SIDC graphic, annotation added TNC
- SC (red dash-dotted) : classical Standard-Curves prediction method, based on a least-square interpolation of Waldmeier’s set of standard solar cycle profiles;
- CM (red dashes) : Combined Method (K.Denkmayr and P. Cugnon), based on a regression technique combining a geomagnetic precursor (aa index) with a least-square fit to the actual profiles of the past 24 solar cycles.
http://sidc.oma.be/sunspot-index-graphics/wolfjmms.php
One or other or both will be wrong.
SIDC have recently altered their sunspot graphics pages. A divergent of sunspot model prediction is being shown. The phrase ‘All models are wrong, some are useful’ comes to mind.
Tim
Tallbloke's Talkshop 
The second peak: probability of second peak being larger than first ……?
But maybe nothing in our universe is probabilistic these special days.
The deterministic worldview wishes everything could be nailed to the floor and, dammit, will live that way until forced otherwise.
We’ll see at least another peak. This cycle is weak and that means long. The next minimum not before ~2021, maybe later.
I think the whole solar thing is at the moment ‘we ain’t a clue’
If you really want to see how difficult it is to predict the behavior of our fickle sun, there’s a good paper here…
Solar Cycle Prediction – Kristóf Petrovay
Click to access lrsp-2010-6Color.pdf
It would seem solar science still has a long way to go to understand how the sun works and be able to make accurate predictions about future activity. This means predicting the future climate of Earth is a futile exercise, and sun has a massive effect on our climate.
My guess is that SC24 will continue to bumble along as it fizzles towards solar minimum. The 20th century grand solar maximum is now behind us, and I think Earth climate will revert to cool mode again. How low it will go I just don’t know.
In “Why does the current Sunspot Cycle stagnate?”
http://www.gsjournal.net/Science-Journals/Essays/View/4598
I made an attempt to clarify why I had put forward the expectation that Solar Cyle 24 would stagnate as of the end of 2011 (based on similarity with prior solar cycles).
I would expect Solar Cycle 24 to resemble most with Solar Cycle 12.
For both cycles, the stagnation started with the reversal of the Sun own acceleration/deceleration.
Solar Cycle 12 showed an increase before starting the decline …
so the CM prediction method (including the Kalman filter) may come closest.
Solar Cycle 5 also stagnated after the reveral but showed a lower overall intensity.
Like it Tenuc
“Starting from the 1980s many researchers jumped on the chaos bandwagon”
and Neural networks, and genetic and every other magical method.
I was there, looked, tested, cynical, bumps into the AI problem which to this day seems to escape comprehension by those in the field. Fascinating area at least to some.
Some things can be done.
Forget about the ‘models’ talk
unless they are on the cat-walk
planetary extrapolation will do the biz
as in this old Vuk’s viz
tchannon says:
July 3, 2013 at 8:19 pm
Like it Tenuc
“Starting from the 1980s many researchers jumped on the chaos bandwagon”
“and Neural networks, and genetic and every other magical method.”
The semi-chaotic base on which everything relies makes short term ‘noise’ the biggest problem.
Removing that ‘noise’ without mangling the signal is the trick. I think it needs a very partiuclar form of circuit 🙂
Easy enough to do. Take a reasonably long dataset, withhold a signficant portion at the end, show predictive capability.
Sunspot data is reasonably well known from late 1800s, withhold start of cycle 23 and later.
tchannon: tallbloke: I have just published this on WhatsUp in response to a thread by Bob Tinsdale.
What do you think?
Ideal digital sampling series for a bandpass splitter in Climate research.
next = rounded(previous*1.3371) starting from 3.
It precisely ‘nulls’ the 12 month signal whilst leaving all of its harmonics and all other frequencies intact.
(The same applies to any sample frequency that has one of the later poles as its direct multiple.)
This is a digital implementation of a ‘brick wall’ cascaded low pass/bandpass splitter circuit (approx 1/3rd octave).
As it is a digital average, it has a ‘square wave’ sampling methodology on the source data.
The well known side effects of this ‘square wave’ sampling are cancelled out by using the 1.3371 inter-stage multiplier.
On the plus side, it has a infinite(?) roll off per octave between stages so the bands are precise, though not tunable.
Also, as it is completely de-tuned in the passbands, it is completely insensitive to internal data distribution ‘in band’.
On the digital side, this is just the well know 3 pole filter arrangement which running means require, extended to the full cascaded filter set.
It can be used on either normalised (at 12 months) or non-normalised data.
When used as a simple 3 stage filter on daily data it can provide a 28 day smooth which is more scientifically accurate than a human Monthly one.
When looking at temperature data, this series and circuit should be used rather than Normals of any period.
Because of the 4 year period it discovers in temperature series, Yearly Normals should be discarded.
When used in Splitter mode (each stage subracted from the previous stage) it provides a ‘DC’/zero referenced view of each passband.
The final stage can be set to ‘DC’/zero or a ramp as required for any very long term trends (such as CO2).
Nodes can be extracted by using, -previous*next between each stage, or other such similar arrangements.
RMS power can be extracted from each passband in the normal way.
The maximum period that can be discovered by this method is limited by the record length and this is a power series.
Getting very long cycles will require correspondingly much longer input data!
Fig 1
Fig 2
Fig 3
tchannon: tallbloke: and if we apply it to hourly data, there goes min, max and half way between for daily data.
I’ll be speaking with Tim for a precis.
Thanks. Sorry to barge in on your blog. Difficult to get heard over the noise.
The thing is, If we can calculate the input power from the sun, distributed correctly across the whole Earth’s surface as the input term (with the appropriate frequency pattern on Land and Sea as required) and then calculate the general, long term average of all the temperature sources we have (recorded and extended) as the final ‘DC’/zero reference point then it will all fit somewhere in this digital bandpass splitter filter series! We can at least try and figure out what is really left to discover.