A Cycles Analysis Approach to Predicting Solar Activity

In desperation I asked Fermi whether he was not impressed by the agreement between our calculated numbers and his measured numbers. He replied, “How many arbitrary parameters did you use for your calculations?” I thought for a moment about our cut-off procedures and said, “Four.” He said, “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”.

So said Freeman Dyson, one of the brighter stars in the physics firmament. When we try to understand the star which supplies all the energy Earth needs to maintain life on its surface, we have a problem. The problem is it keeps surprising us with its unpredictable changes in activity levels. It’s doing this at the moment in solar cycle 24. The top solar physics institutions made various predictions about the timing of the start of the cycle, and how high the ampliitude would be. They’ve been proved very wrong. A couple of individual solar physicists did predict a low solar cycle (low for the modern era anyway), including Leif Svalgaard, who predicted a max monthly sunspot number of around 70 for this cycle back in 2004. Leif based his prediction on phenomenological observation of the solar polar field strength, which has been diminishing for a long time now. Even this is looking a bit high at the moment though. I predicted 35-50 SSN in 2008 based on a planetary method.

In this post, we take a cycles analysis approach to looking at solar activity since 1600. Regular contributor Tim Channon has done work in the past in acoustics, and as part of his toolbox, he developed some software which uses clever techniques to break down complex sound envelopes into underlying frequencies and amplitudes. For fun and interest, we fed it with the Lean 2000 TSI (Total Solar Insolation) reconstruction data to see what would happen. The result is shown in the graph below the break.

TSI Lean 2000 reconstruction using cycles analysis.

The R^2 value for the correlation in the top graph between the Lean 2000 TSI reconstruction and the curve generated by Tim’s cycles analysis software is 0.99 – or in layman’s terms “Near enough perfect”.

The software decided to use seven cycles to get a best fit, and so following Johnny von Neumann’s metaphor, we were able not only to get Nellie to wiggle her trunk, but hold a brush with it and paint Lean’s TSI curve while doing a belly dance.

So, given the number of ‘arbitrary parameters’ to play with, this stuff is easy, right?

Well I haven’t seen anyone else manage it, so hats off to Tim Channon and his remarkable software, though Ray Tomes managed something similar with the temperature record.

Beyond that though, there is the question of whether there is any scientific value or merit in this kind of curve fitting exercise, or as Leif Svalgaard would have it, “cyclomania”. :)

It seems to me the answer to this depends on how ‘arbitrary’ the parameters really are. It has long been thought that the Sun’s activity and Earth’s climate exhibit some longer term cycles beyond the well known ~11 year Schwabe Cycle and the solar magnetic ~22 year Hale Cycle. There is also the ~80 year Gleissberg Cycle, and the ~200 year de Vries Cycle. There are also a couple more lesser known ones. The 55.15 year cycle identified by Roy Martin, and the ~110 year cycle noted recently by Roger Andrews in his SST vs Air temperature analysis. Roger notes in comments that not only is the periodicity right, but the phase is pretty close too.

Another point I want to stress here is that the cycles, which when combined reproduce the Lean TSI curve so perfectly, were not ‘picked’ by a human with pre-conceived ideas or knowledge of real natural cycles in terrestrial climate or solar activity. The software did the choosing, testing and refining all by itself.

Look again at the legend at the bottom of the lower graph, where the frequencies of the cycles Tim’s software found are enumerated:

11.09, 11.51, 57.2, 78.94, 112.5, 238.86, 426.47 years.

Most of these are close to periods or multiples thereof recognised by solar physicists, climatologists and planetary cycle periods identified by researchers on tallblokes talkshop.

Given that the Lean TSI reconstruction is no doubt imperfect, there remains the question of how a curve generated by nearby exact planetary synodic frequencies would look compared to it. This is a project I’ll be undertaking soon.

What happens when we project the synthesized curve further backwards and forwards in time?

Watch this space, it’ll be worth your time, I promise.


About tallbloke

I study climate and weather for interest, And drive at high speed just for kicks, The solar system's my playground, Sunshine and rain my fix. tallbloke.wordpress.com
This entry was posted in astronomy-Planets, astronomy-Solar, climate-Solar, cycles-Analysis, Cycles-General, cycles-Prediction. Bookmark the permalink.

3 Responses to A Cycles Analysis Approach to Predicting Solar Activity

  1. Ray Tomes says:

    I love that Fermi quote.

    Of course we must also be aware of that when analyzing cycles. Fortunately we have the Bartel’s test to determine whether we are curve fitting an elephant or finding a real underlying oscillation. In general, the longer cycles with only a few cycles present in the period being studied will not be significant. That does not mean that they are not real, just that they cannot be proved to be real in that length of time. Take the 239 year cycle identified and it will not be significant in that data. However, as you point out, this agrees with the de Vries cycle found in much longer climate series including the phase. In the article https://cyclesresearchinstitute.wordpress.com/2010/06/27/cycles-in-sunspot-number-reconstruction-for-11000-years/ I found two periods in that range, 224.4 and 207.7 years, the later one is the normal de Vries period. Of course cycles periods can be determined much more accurately with 11,000 years of data because that cycle occurs 50 times rather than twice.

    The IPCC and many climate scientists have totally ignored the fact that the twentieth century corresponds exactly to the rise of the de Vries cycle. They largely ignore the contribution of solar effects on the earth’s climate. The Sun is near enough to 100% of the energy supply to Earth, so that seems pretty foolish.

  2. J Martin says:

    “What happens when we project the synthesized curve further backwards and forwards in time?
    Watch this space, it’ll be worth your time, I promise.”

    Nearly two years now. The projection forward by eye suggests temperatures lower than the Maunder and a cold period of about 100 years.

    So when do we get to see this graph / projection extended ?

    Any projection colder than the Maunder is seriously interesting.

    • tallbloke says:

      Making a serious prediction about long term climate is a serious business, and not to be rushed. However, something worth pointing out is that we are predicting changes in solar activity here, not Earth’s climate directly. The solar activity levels relate to ocean heat content, and because that is currently considerably higher than it was during the Maunder Minimum, built up over centuries since the end of the little ice age as solar activity has increased, it will take a long time for that energy to dissipate too. So a rapid drop in solar activity levels probably won’t lead to an equally rapid drop in global temperature. However, the change in solar regime will have other regional effects on circulation which could bring about rapid change in some parts of the world.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s