This is based on the data provided from the “Solar Cycle Progression” web page, “Provided by the NOAA/Space Weather Prediction Center” except substantial data post processing has been done.
The datasets have been heavily normalised but the F10.7 earth distance problem (or something of a similar origin) has been partially compensated before normalisation.
Normalising brought the three datasets closer to the same.
Two sunspot data r2 > 0.98, SWO/F10.7 r2 > 0.93
I also produced a simple 3 term harmonic simulation and allowed it to extrapolate to 2020 producing an unexpected result. This was based on normalised F10.7 data but the extrapolation matches the sunspot forecast, not the F10.7 forecast.
A more detailed simulation is far more complicated and I don’t think is useful, far too much uncertainty exists.
Result
An extrapolation based on F10.7 data agrees with the official extrapolation for conventional solar forecasting but disagrees with the official F10.7 extrapolation.
A lot of people are watching and waiting to see where the solar cycle is going. Prediction is best done after the event.
F10.7 and earth problem
F10.7 data contains an artefact at 1 year, no sign of 0.5 year. The removal method chosen in view has significance. I could use a novel facility here but on investigation the problem looks to be a variation on the affliction in Wilcox Solar Observatory solar polar magnetic field data. In that case a small area of each solar pole is the target but through the year the sun in effect bobs up and down in the sky. This produces a clear doublet (split peak).
The commonality is signal modulation by the major signal excursion.
I don’t know the full story.
Post by Tim
(article was written some time ago, still valid)
Tallbloke's Talkshop 
“Prediction is best done after the event.”.
Classic !
Certainly the sceptics and lukewarmers are looking forward to 2020 or so, the warmists must be increasingly dreading it.
Not a Dalton, nowhere close to a Maunder. A precursor to a Dalton? A precursor to a precursor to a Dalton?
Hah!
Predictions are where all the fun is.
Tim, is the F10.7 data normalised for 1AU like TSI is, hence a 1 year artefact in the raw data due to orbital eccentricity? I’m more interested in TSI but I’m interested to learn how the F10.7 amplitude varies with orbital radius if at all.
As I recall, was a month or so ago, I played around to get a feel then normalised to for standard deviation = 1, skew and so on, both to the same figure. As much as anything I wanted to see what it looked like. (what if etc.) Nothing can be accurate with data like that.
I already knew the F10.7 has oddness about annual. Wondering whether this was the 6 month matter looked for that, nothing. Looking at the annual no split peak either.
Bit of a mystery finding neither. If it is the 1AU matter it would have a 6 month component.
Decided to mostly remove it anyway.
Things would have remained there, insufficient interest so no post but a casual model of it brought up what I think is good enough. Two forecasts ought to be the same.
The polar magnetic field is probably of more interest. All of a watching paint drying nature.
I saw a graphic based on accumulating the sunspot number over a long period as a proxy for accumulated net energy gain/loss.
I tried to do this with the adjusted data recently provided by the group led by Dr. Svalgaard but did not have much confidence that my processing did justice to the new data, based as it was on eyeballing a graphic.
Can anyone give a reference to a dataset with the adjusted numbers?
FC: http://www.leif.org/research/Revised-Group-Numbers.xls and http://www.leif.org/research/Revisiting-the-Sunspot-Number.pdf
All this is way above my pay scale. Happened to read, once again, the earth is about 6% closer to the sun on January 4th than July 5th. This is during Southern Hemisphere summer yet Antarctic ice is not reaching the normal summer minimums for many years now. Makes one wonder.
Peaks in the F10.7 spectrogram at 0.8 1.6 and 3.2yr look like they could be E-V connected. The remaining peaks at 1.9yr (Mars orbital period), and 2.37yr (same period as QBO) are more enigmatic. The small peak in Tim’s altered data around 1.1yr would be near the E-J conjunction period.
There’s a highly detailed look at solar output periodicity in Scafetta and Willson’s PRP paper
Click to access prp-1-123-2013.pdf
As for the difficulties of prediction – doubly difficult when the Sun is perturbed from its ‘normal’ (or more usual) rhythm as it is now.
tallbloke says: September 22, 2014 at 10:57 am
“Peaks in the F10.7 spectrogram at 0.8 1.6 and 3.2yr look like they could be E-V connected. The remaining peaks at 1.9yr (Mars orbital period), and 2.37yr (same period as QBO) are more enigmatic. The small peak in Tim’s altered data around 1.1yr would be near the E-J conjunction period.
There’s a highly detailed look at solar output periodicity in Scafetta and Willson’s PRP paper
Click to access prp-1-123-2013.pdf
As for the difficulties of prediction – doubly difficult when the Sun is perturbed from its ‘normal’ (or more usual) rhythm as it is now.”
Interesting, can earthlings ever learn of what “is” before “is” changes, only to confuse earthlings?
What a wonderful planet, if you can kick back, relax, with the beverage of your choice, and say WOW!
Leakage of 1 year is incidentally also in most global earth temperature datasets (eg. hadcrut4) but does not match the variation in sun/earth distance. Varies with date too.
Good reason to raise eyebrow and gaze skywards.
Is it really that tough for those paid to do it?
The annual peak-trough in surface temp shouldn’t be expected to match perihelion/aphelion, since there is lag in the system due to ocean heat capacity, and hemispheric differences in landmass react differently to earth’s tilt and the fact that the longest day of the year occurs before perihelion in the south – and after perihelion in the north. This due to the asymmetry of the seasons caused primarily by tilt and distance changes caused by eccentricity.
That’s an interesting point, Tim. I would certainly appreciate a more in-depth treatment of that and explanation of peoples thoughts on the matter.
Earth orbit effects are very complicated as Roger points out.
I’ve had to look at this often, try and decide what is reality, particularly troublesome when the data is a mish mash of actual measurements. This is not helped by the almost universal violation of Nyquist which introduces fictitious artefacts.
For a single eg. met station this can be a fingerprint of station change. Example, ages ago I discovered a wall of fast growing conifers had been allowed to grow sunside of the Steveson screen at the Armargh observatory, then was cut down and photographs of the new library building taken. The tree stumps look like fence posts. (material effect? Not really, data is not that good, many other issues) The point here is that the annual cycle changed, not hugely but enough to me to wonder and look.
At odd times I’ve played around trying to work out the best way of highlighting these things. No conclusion other than it takes time. and probably will show nothing useful.
Thanks, Tim. That’s an interesting example for me to play around with if I make the time etc.
I see the Armargh site temp data is available (as well as non-temp data) as a series of .pdf pages. Is it freely available as a simpler file(s)?