When I last wrote about the solar activity situation, things were (as Jack Horkheimer used to say) “looking up”. Now, well, the news is a downer. From the Space Weather Prediction Center (SWPC) all solar indices are down, across the board:
The radio activity of the sun has been quieter:
And the Ap Geomagnetic Index has taken a drop after peaking last month:
WUWT contributor Paul Stanko writes:
As has been its pattern, Solar Cycle 24 has managed to snatch defeat from the jaws of victory. The last few months of raw monthly sunspot numbers from the Solar Influences Data Analysis Center (SIDC) in Belgium are: January = 12.613, February = 18.5, March = 15.452, April = 7.000 and May = 8.484. After spending 3 months above the criteria for deep solar minimum, we’re now back in the thick of it.
The 13 month smoothed numbers, forecast values and implication for the magnitude of the cycle peak are as follows:
- June 2009 had a forecast of 5.5, actual of 2.801, implied peak of 45.83
- July 2009 had a forecast of 6.7, actual of 3.707, implied peak of 49.79
- August 2009 had a forecast of 8.1, actual of 5.010, implied peak of 55.67
- September 2009 had a forecast of 9.7, actual of 6.094, implied peak of 56.55
- October 2009 had a forecast of 11.5, actual of 6.576, implied peak of 51.46
- November 2009 had a forecast of 12.6, actual of 7.190, implied peak of 51.36
- December 2009 had a forecast of 14.6, actual would require data from June.
Solar Cycle 24 now has accumulated 810 spotless days. 820, which would require only 10 more spotless days, would mean that Cycle 24 was one standard deviation above the mean excluding the Dalton and Maunder Grand Minima.
One standard deviation is often an accepted criteria for considering an occurrence ‘unusual’.
Here are the latest plots from Paul Stanko:
Geoff Sharp says:
June 17, 2010 at 4:40 pm
We need to see a clear graph of the non manipulated base data properly overlaid to gain some sort of comparison.
That is what my plot shows. The ‘manipulation’ bit is just a strawman as the difference is below 1.5%. What he plots is the difference between ‘observed’ and ‘proxy’. This is to make it easier to see.
He also states flux (observed), making it unclear if he has used the adjusted for orbit flux?
‘observed’ is as opposed to ‘proxy’. Of course, he uses the ‘adjusted’ values. All solar physicists do.
along with his own doubts on the accuracy of his method.
What doubt?
Geoff Sharp says:
June 17, 2010 at 4:40 pm
We need to see a clear graph of the non manipulated base data properly overlaid to gain some sort of comparison.
The [almost] correct way of doing is to find a regression formula for 1947-1990 for SSN vs. F10.7 flux. This allows us to calculate the SSN from the flux at all times and check if the SSN during SC23 is lower than that predicted by the formula. I say ‘almost’ because that comparison assumes that the calibration of F10.7 has been constant, which we know from the Japanese data that it was not, but since the error is small, we’ll let that slide. First the regression formula based on 16,000 individual days:
http://www.leif.org/research/F107-SSN-Daily-1947-1990.png
and then the comparison [solar cycle per solar cycle] between observed [reddish] and calculated from F10.7 [bluish] for every single day since measurements began until today:
http://www.leif.org/research/SSN-Daily-derived%20from%20F10.7%20and%20Observed.png
The heavier curves are 27-day running averages. It should now be clear that during SC23 the reddish curves are consistently below the blue, showing a deficiency of spots as Tapping said there should be.
Leif Svalgaard says:
June 17, 2010 at 7:10 pm
An update on the L&P Effect is Debunked story.
“Ken Tapping has performed some analysis on this topic and shows that there is some divergence between flux and sunspots during SC23. http://helios.swpc.noaa.gov/sww/2010/friday/SWW_2010_KFT.pdf He uses the standard Canadian flux data and shows plots describing this variance. The difference in this analysis is he is not showing a trend change as per Dr. Svalgaard but an overall difference covering the entire cycle. If I smooth the standard data further a large anomaly is apparent around 20001/2002 and then the two measures come back together. The reason for this divergence could be high flare activity at cycle max or one of the other factors that drive F10.7 Flux, it is more of a one off event rather than a gradual diminishing of sunspots.”
http://www.landscheidt.info/images/sidc_flux_200.png
There is no need for fancy formula or manufacturing composite data, everything is apparent in the unmodified base data. I have still not seen credible evidence for a decline in sunspots that will render them invisible by 2015.
Geoff Sharp says:
June 17, 2010 at 9:06 pm
“Ken Tapping has performed some analysis on this topic […] The difference in this analysis is he is not showing a trend change as per Dr. Svalgaard but an overall difference covering the entire cycle.
You are misrepresenting his analysis. His slides 5 and 7 show that the difference was smaller in the beginning of the cycle and then growing.
There is no need for fancy formula or manufacturing composite data, everything is apparent in the unmodified base data.
You analysis is misleading [even invalid] as you [it seems] let Excel do the scaling for you [correct me if I’m wrong by telling us how the scaling was done], but this is incorrect if the relationship is non-linear [as it is – his slide 5]. To make a correct analysis you have to find a non-linear function [I use a polynomial – Ken uses an exponential] that is a good fit over the period where you except there is no trend – such as 1947-1990]. Then calculate the proxy and compare the curves. If you do this correctly you’ll find the same result as Tapping as I. In my previous post I show you the deviation for every single day since the measurements began in 1947. The ratio between observed and calculated [scaled] SSNs shows the downward trend I have shown before.
I have still not seen credible evidence for a decline in sunspots that will render them invisible by 2015.
2015 is probably too early. More like 2018. Anyway, the spots may just end up with reduced visibility, leading to an undercount of the sunspot number. It does help if you study the material and try to follow the advice about making correct analysis.
Geoff Sharp says:
June 17, 2010 at 9:06 pm
There is no need for fancy formula or manufacturing composite data
A composite of several reliable datasets is much better than the individual datasets, as they support each other in the composite. Your use of ‘manufacturing’ betrays a deep lack of understanding of data and their analysis, or an attempt of willful smear. Perhaps you could tell me which of the two it is.
Leif Svalgaard says:
June 17, 2010 at 10:54 pm
Manufactured ie assembled from components, just not sure I like your bill of materials.
The F10.7 value on your graph at http://www.leif.org/research/TSI-SORCE-2008-now.png is looking very sick, the sunspot count is beginning to read higher… watts up with that?
Geoff Sharp says:
June 18, 2010 at 4:17 am
Manufactured ie assembled from components, just not sure I like your bill of materials.
You still didn’t tell me which of the two possibilities it is. The ‘material’ is high quality microwave radiometry data, carefully observed over decades by competent researchers.
The F10.7 value on your graph at http://www.leif.org/research/TSI-SORCE-2008-now.png is looking very sick, the sunspot count is beginning to read higher… watts up with that?
The sunspot number is indeed all of 8.
Leif Svalgaard says:
June 18, 2010 at 5:00 am
In the “manufacturing” industry when designing a quality finished product the BOM requirements are paramount. When there are competitors for components the best man wins (when cost is not an issue), the track record of the Japanese component is in doubt and should be sidelined until it meets quality assurance.
Geoff Sharp says:
June 18, 2010 at 5:23 am
the track record of the Japanese component is in doubt and should be sidelined until it meets quality assurance.
It is not in doubt, and I [and Hugh Hudson] have personally assured its high quality: http://www.leif.org/research/AGU%20Fall%202009%20SH13C-03.pdf
Our conclusion:
“The Canadian and Japanese microwave radiometry is stable, robust, and of high quality”.
And from:
http://www.leif.org/research/Solar-Microwaves-at-23-24-Minimum.pdf
• As shown by Tanaka et al. (1973), these data (fixed frequencies, integrated Sun) can be calibrated precisely
• The microwave flux is more objective than the sunspot number
• The main high-quality data sources (Canada for F10.7, Japan for five frequencies at present), have been maintained very well
If anything, the Canadian series suffers from a [small] discontinuity stemming from the move from Ottawa to Penticton. In any event, this small error does not alter anything.
And still:
Your use of ‘manufacturing’ betrays a deep lack of understanding of data and their analysis, or an attempt of willful smear. Perhaps you could tell me which of the two it is.
Leif Svalgaard says:
June 18, 2010 at 6:34 am
You seem hung up on a smear campaign, I am not suggesting you would willfully try to delude the solar community. But perhaps you should do your own QA analysis and question why the Japanese record is below the Canadian standard on the first installment and above on the second. The sunspot record agrees with the Canadian standard which is adopted by most and appears to be the leader in quality assured product.
Leif, loved the powerpoint. Especially the magnetic coiling up as a result of the rotation speed difference between the equator and poles. I am once again reminded of how we made yarn draw strings for our 4H knitting class, or how we used to twist the long chains of our school yard swings as tight as we could (I needed help because my feet couldn’t touch the ground) and then let them untwist with dizzying stomach churning speed.
Geoff Sharp says:
June 18, 2010 at 7:19 am
You seem hung up on a smear campaign
I consider the phrase ‘manufacturing data’ to be a smear. Good to see that you agree.
But perhaps you should do your own QA analysis
We have worked closely with world experts [Tapping, Shibasaki, Hurford] on the QA of both series and we all agree that both series are excellent.
question why the Japanese record is below the Canadian standard on the first installment and above on the second.
Because the Canadian ‘standard’ suffers from a jump when they moved from Ottawa to Pentiction [now, how many times have I explained that to you?]. Such small jumps are hard to avoid [as there are local – and seasonal – influences on the flux measurements] and are not really a sign of poor quality. So, the Canadian series is not of lower quality. You just have to be aware of site discontinuities and correct for them as best one can.
The sunspot record agrees with the Canadian standard which is adopted by most and appears to be the leader in quality assured product.
The sunspot record agrees with both the Canadian and the Japanese series within the uncertainty of the sunspot series. ‘Appears to be the leader’? says who? The director of the Canadian observatory [Tapping] is a co-author on our paper that compares both series and that finds both to be excellent. The Japanese series is perhaps less known, but that does not mean that it is inferior.
The Sun produces most of the microwave flux in a frequency band between 1.5 GHz and 4 GHz. The choice of 10.7 cm [2.8 GHz] by Covington was dictated solely by accident as that was the wavelength that the military surplus hardware he had was tuned to. The composite of 2.0 MHz [Japanese], 2.8 GHz [Canadian], and 3.75 GHz [Japanese] is a very good measure of solar activity. Probably the best we have at this moment.
Geoff Sharp says:
June 18, 2010 at 7:19 am
You seem hung up on a smear campaign
I consider the phrase ‘manufacturing data’ to be a smear. Good to see that you agree.
But perhaps you should do your own QA analysis
We have worked closely with world experts [Tapping, Shibasaki, Hurford] on the QA of both series and we all agree that both series are excellent.
question why the Japanese record is below the Canadian standard on the first installment and above on the second.
Because the Canadian ‘standard’ suffers from a jump when they moved from Ottawa to Pentiction [now, how many times have I explained that to you?]. Such small jumps are hard to avoid [as there are local – and seasonal – influences on the flux measurements] and are not really a sign of poor quality. So, the Canadian series is not of lower quality. You just have to be aware of site discontinuities and correct for them as best one can.
The sunspot record agrees with the Canadian standard which is adopted by most and appears to be the leader in quality assured product.
The sunspot record agrees with both the Canadian and the Japanese series within the uncertainty of the sunspot series. ‘Appears to be the leader’? says who? The director of the Canadian observatory [Tapping] is a co-author on our paper that compares both series and that finds both to be excellent. The Japanese series is perhaps less known, but that does not mean that it is inferior.
The Sun produces most of the microwave flux in a frequency band between 1.5 GHz and 4 GHz. The choice of 10.7 cm [2.8 GHz] by Covington was dictated solely by accident as that was the wavelength that the military surplus hardware he had was tuned to. The composite of 2.0 MHz [Japanese], 2.8 GHz [Canadian], and 3.75 GHz [Japanese] is a very good measure of solar activity. Probably the best we have at this moment.
I direct you to my analysis based solely on the Canadian data [and thus of diminished weight]: http://www.leif.org/research/SSN-Daily-derived%20from%20F10.7%20and%20Observed.png that shows the deviations for every single day since 1947 until yesterday. That should put the doubt to rest. It is now your chance to see the light.
Geoff Sharp says:
June 18, 2010 at 7:19 am
But perhaps you should do your own QA analysis
We have worked closely with world experts [Tapping, Shibasaki, Hurford] on the QA of both series and we all agree that both series are excellent.
This does not mean that there are not issues [especially with the Canadian series], but the data is just the best we have.
As an example of a problem with the Canadian series [and an example of the excellent QA we have performed] I can mention the Sawtooth-problem. There are systematic differences between morning values [1700 or 1800 UT], noon values [2000 UT] and afternoon values [2200 or 2300 UT] in the winter time:
http://www.leif.org/research/F107-Sawteeth.png
[shows the square root of F10.7 for illustration purposes]
The teeth appear every day during winter and is probably related to influence [reflection?] of snow on surrounding mountains. They are a good example of influences that can vary from site to site.
So, I do not want to hear anymore whining about QA. We do the best that is humanly possible.
Leif Svalgaard says:
June 18, 2010 at 9:33 am
The Sun produces most of the microwave flux in a frequency band between 1.5 GHz and 9 GHz. […] The composite of 2.0 MHz [Japanese], 2.8 GHz [Canadian], and 3.75 GHz [Japanese] is a very good measure of solar activity.
I should clarify this a bit and correct the typo: 2.0 GHz, not MHz. The band is where the flux varies the most in response to [rotational modulation of] solar activity. All this is just technical detail. By good fortune, the military surplus hardware that Covington got to work with in 1946 was just in the most sensitive part of the microwave spectrum.
Awesome! Thanks for the info!