Prediction of the Strength and Timing of Sunspot Cycle 25 Reveal Decadal-scale Space Environmental Conditions

Prediction of the Strength and Timing of Sunspot Cycle 25 Reveal Decadal-scale Space Environmental Conditions

Prantika Bhowmik1 and Dibyendu Nandy1,2,*
1Center of Excellence in Space Sciences India, Indian Institute of Science Education and Research Kolkata, Mohanpur 741246, West Bengal, India 2Department of Physical Sciences, Indian Institute of Science Education and Research Kolkata, Mohanpur 741246, West Bengal, India *Corresponding author:

The Sun’s activity cycle governs the radiation, particle and magnetic flux in the heliosphere creating hazardous space weather. Decadal-scale variations define space climate and force the Earth’s atmosphere. However, predicting the solar cycle is challenging. Current understanding indicates a short window for prediction best achieved at previous cycle minima. Utilizing magnetic field evolution models for the Sun’s surface and interior we perform the first century-scale, data-driven simulations of solar activity and present a scheme for extending the prediction window to a decade. Our ensemble forecast indicates cycle 25 would be similar or slightly stronger than the current cycle and peak around 2024. Sunspot cycle 25 may thus reverse the substantial weakening trend in solar activity which has led to speculation of an imminent Maunder-like grand minimum and cooling global climate. Our simulations demonstrate fluctuation in the tilt angle distribution of sunspots is the dominant mechanism responsible for solar cycle variability.

Full paper here

HT/Leif Svalgaard

88 thoughts on “Prediction of the Strength and Timing of Sunspot Cycle 25 Reveal Decadal-scale Space Environmental Conditions

    • Our ensemble forecast indicates cycle 25 would be similar or slightly stronger than the current cycle and peak around 2024.

      No shit Sherlock. I have no super computer models but could make an equally vague prediction off the top of my head from following discussions on WUWT for the last 10 years.

      Doubtless Dr David Evans will be along with his “notch filter” to annouce the imminent collapse of solar activity. 😉

  1. Hmmm. So, other than it’s fun, what is trying to predict the solar cycles 10 years in advance going to tell us? I get that it can “suggest” the strength of upcoming solar storms years in advance, but unless it can predict they are aimed at Earth (or a interplanetary space craft), I don’t see the value. Is it just to show that we are starting to understand the physics of the Sun, or is it deeper?

    Here is another thought… We know that the Sun’s particle streams intensity vary over time. We also know that the Earth’s magnetic field flips now and then. If the Sun were to point a really intense flare at us during an energetic cycle, while the Earth were in the midst of flipping its magnetic field, what would be the likely result? Would there be some kind of evidence in the paleontological or geological records?

    • what is trying to predict the solar cycles 10 years in advance going to tell us?
      People that are launching satellites need to know what solar activity will be up to ten years in advance.
      To know what danger the solar storms will be to their product. The insurance premiums depend a lot on that.

      • If I understand correctly, this is the first century-scale model of the Sun. Let’s hope that it turns out better than climate models. Magneto-hydrodynamics is a difficult beast.

      • Interesting. How should Acuaries judge the veracity of speculative solar activity scholarly papers?

        I mean to ask, (for example) will premiums change based on this paper?

        • Yes, ostensibly. That is to say premiums will rise, period. They inexorably do. If pressed as to why, actuaries can now point to this paper.

      • But given the low accuracy of these predictions, it seems you will either proceed to cautiously (if you believed the forecasts of this last cycle) or too recklessly.

        I realize planners want and will pay for foreknowledge, but it doesn’t seem like a product that’s fully baked.

    • Given the behavior of the Earth’s magnetic field during what most scientists believe is the beginnings of a magnetic reversal.
      It appears that the Earth’s magnetic field never goes to zero.
      Instead it first weakens a bit, then anamolous magnetic regions start to show up. IE areas where the magnetic field is reversed from what appears for most of the planet. Over time, the anamolous regions grow in number and size until eventually until planet has the new magnetic alignment.

      Can you imagine the chaos had this reversal started a few hundred years ago during the age of exploration, when magnetic compasses were still the dominant form of guidance for explorers.
      Would it have slowed down the age of exploration? Would the inability to trust compasses put a damper on other forms of scientific research?

        • Actually, we are where we are right now. However, had the compasses been wrong, we might not be here, at all.

      • MakW,
        Long ago mapped out mentally the bones of a novel about a commercial airliner affected, as were some nearby others, by a reversal of Earth polarity over the course of an hour for the small region where the aircraft were flying.
        Scrapped the idea because Leslie Nielsen beat me to the post with some incomparable lines.

      • MF will not disapper or fall greatly from its present value, magnetic poles may move further equatorialy, generating a quodropolar or even multi polar field, followed by a short ‘violent’ swing in polar directions. Magnetosphere will persist but its form may be substantially different during the perturbations, but all that is many centuries or millena away.

    • Simulations, of the available data, when will we get some real data not made up stuff.
      You can’t have real data before they happen. All predictions are based on some form of simulation or model.

      • Exactly the why all predictions are only educated guesses and only that. A model may are may not be correct, again it only a guess as to what might happen. Remember this as quoted by Enrico Fermi “I remember my friend Johnny von Neumann used to say, ‘with four parameters I can fit an elephant and with five I can make him wiggle his trunk.’

      • As opposed to….. a WAG (wild ass guess)
        Apparently there are also SWAG’s (scientific wild ass guesses).

        • They are not WILD guesses, but based on observed data and well-founded physics. The main uncertainty comes from the very difficult measurements of the flows of plasma in the interior of the sun. With time these will improve, so the predictions will improve too. We already had successful predictions of the last several cycles. Each new success will bolster our confidence. Each failure will teach us how to do better next time.

          • Very good, Leif. Science progresses one mistake at a time.\

            I appreciate that you don’ t make wild guesses and help us undereducate folk get some sense of what is happening.

          • Thanks Leif, what if they are not cyclical and just leftover harmonic resonations from an earlier perturbation ?

          • Leif,
            “We already had successful predictions of the last several cycles”
            For this to be true, the theory should be several cycles old. Is that the case? Forecasting is not the same thing as hindcasting. Getting a theory that fits past observations is not really predicting.

          • You should not just assume things without checking first.
            I don’t, that’s why I asked. I asked because I have seen papers that claimed to have some new predictive technique that worked with past events, although I don’t remember if it was about earthquakes or about volcanic eruptions. So while I could assume that things get done right, the realitiy is that some people don’t do things right.

          • So while I could assume that things get done right, the reality is that some people don’t do things right.
            You can’t just generalize ‘some’ to ‘all’. The default should be that specific people did things right, unless you have evidence to the contrary.

  2. Quickly read the model description. It looks like they ran the simulation multiple times and picked the one closest to the actuals as their model.

    Hopefully I misread this because it would appear to introduce the possibility of spurious correlation via “Selecting on the dependent variable”.

    This was the problem with tree ring calibration. How can you distinguish between a model that has predictive value and one that simply guesses the past by chance.

  3. Indeed, H/T to Leif Svalgaard for enduring us. Thank you.

    I appreciate Nassim Nicholas Taleb on prognostication and skin-in-the-game doxastic comittment. Prognosticators need the Sword of Damocles’s truth hanging over their head. Scepticism is the chastity of the mind. George Santayana

  4. We select the simulation for
    which the correlation between the
    observed and simulated polarflux is maximum as our calibrated, standard
    This reads to me like tree fing calibration. You because one simulation/tree tracks the past doesn’t means it can predict. The correlation may simply due to chance over a large number of samples.

    • I am the second (and senior) author of this study. First of all, thanks to everyone for your our work. Let me respond to this comment by Ferdberple to elucidate why it is important to calibrate a simulation or observation, and why we believe our method is robust.

      First a model generates data which is not absolute, neither quantitatively equal to what is observed, especially when a proxy is being used for some physical quantity. Note that polar field or polar magnetic flux observations exist only over the last four decades. So we used an observed proxy for polar field, namely polar faculae, for the long term comparison. The polar favulae was already calibrated with polar flux observations so that it could be considered as an extension of polar flux observation over the last century.

      Our polar field simulations produce simulated flux based on integrating forward in time a differential equation representing the surface field evolution; such equations always need an initial condition and of course we did not know what that initial condition was too accurately (look at the increasing error bars as we go back in time in the observed polar field in Figure 2 of the paper). So we did what is the standard technique of performing multiple (ensemble) runs with slightly different initial conditions.

      Within a close range of initial conditions many simulation runs would show virtually similar results with small differences that a visual inspection (chi by eye!) cannot differentiate. So then, if Tom, Dick and Harry pick up what they think would be the best initial condition based on visual perception, then reproducibility is lost and the choice is not quantitative! Our choice is quantitatively stated, makes the most sense for polar field calibration. Furthermore we do not vary any model parameter for the duration of the runs, so the sanctity of the simulation is maintained. Now this is only half the story.

      Next we feed the polar field extracted from the surface magnetic field evolution model at every cycle minima to a dynamo model for magnetic field evolution in the Sun’s interior. It is this model that creates the sunspot cycle. Again other than this data assimilation from the other model to the dynamo model, no parameter in the century scale dynamo run was fine tuned over the period of simulation. It is the dynamo model that really does the prediction for the sunspot cycle based on observational data driven inputs from the surface flux transport model. Figure 4 from our paper, which has been posted here is the output of that dynamo model.

      Summary: Calibration with observations is important in data driven simulations. Idea of the surface model was the use observed data to simulate past polar magnetic field accurately. Then the dimulated polar field was fed in to a *different* model for the Sun’s interior to generate the century scale sunspot cycle simulation and prediction.

      Hope this helps.

      [Thank you for your time, your efforts in replying here. .mod]

      • Hi,
        Thank you for additional comment.
        Your estimate is in line with Dr. Svalgaard’s prediction, but I assume you are not contemplating possible amplitudes of the following two cycles.
        On the other hand, I think that the sun has ‘ejected’ such amount of energy during the last 100 or so years that it might take at least two or more cycles for another burst to perchlorate from the depths of the interior to the surface to be visible in a way of sunspots numbers as seen in SC24 or 23.
        I’ll put my head on the block and suggest that neither SC25 or 26 will surpass the SC 24’s amplitude, hopefully the the sun will not slide into Maunder minimum type of interlude, else we may see possible rise not before SC27.

      • Within a close range of initial conditions many simulation runs would show virtually similar results with small differences that a visual inspection (chi by eye!) cannot differentiate.
        Thank you Dibyendu for the explanation. I haven’t studied the report in detail so I may well be completely wrong in my assessment.

        My concern would be sensitivity to initial conditions. As I understand the problem you are trying to recreate the initial conditions. This is so your 2nd model will deliver consistent results. As such, you go through a process to locate the initial conditions that most closely match observations.

        However, as you note, many sets of initial conditions return very similar answers. So there isn’t very much to say one set is “best”. Even very small errors in the historical data could change your calculated initial conditions.

        So by this it must be very likely that the initial conditions you have selected are not “correct”. Rather they are the result of chance operating on errors in the observations.

        However, this may not matter if your model is insensitive to initial conditions. If your model generates the “same predictions” with only minor differences regardless of initial conditions then it is likely robust in this regard.

        The problem is that your initial conditions are likely to be wrong simply due to chance and the imperfections in the historical data. The question is whether the predictions are sensitive to initial conditions.

        Treating the initial conditions as a probability, rather than a unique value, might satisfy this question by revealing whether the prediction is sensitive to the initial conditions.

        For example, re-run the model with the lowest and highest initial conditions reasonably possible and check the spread in the predictions. If the spread is wide, then the prediction may not be robust.

        If the spread is narrow then I believe you may well take some confidence in the result. (Excluding potential errors elsewhere not considered.) Again, I may be completely wrong as I have not studied the question in any detail.

      • “Idea of the surface model was the use observed data to simulate past polar magnetic field accurately.”
        Just like a Stockmarket or a greyhound betting programme.
        Past results are no guarantee of future outcomes.
        Although they are all we have.
        You have put in conditions that lead to some expectation of a future result.
        You have reasons why it might work. Retrofitting removes those reasons and subjects you to a poorer choice through accidental past fitting.
        Better to take the median of your expected ranges.
        This might be the same by accident of course.

        • Just like a Stockmarket or a greyhound betting programme.
          Past results are no guarantee of future outcomes.

          You clearly have not understood the process.
          Let me give you an example of how it works:
          Imagine you have an automobile. You put a certain amount of fuel in the tank and drive until you are out of gas and note down the distance traveled. You do this for several different amount of fuel. That gives you a calibration curve, that allows you to deduce [predict] how far the automobile will travel for a given amount of fuel.
          The solar cycle works in a similar way. The sunspots are generated from the magnetic fields in the polar regions of the sun with a lag of several years. Using measurements of the polar fields for past cycles we can calibrate the cycle for given amounts of polar field. Then given a measured polar field we can (just as for the car) deduce [predict] the number of sunspots several years later.
          Now, the polar fields are produced by magnetic flux from decaying sunspots being carried to the poles by a circulation of plasma from where the sunspots died to the polar regions. That magnetic field is measured [both from the ground observatories and from spacecraft instruments] and can be followed as it flows towards the poles. From those measurements and from measurements of the plasma flow the paper predicts how much magnetic flux eventually reaches the polar regions and so predicts the polar fields, and hence the coming sunspot numbers.
          All of this is straightforward and simple and well-understood physics. Not at all like ‘predicting’ the stock market.

          • The assumption is that the sunspots represent a predictable amount of magnetic energy and that the fluxes in the sun can be correctly measured to build up an idea of where the sun and sunspots will be going in future.
            Now there is a repetitive cycle, unlike say El Niño/ La Niña where random, to us, extra conditions means the cycle appears random.
            But the same inner turbulence of the sun means that the energy does vary more randomly than you assert though on a more stable cyclical heartbeat.
            Hence solar flares for instance I presume are basically unpredictable at least over a month in advance.
            The same unpredictability of energy build up means that an over exuberance or under exuberance, a run of cold cycles or a run of hot cycles can happen purely by chance despite your best efforts.
            This is not to say that your predictions for a couple of years in advance of sunspot number is wrong. You obviously have a much better chance of short term prediction accuracy due to your study of the “form” than most people here will ever have. Hubris compels me to caution you on anything outside this short window. I hope I can look at a blog of yours here or elsewhere in 10 years and have you confirm that you are indeed right.
            On the other hand I want a cool sun and cool extended cycle to make climate scientists come to the realisation that they have far more factors and complications to consider.
            Thank you for all your work and go easy on the opposition as it is Xmas.

          • Now there is a repetitive cycle, unlike say El Niño/ La Niña where random, to us, extra conditions means the cycle appears random.
            There is, indeed, an element of randomness in the ‘cycle’. The sunspot cycle is not a repeating cyclic phenomenon, but rather an eruption of activity. We do not rely on the assumption that the activity is cyclic, but treat each eruption as a unique event. The properties of the event are largely fixed by the amount of observable magnetic flux left over from the previous eruption which gives us the predictive power. Because of the random element we can only reliably predict one cycle ahead. There is no hubris here nor ‘cyclomania’, just application of known physics and direct [but difficult] measurements.

          • Thank you Leif. This is a brilliant elucidation of the idea of calibration in a language that would hopefully be understood by most in this community.


      • Dibyendu

        Thanks for taking the time. I was surprised to read that you view model outputs as “data”. At the end you refer to “data driven simulations”. This adds more confusion in my mind. To me, data is a measurement or is perhaps calculated from a measurement. A data driven model is one that takes data and processes it in some standard manner to generate an output or projection; a prediction, if the model has been validated.

        The output from one model, being simulations, goes into another model, if I read your message correctly. That is not “data driven” in my arena.

        Surely there is some data involved in your simulations. I read that you have some years of observations, then a model that projects backwards for a century to give modelled outputs.

        Sorry to be pedantic about this but in the field of climate science there are many things claimed to be “data” which are guesses, and often bad guesses.

        The WHO is on a new jag about “air pollution” and “deaths” with turns out to be a stack of no less than six models and numerous heroic assumptions, with each model’s output assumed by the next to be “data”. It is puffery, and puffery should never be attached to your good works.


        • Hi Crispin,

          Thank you for your interest in our work.

          Well, at least in astrophysics, modelers also call simulated output data or simulation data, so we dealienate observations by explicitly referring to them as observations or observational data.

          We use two models. One for magnetic field evolution on the Sun’s surface (say model 1) to simulate the Sun’s polar field (hard to observe and with direct observations only for last 40 years or so). This model is data driven, by assimilating observed sunspot emergence for the last century. The second dynamo model (say, model 2) is for magnetic field evolution in the Sun’s interior driven by internal plasma flows. Model (2) is driven by using inputs from observational data driven model (1) to predict the sunspot cycle.

          So yes, our methodology is such that it starts with observations, goes through intermediary steps of computational modeling, and finally generates the sunspot cycle prediction.


  5. There are many things to be skeptical of. I’m not a fan of the “Most Scientists” meme, as it is often misused by headline seekers.

  6. The closest many people come to a simulation exercise is when they hire a financial planner. Your great ideas about stock, bonds, real estate, and works of art can be examined.
    The big issue is — will you out live your money.

    What is Monte Carlo Simulation?
    Monte Carlo simulation is a computerized mathematical technique that allows people to account for risk in quantitative analysis and decision making. The technique is used by professionals in such widely disparate fields as finance, project management, energy, manufacturing, engineering, research and development, insurance, oil & gas, transportation, and the environment.

  7. If I remember correctly, initial predictions said solar cycle 24 was going to be very active. That was prior to Cycle 24 starting. It turned out to be about as wrong as it possibly could be. I guess we shall see.

  8. We wouldn’t become Paris. We’d become Carthage, at our own hands. At least, parts of the map would.
    The fed- up factor in Paris is a candle, lit from modern fervor to cast out the usurpers of Liberty.
    The world’s latest revolution is rooted in information, available to any who seek it, as never before.

    On these shores, our pilot light is lit and the gas is full on.
    Some have visibly ramped up their efforts to put out that ubiquitous, hidden flame.
    Here’s hoping they are defeated before this world is set ablaze.

    • Mods!

      I do not know how my comment came to be posted in this thread.
      It was not aimed here and does not belong here, being also irrelevant to the topic at hand.
      Please delete my immediately previous comment, if you can.

      [Actually, the mods were thinking that reply could be related properly to any number of threads… .mod]

      • I had several thread tabs open at once and thought I was posting in another thread, but accidentally posted here, instead. By my own hand, sure.

  9. Timing of next solar grand minimum is the critical factor of the natural variability that governs the North Hemisphere’s temperature trends. However there is an elephant in the room, nothing to do with the CO2, in form of a undercurrent periodicity which is due to peak in about 60 to 70 years. Since the ~60 year AMO periodicity characteristic for the NH is at its peak the next 30 years will lead to some cooling but the exetent of it depends on if and when the next solar grand minimum occus. In this link
    I looked at possible alternatives:
    – If the grand minimum doesn’t start during next 4-5 cycles than cooling will be minimal, about 0.2- 0.3 degrees (faint blue line)
    – Since grand minima tend to last (intriguingly?) around 60 years, if a GM is about to start with SC25 then it’s greatest effect might coincide with the AMO just lifting of the floor with total fall of about 0.7- 0.8, since the cumulative solar GM cooling is about 0.5C (dark green line).
    In about 60+ years time the AMO will be hitting next peak and any pending solar GM may well be over, resulting in the temperature rise of about 0.5C on the current level. It could be expected that the NH’s temperature gets up to about 1.5C on preindustrial levels. From there on is all way down hill.

  10. It should be a good paper since it is from the “Center of Excellence in Space Sciences ” :))

  11. Dibyendu Nandi: I’m wading through your extremely well written and readable paper. A bit slowly as my physics and suffered from considerable oxidation over the past six decades. What are the fundamental differences between your team’s approach and that of Valentina Zharkova? From a distance to sounds like both teams are looking at the magnetohydrodynamics of the sun’s interior but with different dynamo models.

    • Science is a belief in the ignorance of the experts.
      – Richard Feynman

      I’m with Feynman on SC25. We know too little to expect the predictions to have skill.
      And in this big world, there will always be some expert’s prediction that will be correct in one instance. Too often we mistake that one-off luck to be skill. So who is the fool?

  12. I will take this prediction with a large grain of salt. SS24 was supposed to one of the strongest ever. And they kept revising it downward and downward until it is one of the weakest in over a 100 years. Something tells me that they are still a long way away from being able to get even close to what is going to happen. About the only natural events that we can seem to predict accurately beyond a month or so are the movement of the planets.

      • Sorry doc, I put your in 2005 (see my comment further above); as a matter of record I bit you by nine months, but what is few months between friends. In contrast to your work at the time I hardly knew what I was doing, not that much has changed since, so just consider it a confidence. I’ve put my marker on SC 25 to be lower than SC24 , you said it is going to be higher, both can’t be correct this time. I’ll raise a class of a Napa Valley red whatever comes about.

        • so just consider it a confidence
          With a coincidence timing doesn’t matter.
          What is of more interest is the prediction for SC25 [depending on the polar fields].
          Your prediction for the polar fields for 2018 is wildly wrong [less than half of what is observed]:

          Not to speak about the failings around 1900 AD and 1780 AD.
          Looks to me that your ideas have been strongly falsified.

        • No solar cycle 24 was predicted to be one of the strongest cycles ever.
          It was soon realised that that old prediction was wrong and that a low cycle was a better deal.
          If you don’t know this, it is time to educate yourself a bit.

          • The original prediction is what counts
            Nonsense. The latest one [with newest and best data] before the minimum is what counts.
            Especially if based on correct physics.

          • I had the prediction of solar cycle 24 up on the wall. It was very well suppose to be stronger than 23. It was the year of quite that took everybody by surprise. And it wasn’t until the end of the quite did I see new predictions that were lower than 23.
            I for one think global cooling is a disaster and global warming a benefit.
            I hope you are right Lief on the upcoming solar cycle. It will be interesting. We will see.
            I did look at all the available work you’ve done.
            If you are right, the increased confidence level will be very high for cycle 26.

          • And it wasn’t until the end of the quite did I see new predictions that were lower than 23
            Our prediction of a low SC24 was made in 2004 and published in 2005, well before the end of SC23.
            And you are right: if we predict SC25 correctly, there need not be a big debate about SC26; just observe the polar fields and you have it. That is, of course, a big ‘if’.

  13. Only three divergences in 240 years, not bad for a bloke who had no idea what he was doing. Remember I told you at the time (some 10+ years ago) all I knew about the solar activity I learned from my daughter’s homework.
    I wonder if anyone else is around from the good old SC24 blog.
    btw. sooner or later the Babcock–Leighton solar dynamo cycle is going to get a puncture, the next one just might do the job; any idea about what happened to ‘many peaks in SC24’?

  14. Only three divergences in 240 years
    No, much worse than that. The 18th century was completely wrong (phase-wise). The 1900, 1780, [and 2025s] were about the failures to reverse the polar fields.
    The many peaks in some cycles are due to the intermittent explosions of the solar eruptions. Quite normal for such a process.

    sooner or later the Babcock–Leighton solar dynamo cycle is going to get a puncture
    No, the BL-mechanism is based on physics and physics does not ‘puncture’.

    • So far the ‘Quiet Sun’ and its predictions preceding are working out for those of us following events. Like P. Corbyn and this paper’s authors etc.. However, now it has started, I’ll just follow the data as it reveals itself….. Brett

  15. My empirical model shows SC25 maximum occurring in 2025 and with a likely second peak in 2027. And that from SC26 onward will see the return of larger sunspot cycles until the next centennial minimum from the late 2090’s, which is a majorly long one, as is the following one from around 2200.

    Roughly agreeing with the FFT element of Steinhilber 2013:

  16. A related paper is of interest:
    The polar magnetic field of the Sun is a manifestation of certain aspects of the dynamo
    process and is a good precursor for predicting a sunspot cycle before its onset. Although actual
    synoptic measurements of this field exist only from the mid-1970s, it has now been possible to
    determine its evolution from the beginning of the twentieth century with the help of various
    proxies. The recently developed 3D kinematic dynamo model can study the build-up of the Sun’s
    polar magnetic field more realistically than the earlier surface flux transport model.

Comments are closed.