Predicting changes in Solar activity

Posted: January 2, 2010 by tallbloke in solar system dynamics
Tags: , , , , , , , , ,

Here’s a prediction graph I produced a little while ago which seems to be more or less on course:



It uses the fact that changes in Earth’s length of day seem to precede changes in solar magnetism and sunspot production by several years. The yellow curve was generated by combining Sunspot data with LOD data to create a prediction for Ap out to 2015. The recent burst of sunspot activity has arrived on cue.

Here’s another graph which shows a possible correlation between sunspot activity averaged over the length of the solar cycle, and motion of the solar system’s centre of mass relative to the solar equatorial plane averaged over two Jupiter orbital periods:

Sunspots graphed against SSBz-solar equator

What caused the collapse in solar activity at the start of the 1800’s known as the Dalton minimum? Could it be the conjunction of Uranus and Neptune which seems to accompany each of the grand minima? Does that mean we are due another one now?  I’ll investigate that in another post soon.

Why does the average sunspot number fall when the average mass of the planets is heading south? Speculatively,  could it be that the ‘lensing’ of an electro-magnetic effect emanating from the galactic centre diminishes when the planets are ‘on the wrong side of the sun’?

Answers on a postcard, or post your thoughts below.

  1. Daryl M says:

    tallbloke, since the LOD is unlikely to be causing changes in the sun, there must be something that is causing both.

    Some time ago, Ian Wilson released his “chicken and egg” presentation which took things a step further making the case that changes in LOD could be causing fluctuations in the earth’s climate by affecting ocean currents. I corresponded with Dr. Wilson several times and he even sent me his paper prior to it being published. His theory is very compelling but it does not seem to be getting any traction. Do you correspond with him? If so, I would be very interested to hear his thoughts on this.

  2. tallbloke says:

    Hi Daryl, yes, Ian sent me a copy of the pre-print too, a fascinating paper. Since variation in length of day on Earth (LOD) tracks the changes in the motion of the sun relative to the solar system barycentre, but around 30 years behind, we can be sure it doesn’t cause changes in the sun. I think the ‘something else’ is the motion of the other planets, time will tell.

    A couple of the other posts on this site link changes in LOD with other Earth phenomena too such as the rate of change of the magnetic north pole, and changes in global temperature, have a look around. If these links are real, we have a lot to learn about how the solar system ticks.

    I think Ian’s paper will only gain traction through the interest of people like yourself. If enough of us discuss these ideas, more qualified scientists will investigate. I’m hoping my second graph on this post will help spark more interest, as it potentially links sunspot numbers to the motion too.

  3. Semi says:


    Regarding the LOD and SSB-motion relation and relation to the AP index, mentioned by you, I was little suspicious of detrending and time-shifting the LOD, but today I found it is probably a common practice (seeing it in mr. Vaughans work) – it is probably possible, that the effects lag the cause, even by more years? (The AP index is a measure of Earth magnetism, Solar activity or both??)

    It is not completelly impossible, that the Earth rotation does cause the Sun’s activity changes (as it precedes it’s changes?) – if the Earth’s influence onto Sun was magnetical, the change in Earth’s rotation could cause change in Earth’s magnetism (?) and thereby change the Earth’s influence onto the Sun?

    I failed to reproduce your SSB z-axis chart – did you really mean to rotate SSB to a reference frame of Solar equatorial plane and taking the Z component? See my version here , possibly inverted over the axis? (Computed as “vz(veclrotate(sun))” with SolarEquatorial “ecliptic” selected in my EphView program, in barycentric coordinates. I just hope there is not an error in the SolarEquatorial plane definition in my program? But I tried to verify it visually in the Tidal viewer window in my program to match the position of Solar poles from Earthly point of view earlier and now to verify the SSB position relative to the Solar equator with the unsmoothed chart…)

    So, what then really is the bold-blue line on the second chart, as it’s relation to the red-line seems obvious and interesting?

  4. Semi says:

    Sorry, the link to my picture should be here

  5. tallbloke says:

    Hi Semi,
    I can see your curve is produced from monthly JPL data.
    I downloaded two datasets. Monthly from September 1782AD and yearly from 800AD. The graph in my post is the annual data averaged over 24 years and inverted

    Here is another graph with my monthly data of z-axis distance of barycentre to solar equatorial plane, smoothed over 24 years, (inverted) and overlaid over your (inverted) graph. The sunspot curve runs from 1749.71AD.

    The agreement between the z-axis curves is good, but 1782.75AD on my graph seems to match with around 1792 on your graph, a ten year difference. I think the calendar dates were pasted into my spreadsheet with the x,y,z barycentric distance data all at the same time from the JPL download. I do have memory and other cognition problems after a serious accident three years ago, so maybe I messed something up…

    In the second graph on my post I used the annual z-axis data smoothed over 24 years. I’m not sure why the curve looks so different to the monthly data smoothed over the same period, maybe someone who understands statistics better than me can help me there.

    I agree it is strange that the sunspots seem to be happening before the matching changes in z-axis motion, or is it possible they are lagging behind by a long time? The repeating cycle of z-axis motion over ~170 years is a little self similar. Maybe the motion of the outer planets has a deep effect on the sun which comes to the surface many years later and then the sunspot cycle is modulated by the electromagnetic connection between Earth and sun in the present day? I don’t know, this is only speculation on my part.

    Here is another graph like the one in my post but also with the sunspot data smoothed over 12 months in addition to the original red curve.

    The match between the amplitudes of the solar cycles and the averaged changes in the z-axis motion seem too good to be explained by a long lag of sunspot activity behind an earlier z-axis cycle, but it doesn’t seem possible that sunspot activity drives the motion of the planets unless we live in a universe much stranger than we realised!

    So what is going on? I would really like an answer to this puzzle. 🙂

  6. Semi says:

    Well, I’m not using smoothed datasets, either monthly or yearly, but calculated it from the DE406 ephemerides using my program in 21-day steps (but in 1-day steps the curve is almost identical).

    Care must be taken when using discrete values, picked over longer intervals – say that you take 1 value every January 1st and assume, that this value is valid for the next year – then it depends, if just the January 1st was deep in the trough or some way beside – this can create false frequencies/trends. (I’d recomend taking at most monthly values and averaging them into yearly values or otherwise smoothing and using these, instead of one value for January 1st. Surely the sampling frequency needs to be considerably shorter than half of the frequency of the evaluated value, see “Nyquist-Shannon_sampling_theorem” in wiki). (Same would apply for Sunspots, you would not pick one daily value and assume it is valid for a whole year, but rather taking already smoothed or unsmoothed data at good precision and averaging…)

    But I’ve tried to simulate this, taking 1 value every January 1st, and it does not differ much from the “monthly” or 21-day values (the function is essentially 12-year, so 1-year sampling should be good enough), so I’m not sure, why does it differ? – Now I see, your yearly chart (bold blue line) seems to be time-shifted by some 20-40 years to the left? (That’s why I failed to recognize it…)

    Otherwise, the sunspots ultimately may not lag 179-40=139 years, and they do not influence planetary motions (calculated solely from gravitational attractions), at least not very much. I’ve already heared of effects, that may lag 1 or possibly 2 cycles – by influencing the transport of magnetic field toward poles in current cycle, it may affect the next cycle strength. This is studied by K.Georgieva, but she argues that this influence to yet another cycle (ie. over 2 cycles) is negligible…(?)

    Anyhow – here the Effect (Sunspots) precedes the Cause (SSB motion), so it does not seem very probable…?

  7. tallbloke says:

    “Now I see, your yearly chart (bold blue line) seems to be time-shifted by some 20-40 years to the left? (That’s why I failed to recognize it…)”

    yes, I said in the other thread it’s is shifted by the Hale cycle, but it is nearer two Hale cycles. This was an error on my part. I agree the causation problem is a conundrum, but it is still striking. After the Dalton minimum to the present, the R^2 value is around 0.8 when the sunspot data and z-axis data is smoothed over 23.25 years.

  8. Paul Vaughan says:

    A few notes in response to what has been posted:
    1) 1 year sampling is ‘good enough’ (for some purposes) for the low-frequency of the jovians (but certainly not for the inner planets).
    2) Please see my notes on lags & alternate (to z) axes in other threads. [Misinterpretation of lags is very common due to deficiencies in our education system.]

    tallbloke, you should be able to find another axis-orientation that will eliminate the need to use a lag to achieve match (for a portion of the record). I offer no comment on mechanisms; my comments here are based ONLY on abstract consideration of harmonics, smoothing, & rotation.

  9. tallbloke says:

    Well that gives me hope. I did geta brain mix up over the direction of the lag – mea culpa. I’ll have to start afresh with newly downloaded data to ensure I have it right from the start. Your comment about axis orientation is very interesting, but you’ll have to give me more clues…

    One thing I gleaned from today’s reading is that the sun’s tilt angle with respect to the invariant plane changes, as does it’s precession, though I don’t know by how much or over what timescales yet. More research needed…

  10. Paul Vaughan says:

    I recommend obtaining monthly SSB(x,y,z) data from Horizons. To do spatial rotations, just use trig functions. You may benefit from reviewing a wiki page on transformations from Cartesian to polar coordinates for starters. You can set up “spinner” buttons in Excel to rotate view. Everything you need to know is on wiki &/or in Excel’s help files. It may take a while to put the pieces together.
    Bear in mind that the axis you seek may not have a static orientation (but I can already see [from graphs upthread] that you should be able to find a *partial match with a *static axis).

  11. tallbloke says:

    Got it, thanks. I already did something similar to set up a solar cycles emulation which enabled me to change the phase angle.

  12. Paul Vaughan says:

    Once your results start pouring in, you’ll no doubt see why I’m always going on about thick confounding of indices of SSD. It all comes down to this: Lots of symmetry in the SS. Bear in mind that there are different types of symmetry: translation, rotation, reflection. Try to take note of when time can substitute for space due to stationary periodicity.