around 1.2y.

    This corresponds to the Chandler nutation on the precession of the Earth’s axis or rotation.

    If precession is accepted as being behind Milankovic cycles governing cycles of glaciation, it is implicit that the Chandler nutation is of the right order of magnitude to produce climate cycles within an interglacial.

  101. Greg Goodman says:

    figure6 : there are two peaks vying to 2nd position. One around 1.25y seems to be the 3rd harmonic of Jupiter.

  102. Greg Goodman says:

    In recent decades precession nutation models have been revised more fre-
    quently, and in the process, grown quite complex. The 2000A version now
    includes almost 1400 terms (Bangert 2002) yet it still does not accurately
    predict changes in the precession rate trend, and it has failed a key test: dy-
    namic equivalence. In 2006 the IAU noted the precession nutation model is
    “not consistent with dynamical theory” (IAU Resolution P03). This same res-
    olution recognized that the planets must make a “significant contribution to
    the motion of the Earth’s equator making the terms lunisolar precession and
    planetary precession misleading.”

    At least in astrophysics we don’t pretend the science is settled. 😉

  103. Greg Goodman says:

    … After modifying this theory with an array of new inputs, and making an increasingly wide number of assumptions about the inner earth, it has now been decided that the moon and the sun must have so little to do with “changes” in the precession rate that the term lunisolar precession be eliminated (P03 recommendation) and the planets really do make a “significant contribution.”

    And if they can affect precession they can affect climate.

    Game on !

  104. tchannon says:

    “Don’t understand where you 1.44 came from but 18.6/1.44=12.9 : lunar cycles/earth year?”

    Doublets. The period spacing of the doublet calculates to show the doublet is a modulated single where the modulation in LoD data is usually 18.6y (from f1.f2 / (f1 +- f2)

  105. tchannon says:

    The free version of Solex *is* Solex.

    “Basically, SOLEX is a free computer program modelling the N-body dynamics of the Solar System, and it is the result of a long and patient amatorial work by the author (Aldo Vitagliano).”

    ” Download the installation file for Solex 11.05 Light: SETS110L.ZIP (updated October 16, 2010)”

    Light here means large things omitted. There were additional database style items which were available as an addition. These seems to have vanished.

  106. Greg Goodman says:

    Thanks. Reading the manual I got the impression there was a paid for version with more precision. Maybe it’s out of date.

    Could you give an accurate frequency for that strong peak around 1.2y. Is it an obvious harmonic of J or S ?

  107. tchannon says:

    This is not made up so it is more tantalisation from earth data, rated bizarre+. What on earth, maybe not, what off earth is creating these patterns? I still don’t know if what I am doing is valid.

    Earlier I decided to just scruff of neck, crucify computer time. Produce a large daily run, >100M of output. Set the translation program going and went out to a supermarket. On return it hadn’t quite completed. I then extracted from the result to more series.

    Then the game I am playing, I decimated from daily to whatever is useful. At least this operation is split second. First lesson was I hadn’t left enough guard band: I’d allowed extra time before and after the period of likely interest (1400-2200). Doesn’t matter for now.

    Went for annual, which is on Julian year, so years do not align with Gregorian year, can of worms in this, fractional decimation or other to be decided scheme.

    I then fed that into the analyser software, which munched it double quick, all the usual terms except for one. It seems DE406 includes one long term factor (on removal dead straight line, nothing else out there). Won’t be spot on, was a quickie, does 6245.24 years Julian. Does this mean anything? (not looked anything up) A recent peak was during MWP and now we are far from another.

    Took a squint at the analysis result, uninteresting.

    Pull in the residual, ie. everything not taken into account.

    Trace above is a 16y low pass on that. (just a guess) and the hilbert derived envelope.

    This makes no sense. Which data? Z-axis gravitational force but my math might be wonky. I assume the sun will get a similar yanking.

    I don’t want to make more of a mess so I am still wondering how to figure out whether any of this is valid.

  108. tallbloke says:

    Odd looking periodicity in the blue curve. I don’t think the curve for the sun will look very similar. The inner planets will affect Earth much more than they will affect the sun I would have thought.

  109. tchannon says:

    First three terms are these. Other two given if you want to try and figure relative time. This is Julian time.

    [added: measured via FT, not DFT, there is no binning whatsoever]

    Our friend motls mused on the julian gregorian problem at the beginning of the year

  110. Greg Goodman says:

    Thanks Tim, so that 1.2 peak matches the nutation called Chandler wobble. IIRC you were plotting z-axis cmpt so this is simply a result of the axial deviation on the net force of the planets due to nutation and presumably is not present in spectrum of R.

    As the paper I linked above points out, axial precession does not apply within the solar system , it is only relative to fixes stars. This is apparently implicitly accepted in astronomical practice though current precession theory is still struggling with its 640 parameter model 😉

    So, without precession, nutation is the larges axial deviation and is comparable in magnitude to the variation of net G due to the major planets.

    That helps gauge the relative importance of all this.

  111. Greg Goodman says:

    1.2y plus annual 1.0y cycle will produce beat freq. of 9y

    The main frequency of AMO (real SST, not detrended), revealed by lagged autocorrelation is close to 9y

    There is also a clear fundamental around 5.5 – 6y in N. Pacific 20N-60N SST (real SST, not the mystical PDO index)

    There is evident linkage of about 20% contribution of NP getting subtracted from AMO. Such teleconnections are not controversial in principal, though I have not seen such clear evidence before. Possibly due to the general reluctance to do this kind of cyclic analysis.

    About 35y back in the record NP osc dies out and then starts to rebuild. This looks like a classic interference pattern with roughly 70y period. This data ran to 2009, so -35 is around 1974, the beginning of the most recent warming surge. This is the time of a recognised “phase change” in PDO. Such a phase change is also a characteristic of the null point when two harmonics produce beats.

    [ As shown in the third panel here ]

    The confidence line in the lag plot reflects the reduction in the sample size as lag over-lap increases. This shows the auto-correlation in AMO to be about constant over the range of the graph.

    I need to pull some more precise estimates of these periods, particularly for NP, to work out the second frequency from the resonance.

    It would appear from this plot that NH SST could be characterised by just three main frequencies, one of which is already identified as corresponding to a resonance of annual with Chandler nutation. Since Tim’s plot above suggests J and S peaks of comparable or greater magnitude than Chandler, (even after filtering) it would seem that there is reason to hypothesise a possible gravitational link with the other main climate cycles in NH SST.

  112. tallbloke says:

    1983 article

    Followed up By Buffa and Poma with a planetary and lunar angle here

  113. Greg Goodman says:


    Well I did not want to jump the gun on this because I still have a limited understanding of how “free” this free nutation is. I already noted the significant energy around the 1.2y region (rather than the peak itself). If someone needs some time variant, external noise with frequencies spread in a broad range (not white) there you have it. Planets.

    The true free nutation is supposed to be about 330 days and has been rigged up and down to fit the data by playing with the Earth’s presumed internal structure. This an elegant way of adding parameters and making up your unverifiable constants to fit a model to the data.

    I have a suspicion this nutation may be planet induced but have not read enough for that to be more than an idea.

    from Buffa you linked:

    “La Nina corresponds to negative anomalies in LOD.”
    That is something I have been working on. I think I have a mechanism but it needs more work before I spill the beans on that. This is why I’m interested in planetary effects.

    Needs numbers to see whether it’s credible.

  114. Greg Goodman says:

    Nutter (L. nutare to nod or sway)
    1. Mentally disturbed person, characterised by habit of swaying back and forth or nodding
    2. Mentally disturbed person, who thinks planets can affect climate.

    [1] Lewinsky et al 2013

  115. tchannon says:

    Mystical index (grin) rather my thoughts on certain “datasets”

    The Chandler parameters are not those, wherein lies a problem and why I looked at all at extraterrestrial shoving around of earth.

    1y, 1.19y 6.4y
    1.19 is not 2 1.2

    I stopped when it became clear there were dataset problems appearing on a single source, no cross check.

    I don’t recall doing a post on the Chandler wobble so maybe this is a good time, see if anyone can shed more light.

    Now for an oddity. The large gravity term from the gravity work is very close to the 5th harmonic,
    5.93/5 = 1.19
    Uneven harmonics are not so common but in this case there is a twist, Chandler has strong visual clues there is an asymmetric process at work, perhaps frequency doubling, yet that makes even less sense.

  116. Greg Goodman says:

    “1.19 is not 2”

    did you mean “is not 1.2” , if not what is the 2 here?

  117. tchannon says:

    Fair cop.

  118. Greg Goodman says:

    OK, so if I get your point, what you are saying is that the peak you reported as 1.20371210288 should not be confused with “1.19” of Chandler.

    I think you need to qualify that by stating the expected accuracy of both figures. I would not expect either to be so accurate as to preclude this being the same thing.

    Please correct me if I’ve missed the point.

  119. Greg Goodman says:

    You seem to be applying A.M. formula, I was assuming addition (beats). Addition would be correct for gravity but maybe some assumptions need to be stated for Chandler plus annual. Annual what, and why would this be additive or modulated?

    I’m not pushing one or the other but I think it’s important to say what assumptions are being made. Otherwise someone is going to say “numerology” (and they may be right).

  120. Greg Goodman says:

    Now for an oddity. The large gravity term from the gravity work is very close to the 5th harmonic,
    5.93/5 = 1.19

    chandler = J / 10.00 = J tide / 5.00 ? Hmm , that close a match makes me stop but 5th harmo??

  121. tchannon says:

    How accurate?
    If I have used the analyser software here on a definite item, probably to ppm. I think you will understand more than most, in the simple case this subtracts an exact sine from whatever internal working data it has in least squares sense. Hence phase and amplitude are correct too.
    Hence too terms can be subtracted out of a dataset.
    If the data is probably chaotic is more a question of what it means.
    Experience tells me it works.

    A small plot in thread . Chandler and minimal model overlay.
    Without rounding and without dealing with tiny timebase error

    The plot is adding together those sines (plus offset plus one longer term)

    Not perfect math but sum and difference 1 and 1.19 produces 6.3 and 0.54, with other combinations giving complements.

    The signal is tiny, not happy all is well. For example, we learn the earth did a loop d’ loop turn 2006/2007 which is that nick in the wave, seems to go off and then slowly return to the former trajectory.

    That seems a peculiar thing, shrug. When I eventually followed this up (I did most of this some time ago so the data was new) I discovered a worrying co-incidence. Slap bang at the point of loop is a leap second. Other leap seconds are not so associated. I have other reasons to be wary.

  122. Greg Goodman says:

    OK, so the resolution and accuracy of the analyser seems impressive. I wasn’t expecting that.

    I don’t see the 2006 discrepancy as odd, there’s just another term with smaller amplitude that is not accounted for.

    This is most noticeable when the two model terms are interfering destructively but it is there throughout. It’s additive during 80’s where it’s in phase, decade centred on 2010 is out of phase. 70’s minimum is skewed left, 68 peak gets hit just before max etc. If there is something odd in 2006 it may become clearer if that is accounted for.

    What are you calling “Chandler” here?

    There is much discussion about whether Chandler is a doublet , whether it’s frequency changes. In any case it’s not stable. Your original plot was something like 400 y , now you’re plotting 50y. Chandler, whatever it is, will have changed in that time frame.

    So I’m still not convinced that your comment 1.19 1.2 means the 1.2 is not Chandler (which I’m still guessing was what you meant).

    However, before getting into drawn out discussions and examination of the detail, I don’t think changes in lunar gravity can be excluded. The E-M couple will be affected by the planets, not just Earth. So perhaps the current run could be regarded as being close to that, though it ought to be re-run with EM bary, not centre of Earth, to maintain accuracy.

    At that point I’m not at all sure that we should be restricted to looking at z-axis. Z-axis may be interesting for effects on sun where small deviations to the plane may be relevant. Since E-M is highly inclined and precessing, I really see not special meaning in looking at z only. It seems inappropriate and disruptive unless there is a solid reason.

    Accepting that lunar orbit will be affected brings us back to the need for detailed MOON.OUT , a prominent influence is the tidal one which 1/r^3 , so planetary influence will probably be more significant via the influence on lunar precession than directly. It is the inv. r^3 argument that has been used to exclude planets up till now.

    Bottom line: planets w/o moon needs to be done on EM barycentre to see gravitational effects on the EM couple. Detailed moon run should be used to assess planetary influence on tides.

  123. tchannon says:

    Prior to the 1962 dataset, going back to 1960 is fair but before that the data is plainly terrible yet there is occasionally more coherence. The explanation will lie in the lack of technological competence, was beyond our measurement capability. This pretty much applies to all data fields but doesn’t stop attempts at making a silk purse.

    Messed up data is more the norm than exception, a view from many years in swamps.

    In the case of the wobble it is remarkable it can be measured at all.

  124. tchannon says:

    Rog has an interest in this so I try to help out with things I can do leaving others to do what they do best.

    My original intent here was trying to rough something out a gravitational force as seen by the sun in the z-axis.A look without plane rotation was a first step but I screwed up on the figuring out what I had already done.

    What has now happened is pushing me to automate some of the messy parts, figure out how to do more.

    As it stands I still don’t know the details of the rotation needed. There is more information yet to be dug out. The sun has a fluid-like rotating outer which is going to be electrically conductive all the stuff is there for EM works. An additional question is the unknown core, what does this seem to do and I recall coming across mention of rotation in connection with neutrino work. Need to find that again, whether it gives a clue on the spin axis.

    For now a question is whether it is worth producing helio dataset as is.

  125. tallbloke says:

    I would say yes. But then, I would. 😉
    if by rotation you mean the issue of the 7 degree tilt of the sun to the plane of the planetary orbits, I think it’s important to measure angles to the solar equatorial plane for consideration of Ray Tomes hypothesis, which I regard as one of the best of the current crop.

    I think the ~60 year oceanic cycles being at the same frequency as the tri-synod of J-S is a big clue here. The angular relationship of the alignments in the z-axis to the SEP is repeated every third conjunction, on a slowly precessing cycle which matches the other obvious terrestrial cycle, the ~974 year RWP-MWP-ModernWP

  126. Greg Goodman says:

    OK, for sun I would say heliocentric, use E-M combined as one object. Once you have a full set of s,y,z time series with sun at 0.0.0 it should be straight forward to do the vector addition. That gives one x,y,z time series.

    Then some information about how the solar axis is orientated and if it’s fixed, precessing or whatever. This gets a bit messy already, since there’s no (or little) earth precession of the equinoxes within SS , but there is w.r.t to fixed stars and since all this output is fixed frame that means Earth is precessing. That is apparently equivalent to to SS rotating as a whole. So I guess that means that even if the sun is not processing relative to the planets , it will be now.

    This will not be ascertainable from output that only give positional information of the sun.

    I suppose one trick would be to set obs pt to solar S pole and plot the position of the sun (ie its centre) and should give a means of establishing the rotation vector and any axial drift or precession. Divide by solar radius to get the unit vector.

    Once the solar axis is defined the projection is a trivial dot product.

    If it does move, you just have to do the dot product for each value rather than onto a fixed vector.

  127. Greg Goodman says:

    TB: I think the ~60 year oceanic cycles being at the same frequency as the tri-synod of J-S is a big clue here.

    AMO is about 60 N. Pacific , less clear, not 60y.

    What is interesting is the coincident peak of neg. correl at c.93y. Suggests they will do full cycle and both +ve correl at around 186y.

    Some caution since gaussian filters tend to make everything look like a sine wave.

  128. Greg Goodman says:

    similar, with 12m filter. Inflections indicating change over marked. AMO slightly nearer 60y.

    PDO infection corresponds to 1973 , close to recognised PDO phase change is notable tornado burst of ’74.

    AMO short period seems to get hit by %age of PDO signal putting a dent in the peaks. Keeping within the limits of the phase changes in each, est. AMO at (52.1-9.1)/5 = 8.6y; PDO (32-3)/5 = 5.8y

    Long term decorrel may indicate a natural frequency of the basin resonating with a driver of a slightly different period. It’s tempting to suggest Jupiter may set up a resonance in the heavier waters under the thermocline. Solar pull at perihelion may accentuate this.

    So in a similar way as was discussed for Sol above, maybe the net Earth dataset (with moon) needs to be projected onto the axis, rather than taking z of the orbital plane was done in Tim’s data at the top.

  129. Greg Goodman says:

    looking again at the long autocorrelation plot the two curves have coincident trough (max anti-correl) at 93.6, for AMO this is 1.5 cycles for NP it is 2.5

    93.6/1.5 = 64.2 since these are not symmetrical cosines that is close enough agreement to 63.3 peak.
    93.6/2.5 = 38.52 (this leads me to think I may have mis-identified the crossover earlier due to the annual ‘noise’, -38.5y is probably where the phase change happens).

    5:3 ratio

    NP long autocorrelation would seem to be a smaller proportion of the same c. 64 with a slightly dominant c.38.5y

    The ratio of short cycles I noted : 8.6/5.8 = 1.48 ; 3:2 ratio?

    Any atmospheric linkage would probably have negligible lag on these time scales. These neat ratios may support the idea of linked harmonic oscillators. It is interesting that NP seems to affect AMO in the short scale, the opposite being true in the long terms.

    I’ll let the resident “numerologists” speculate on the origins of these frequencies 😉

    Be warned, post WWII cooling period shows different NP oscillations. Peaks@ (66.76-45.48)/3 = 7.1y rather than 5.8y in the late century warming period , post war AMO is harder to follow since it hits its long term decorrelation when it goes into anti-phase with the stronger NP cycles.

    This latter observation suggests the reason for the long term decorrelation in AMO is NP linkage.

    Final note, I forgot to put the confidence level on those graphs. Major peaks are firmly about 95% significance but to not get more significant as time goes back, they remain about the same. The amplified correlation coeff is due to the reduced window of data as the lag increases. Thus signif is roughly const over the century of data.

    There may be some cyclic sanity behind all that but it’s a complicated web of interactions.

  130. tallbloke says:

    Thanks Greg. I think the oceanic oscillations are a moving target. The Greenland ice core data suggests the 60yr term has great longevity over the long term, but drifts in period and amplitude due to several other cyclic and chaotic factors on longer and shorter timescales. The North atlantic is more affected by ~74yr Lunar tidal effects when PDO is -ve and gets a 45 year inner planet return surge too.

    I don’t know much about the North Pacific, but I’d expect it to be more strongly affected by the timing of strong ENSO events than the Atlantic.

    ENSO itself seems to be solar related, roughly 3 El Nino’s per cycle, strong ones start at solar min, often follwed by La Nina near start of solar max. Then a smaller El Nino at the end of solar max, and another on the mid downslope.

  131. Greg Goodman says:

    “The Greenland ice core data suggests the 60yr term has great longevity over the long term”

    Due to its position Greenland will be dominated by AMO. This is useful confirmation that the c60y cycle I found in AMO autocorrelation is a long term pattern, not just a century scale event.

    PDO is a voodoo index. It’s one of these million dimensional eigenvector principal component gadgets that people (esp. climatologists) use without having the slightest idea of what its effects are. It is also the (negated) deviation of NP temp from the global average of which it is part. What this actually means is anybodies guess.

    I prefer to use real temps over the same area. That’s what I plotted here.

    Also I prefer to look at dT/dt since we are interested in climate _change_ , so let’s plot change not guess it from a time series.

    Thus doing, “trends” become a constant without dubious “detrending” that depends heavily on where you choose to start. Quadratics become linear and don’t mess with the correlation. This helps study the cyclic components. Once we have quantified and have some understanding of the oscillations, we can start to look at the residual. Then and only then can we speculate on attribution.

    TB: “ENSO itself seems to be solar related, ”

    I’ve just been reading a paper detecting some correlation ENSO / LOD. Greater magnitude link to el Nino than La Nina but both showed some correlation. LOD shows some correlation to SSN (at least Svalgaard’s version). Be careful not infer solar caused from your “solar related”, the latter does not imply to former.

    Nino/Nina provides a mechanism getting large amounts of heat in and out of the ocean, ie global heat content. Whatever controls/causes it has a means to strongly affect metrics like OHC and mean global temperature.

    TB:” The North atlantic is more affected by ~74yr Lunar tidal effects when PDO is -ve”

    NB -ve PDO is N.P warmer than global mean.

    I think what you say is related to what I showed in the autocorrelation. NP decorrelates in about half the period you mentioned. The decorrelation of NP corresponds to strongest, cleanest autocorrelations cycles in N.A. once the annual scale signals are filtered out. This reflects the fact that variations in N. Atlantic temp have a strong component linked to NP.

    What PDO has going for it is that, in being a difference from the mean, it removes sampling biases that are _common_ to all regions as well as any questionable bias adjustments that have been applied. It can be seen as a measure of whether the NP is out of phase with the rest of SST and by inference whether tropical pacific is in phase.

    What you note is probably the bleed-over of NP into NA . If both are affected by this cycle, they will tend to cancel when PDO is +ve, add when -ve.

  132. Greg Goodman says:

    Nino/Nina provides a mechanism getting large amounts of heat in and out of the ocean, ie global heat content.

    There’s a Mann 2000 paper on this showing linkage between volcanoes and La Nina. Probably aimed at hyping up the cooling effect so that they can put more cooling into models and boost CO2 climate sensitivity.

    However, Nino/Nina is not symmetrical since heat exchange is not the same process in each direction.

    A unified proxy for ENSO and PDO variability since 1650 : MGregor et al

    Click to access cpd-5-2177-2009-print.pdf

    Consistent with earlier studies we find that volcanic forcing
    can induce a statistically significant change in the mean state
    of ENSO in the year of the eruption and a doubling of the
    probability of an El Nino (La Nina) event occurring in the
    year of (three years after) the eruption.

    Negative feedback. El Nino dumps to heat into atmosphere to counteract v. cooling ; La Nina events restock OHC.

    Now if there is a tidal or inertial component (LOD) to La Nina, that provides a powerful hook into a ‘global warming’ mechanism. Which finally brings us back to the object of this thread.

    Projecting net gravity onto polar axis or looking for indirect planetary influence on position of moon would be worth investigating.

  133. Greg Goodman says:

    TB: “Thanks Greg. I think the oceanic oscillations are a moving target. ”

    Indeed, and I think that any speculation about planetary linkage needs to be able to deal with that. However, I think there is strong indications in the autocorrelations of NP and NA of repetitive patterns, mutual cross linkage and resonant type interference patterns.

    If you want to follow the planetary-solar z-axis idea or whatever, you’ll be met by the same problem once you try to link it to climate.

    SSN and similar propositions have always fallen down due a phase crisis at some point in the last century.

    Clear phase reversal is commonly found in climate data. NP temperature is good example. Around 1975 ,when PDO went +ve, this was because of phase flip. The usual temperature cycle did a flip and had two positive half cycles in a row. This resulted in a sudden step up. Long term Pacific temps are characterised by this step behaviour rather than a nice rounded sine wave pattern. This is one reason I don’t like PDO which rounds off these changes, which I think is masking a key climatic feature.

    People often misinterpret references to phase change in the Pacific as being “positive phase” and “negative phase” which of course is not a phase change at all, it is an amplitude feature not phase.

    Phase changes must not be swept under the carpet as an inconvenient detail, they can provide a key to detecting underlying patterns.

    It is important to note that just such phase changes happen at the crossover point in the beats phenomenon, which is illustrated in the link I gave above.

    If you are interested in following harmonic patterns in climate it is essential to understand the phase flip in interference patterns since this breaks frequency detection by simple peak identification and results in longer interval between peaks.

    For instance, if some solar (or other) driver is in a beats pattern with a natural frequency in an ocean basin, from time to time there will be phase change and the resultant climate variation will undergo a phase flip and go out of phase with the driver for a few cycles while is drifts back into phase.

    This kind of behaviour is commonly seen in climate.

    The observation that smaller SSN cycles are long one is likely another manifestation of this.