Updated sunspot group number reconstruction for 1749–1996 using the active day fraction method

Posted: May 17, 2017 by tallbloke in Analysis, Dataset, Solar physics, solar system dynamics

Ilya Usoskin has kindly sent me the data for the new group sunspot number series he and his colleagues have published. I’ve done a rough and ready plot below. Excel file here in case you have problems wit the links below.

group-sunspot

Group sunspot number average value. Missing values given as zero

T. Willamo1, I. G. Usoskin2,3 and G. A. Kovaltsov4

1 Department of Physics, University of Helsinki, 00014 Helsinki, Finland
2 Space Climate Research Unit, University of Oulu, 90014 Oulu, Finland
e-mail: Ilya.Usoskin@oulu.fi
3 Sodankylä Geophysical Observatory, University of Oulu, 90014 Oulu, Finland
4 Ioffe Physical-Technical Institute, 194021 St. Petersburg, Russia

Received: 4 October 2016
Accepted: 6 March 2017

Abstract

Aims. Sunspot number series are composed from observations of hundreds of different observers that require careful normalization to standard conditions. Here we present a new normalized series of the number of sunspot groups for the period 1749–1996.

Methods. The reconstruction is based on the active day fraction (ADF) method, which is slightly updated with respect to previous works, and a revised database of sunspot group observations.

Results. Stability of some key solar observers has been evaluated against the composite series. The Royal Greenwich Observatory dataset appears relatively stable since the 1890s but is approximately 10% too low before that. A declining trend of 10–15% in the quality of Wolfer’s observations is found between the 1880s and 1920s, suggesting that using him as the reference observer may lead to additional uncertainties. Wolf (small telescope) appears relatively stable between the 1860s and 1890s, without any obvious trend. The new reconstruction reflects the centennial variability of solar activity as evaluated using the singular spectrum analysis method. It depicts a highly significant feature of the modern grand maximum of solar activity in the second half of the 20th century, being a factor 1.33–1.77 higher than during the 18 and 19th centuries.

Conclusions. The new series of the sunspot group numbers with monthly and annual resolution is provided forming a basis for new studies of the solar variability and solar dynamo for the last 250 yr.

Key words: Sun: activity / sunspots


Monthly values of the reconstructed sunspot are available at the CDS via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via http://cdsarc.u-strasbg.fr/viz-bin/qcat?J/A+A/601/A109

Comments
  1. oldbrew says:

    The G-values declined rapidly after the 1988-1992 burst of high numbers.

    http://cdsarc.u-strasbg.fr/viz-bin/nph-Cat/html?J%2FA%2BA%2F601%2FA109

    Mostly below 2 in 1995-96 (end of cycle) compared to mainly double figures from late 88 to early 92.

  2. USteiner says:

    What is the reason for the end in 1996, 20 years and almost 2 cycles ago?

  3. tallbloke says:

    Usteiner, here’s part of an email I got from Frédéric Clette, Director of the World Data Center SILSO earlier today:

    The Group number was created in 1998 by Hoyt & Schatten using archived data going up to 1995. They never considered extending this series as they were only interested in the reconstruction of past centuries, in particular the historical period before 1800, when telescopes were of poor quality and the counts of individual spots were less reliable. Even the more recent versions of the series keep this philosophy. (NB: you can find all the group numbers series in our dedicated page: http://www.sidc.be/silso/groupnumberv3)

    However, given the renewed interest for this series, we thought that it would be worth starting the routine production of group numbers extending this series, in order to provide a longer overlap with all other modern solar indices, which only appeared in recent decades. So, this is in preparation in our World Data Center and it will be implemented probably in (early?) 2018. Note however that this group number would be part of the production of the existing international sunspot number that our WDC-SILSO is routinely producing using the counts from our worldwide network of observers (85 stations). It is just a simpler “byproduct” quantity that only uses the group counts, while the sunspot number uses both the group and spot counts (the latter give a measure of the actual size of the groups).

  4. lsvalgaard says:

    The ADF method does not actually work so well:
    http://www.leif.org/research/Assessment-Active-Days-Fraction-Method.pdf

    “We have identified several pairs of ‘equivalent’ observers and shown that the group
    numbers computed using the ADF-method do not reproduce the equality of the group
    numbers expected for equivalent observers, rendering the vaunted ADF-methodology
    suspect and not reliable nor useful for studying the long-term variation of solar activity.
    We suggest that the claim [Willamo et al., 2017] that their “new series of the sunspot
    group numbers with monthly and annual resolution, […] is forming a basis for new
    studies of the solar variability and solar dynamo for the last 250 years” is premature, and,
    if their series is used, will hinder such research. It is incumbent on the community to
    resolve this issue [Cliver, 2016] so progress can be made, not just in solar physics, but in
    the several diverse fields using solar activity as input.”

  5. tallbloke says:

    Hi Leif. I’m not at all surprised you don’t like the new GSN series. 🙂

  6. lsvalgaard says:

    I would if it were correct. Unfortunately it isn’t:
    http://www.leif.org/research/Sunspot-Group-Numbers-Since-1900.pdf
    http://www.leif.org/research/Spoerer-Not-Perfect-Observer.pdf
    and this very long one:
    http://www.leif.org/research/On-the-Backbone-Reconstructions.pdf
    (there has been progress; see Section 16 of the above)

    The bottom line is that we have a perfectly good GN [also agreeing with the official SN], but that that there is a whole cottage industry dedicated to show that the old Hoyt & Schatten series is good [because it supports the idea of significant solar influence on climate]. Not even Ken Schatten believes that anymore.

  7. tallbloke says:

    Thanks for your links Leif. I’ll let Ilya know you say he has it all wrong.

    As for the solar influence on climate, I’d say that was pretty undeniable, simply looking at the 10Be record (in blue). Our simple orbital resonance model (in yellow) replicates it very well too. So as well as a solar influence on climate, we have the planetary influence on solar variation. I know you don’t like that either, but there it is. Science moves on, and the days of ironing the solar record flat are numbered.

  8. lsvalgaard says:

    ” I’d say that was pretty undeniable”.
    I hereby deny it. And your graph does not show the climate record at all, apart from some vague verbiage…
    Here is a better comparison:

  9. tallbloke says:

    Thanks Leif. the link between solar activity and climate becomes much more obvious when you integrate the sunspot data as a running total departing from the longterm average to create a reasonable proxy for Ocean Heat Content (OHC). Then if you add in the AMO, PDO and a reasonable 0.4C per doubling for CO2 (seen to be lagging solar input as it does at all timescales), you get a good replication of SST. (Peason R^2=0.93 for monthly values in this simple model I made)

  10. tallbloke says:

    At around half the timescale your plot refers to, here’s what you get by integrating the 10Be data in the same way vs Mann’s later effort on millennial temperature.

    The Sun controls the climate. You backed the wrong horse when you threw in your lot with Steven “Truth vs Effectiveness” Schneider and his carbonated climate cronies I’m afraid.

  11. lsvalgaard says:

    Integrating the deviations from the long-term average always gives a zero result, so that is no good. The average, the window size, the end points, the AMO, etc are all free parameters and with enough of those you can fit anything. von Neuman: “with four parameters I can fit an elephant, with five I can make him wiggle his trunk”. As far as solar activity is concerned, the sunspot number [or the group number] reached the same heights in each of the three centuries 18, 19, and 20th within 15%, but the climate was vastly different. Etc, etc, etc. Solar activity now is as low as a century ago, yet it is much warmer now than then, etc, etc. All this are the same old points that everybody knows.

  12. tallbloke says:

    Leif: Integrating the deviations from the long-term average always gives a zero result, so that is no good.

    Quite clearly it doesn’t zero out on the millennial timescale, as you can see from the 1200-2000 CE plot above.

    All the criticisms you offer regarding my simple SST model are well received, providing you admit they also apply to everyone else’s models and plots too. You will note the R^=0.93 applies to MONTHLY data, so not a very free parameter there anyway. 0.4C per doubling of CO2 is Lindzen’s estimate, and I’d go with his expertise in atmospheric science over yours any day of the week. The timing of the ~65 year cycle in the AMO isn’t a ‘free parameter’, and the amplitude is set by the estimate of the effect of the AMO on global temperature empirically derived from the thermometer record.

    Your comments about solar activity are quite funny coming from a solar expert who has an interest in climate. To assess the overall effect of solar variation looking solely at the maximum amplitudes of sunspot numbers in solar cycles is a piece of pure misdirection. The cycle lengths are at least as important when assessing the impact of solar variation on decadal or longer timescales.

    This comment is particularly egregious:
    “Solar activity now is as low as a century ago, yet it is much warmer now than then”.

    After 70 years of higher than the longterm average of solar activity building up excess energy in the ocean heat content (OHC), you wouldn’t expect the temperature to suddenly fall back to 1910 levels after one low amplitude solar cycle, because the oceans have huge heat capacity, and that will take a long time to eke back out into the atmosphere during the coming period when solar activity is below the longterm average.

    Your thinking hasn’t progressed at all, I’m glad I don’t bother arguing with you at WUWT any more.

  13. lsvalgaard says:

    If you subtract the mean value from a function over an interval, the result will be positive some of the time and negative some of the time. Adding up [integrating] those will get you a round zero, if you had calculated correctly.

  14. tallbloke says:

    Adding up those values and dividing by the number of records will get you a round zero. But that is not an integration, it’s just an average, and the fact it sums to zero should surprise no-one. I think you’re just trying to sow confusion, because the evidence that the Sun controls Earth’s climate is clear by this method, and your aim is to hide the truth, or at least obscure it. Once again, here’s the integration of solar activity as recorded by the 10Be proxy, vs Mann et al 2008 millennial temperature. Ain’t it beautiful? Bumps in all the right places. 🙂

  15. lsvalgaard says:

    “Adding up those values and dividing by the number of records will get you a round zero. But that is not an integration, it’s just an average, and the fact it sums to zero should surprise no-one.”
    If you don’t divide by the number of records, then the result is the integral which is, of course, also zero.

  16. lsvalgaard says:

    Integral I = sum(x-m) where m=mean= sum x/n where n=number of records; so n x m = sum x
    so: I = sum (x-m) = sum x – sum m = sum x – n x m = n x m – n x m = 0

  17. tallbloke says:

    You’re confused. Allow me to help you.

  18. lsvalgaard says:

    An example to help you out:
    consider the series 1, 2, 3, 4, 5; mean = 3, now subtract the mean:
    1-3, 2-3, 3-3, 4-3, 5-3 = -2, -1, 0, +1, +2;
    add up the differences and you get 0 as (-2) + (+2) = 0, (-1) + (+1) = 0.

  19. tallbloke says:

    I never thought I’d see the day when Leif Svalgaard argued against the idea that there might be no secular trend in solar activity. 🙂

    Leif, for your information, the long term average of the sunspot number over the period of record is also the sunspot number you get empirically from a comparison of SST and SSN as the value at which the ocean neither gains nor loses heat.

    That’s why it’s the right value to use for the integration of the solar data. It’s also why it’s a reasonable assumption that an integration of the solar data will give a fairly good representation of ocean heat content.

  20. lsvalgaard says:

    There has been no long-term trend in solar activity since AD 1700.

    As for the integral: you still don’t get it. But I think that no amount of education will get you there.

  21. tallbloke says:

    Ah, the retreat to condescension without addressing the substantive argument. A classic Svalgaard distraction tactic.

    The 10Be data says there has been a trend in solar activity since 1700. As does everybody else’s sunspot series except the one you ironed flat. You are the outlier, soon to be left out in the cold when deliberations by the international team led by Matt Owens has completed its work.

  22. lsvalgaard says:

    On the no long-term trend since 1700:

    [Reply] Where are these published Leif? I can’t see the author’s names.

  23. lsvalgaard says:

    Here is Matt Owens’ latest result about the Heliospheric Magnetic Field [which controls 10Be flux]:

    As you can see: no long-term trend since 1700.

  24. tallbloke says:

    End point effect Leif. And those plots run from 1750, not 1700.

    1700 to when the modern warming ameliorated in 1998 would show a trend. And unlike you, Matt Owens is willing to work with other solar experts, which we’ll see the results of in due course.

  25. lsvalgaard says:

    You didn’t look at

    [Reply] Yes I did. Which is why I asked for the citations you’ve clipped.

  26. lsvalgaard says:

    http://www.leif.org/EOS/Muscheler-Cosmic-Ray-Topical-Issue.pdf
    “We presented an update of 10Be and 14C-based solar modulation reconstructions for the past 2000 years and a comparison to the revised sunspot records. […] In
    general, the sunspot and radionuclide records agree well. Especially the 14C-based record
    agrees very well with the revised sunspot data, lending strong support to these revisions.”

    [Reply] 404
    The requested item does not exist.

  27. lsvalgaard says:

    Usoskin, I. G., Arlt, R., Asvestari, E., Hawkins, E., Käpylä, M., Kovaltsov, G. A., Krivova, N.,
    Lockwood, M., Mursula, K., O’Reilly, J., Owens, M. J., Scott, C. J., Sokoloff, D. D., Solanki, S.
    K., Soon, W., Vaquero, J. M.: The Maunder minimum (1645–1715) was indeed a Grand
    minimum: A reassessment of multiple datasets, Astron.&Astrophys. 581, A95, doi:10.1051/0004-
    6361/201526652, 2015.

    [Reply] Ah, ok, so these are papers which have included your assessments among others?

  28. lsvalgaard says:

    Last word on this:

    The very weak trend is not significant: R2=0.0067

    [Reply] This is not “the last word” on GSN because other experts disagree, and your reconstruction is the outlier of a number of other reconstructions. Peddle your propaganda at WUWT, you don’t get to do that here.

  29. lsvalgaard says:

    No, those papers are by other people (Usoskin, Owens, et al.) and don’t use my stuff.

    [Reply]
    OK, I’ll go through them as I find time.

  30. lsvalgaard says:

    The Muscheler paper not found is WordPress not cooperating.
    Here is the reference: Solar Phys (2016) 291:3025–3043
    DOI 10.1007/s11207-016-0969-z

    [Reply] Thanks Open access here https://link.springer.com/article/10.1007/s11207-016-0969-z

  31. lsvalgaard says:

    “your reconstruction is the outlier of a number of other reconstructions.”
    your reaction is a typical example of confirmation bias. You accept what matches your bias.
    A have shown you that the ‘competition’ actually agrees with me. Slide 11 of http://www.leif.org/research/Debate-Material.pdf

    [Reply] I’m happy to await the reviewed and published results of the international co-operation effort underway.
    Your dismissal of Usoskin’s GSN for a tiny error in counting spotless days over a two year period by one observer is typical of your deceptive bullying. Go back to WUWT, you won’t get away with it here.

  32. lsvalgaard says:

    Cherry picking. You should have picked Figure 11. Do me a favor and show it here.

    [reply] Sure. Do you have a hi res version of fig 11 and fig 10 as well? Springer won’t let me enlarge the open access versions.

  33. lsvalgaard says:

    Figure 10 actually. The numbering has changed in the final version. Their conclusion:
    ” In general, the sunspot and radionuclide records agree well. Especially the 14C-based record agrees very well with the revised sunspot data, lending strong support to these revisions”

    [Reply] Yes, the revised sunspot data (Clette et al), not the revised group sunspot record (svalgaard & Schatten).

  34. lsvalgaard says:

    “Your dismissal of Usoskin’s GSN for a tiny error in counting spotless days over a two year period by one observer ”

    Nonsense. Study http://www.leif.org/research/Assessment-Active-Days-Fraction-Method.pdf carefully.
    “We identify several pairs of ‘equivalent’ observers defined as observers with equal or nearly equal ‘observational threshold’ areas of sunspots on the solar disk as determined by the ‘Active Days Fraction’ method [e.g. Willamo et al., 2017]. For such pairs of observers, the ADF-method would be expected to map the actually observed sunspot group numbers for the individual observers to two reconstructed series that are very nearly equal and (it is claimed) represent ‘real’ solar activity without arbitrary choices and deleterious, error-accumulating ‘daisy-chaining’. We show that this goal has not been achieved (for the critical period at the end of the 19th century and the beginning of the 20th), rendering the ADF-methodology suspect and not reliable nor useful for studying
    the long-term variation of solar activity.”

    Decades of discrepancy by several observers.

    [Reply] Usoskin is preparing an addendum to their paper. Feel free to comment on it in the literature. You’re not getting a platform to snipe at them here.

  35. tallbloke says:

    Here are figs 10 and 11 from the Muscheler paper, showing the radio-isotope series and the revised sunspot (Black: Clette et al 2014) and group sunspot (Blue: Svalgaard & Schatten) series. Clette appears to be a more ‘middle of the road’ reconstruction compared to the Svalgaard outlier. The 14C and both 10Be curves show a definite uptrend from 1700, as does the (Clette et al) revised SSN series.

    Finally, a comparison of Svalgaard/Schatten and Usoskin’s new reconstruction of GSN, showing quite strong divergence in late 18th and mid 19th century peak amplitudes. As usual, Svalgaard’s reconstruction is tending towards a flattening of the solar record, and disagrees with the empirical evidence from isotope records, particularly C14, which Clette follows much more closely.

  36. lsvalgaard says:

    Clette and S&S agree exceedingly well from 1749 on. Before that the uncertainty is very great. as the data is VERY sparse, e.g. in 1738 there were observations on only a single day. The group number for that year has an uncertainly of plus/minus 20%, thus a span of 40%. Any difference or discrepancy would be in the noise.

    And Usoskin’s new GSN is precisely the one that I show is wrong:
    http://www.leif.org/research/Assessment-Active-Days-Fraction-Method.pdf

    Also see how much it deviates from the official sunspot number.
    As Muscheler said: “the sunspot and radionuclide records agree well. Especially the 14C-based record agrees very well with the revised sunspot data, lending STRONG support to these revisions.

    BTW, the 10Be record is not so good [contaminated by climate].

    [Reply] “Clette and S&S agree exceedingly well from 1749 on” Not according to the black and blue curves on fig’s 10 and 11.
    “Before that the uncertainty is very great.” Maybe that’s why Usoskin’s series starts in 1749.
    “And Usoskin’s new GSN is precisely the one that I show is wrong:” As I already told you, they are preparing an addendum, and you should respond to it in the literature, because I won’t give you a platform to attack them here.
    “BTW, the 10Be record is not so good” The Antarctic 10Be appears to closely match Clette’s reconstruction (especially where your GSN goes way over the top in the 1700s).

  37. Geoff Sharp says:

    Pretty tragic watching Svalgaard trot out the same old tired graphs for at least a decade..he definitely the outlier with his sunspot re creation. The solar experts not liking his “backbone” method.

    Also interesting that he still has not come up with any answers to the following graph presented to him 3 years ago. (McCracken et al 2014)

    No grand minima across the Holocene occurring in the first column.

    The first column is the period when Neptune and Uranus are opposed.

    If the data cannot be falsified it is clear there is a planetary influence on on solar grand minima.

  38. lsvalgaard says:

    “Not according to the black and blue curves on fig’s 10 and 11.”
    The black and blue curves agree exceedingly well after 1749.
    As I showed you before. Here it is again:

    The pink curve is Clette, the blue S&S.
    Show the graph here if you have any integrity left. It is amazing that you could deny this.

    On Usoskin: my papers are all submitted to journals for peer review and will probably be published after the usual delay of about a year. And I don’t need your platform for showing that Usoskin et al. are wrong. Even the sunspot number [Clette] and the solar modulation [Muscheler] show that already. My emphasis is to find out WHY they are wrong, not IF, as that is by now ell-established.
    But, don’t despair: with your curve fitting I’m sure you can fit our new series just as well as with the old, perhaps, you need yet another cycle or two, e.g. the 1000-yr cycle some people are peddling.

    Steinhilber’s 10Be was mostly based on Greenland. Before 1749 the data are to poor to even discuss discrepancies, except, of course, if they fit one’s bias.

    [Reply] So far as I can tell from the text, the black curve on figs 10 and 11 is Clette’s SSN reconstruction not GSN.

  39. lsvalgaard says:

    like this one:
    Dr Norman Page June 7, 2017 at 9:19 am
    See comment at 7:05 AM above. Temperature is driven a combination of the various orbital cycles and the cycles in solar activity with the millennial cycle being the most significant.

    https://wattsupwiththat.com/2017/06/06/solar-update-june-2017-the-sun-is-slumping-and-headed-even-lower/comment-page-1/#comment-2521827

    [reply] Sorry Leif, I’m not in the least bit interested in the spats you have on WUWT

  40. lsvalgaard says:

    Geoff: “The solar experts not liking his “backbone” method.”
    None of those folks are ‘solar experts’. However, Frederic Clette, Ken Schatten and I am bona fide solar experts. And we agree on the new official versions

  41. lsvalgaard says:

    [Reply] So far as I can tell from the text, the black curve on figs 10 and 11 is Clette’s SSN reconstruction not GSN.

    Instead of squinting at a small figure, simply go to the SILSO website and download Clete and GSN and you’ll see that they are identical after 1749 apart from a constant scale factor as I showed here:

    Bottom line: Clette and I agree completely after 1749 and the differences before that is entirely in the noise.

    [reply] Sorry Leif, I’m not in the least bit interested in the spats you have on WUWT

    Not mine, but Norman’s where he claims a 1000-yr cycle. You may need that cycle too for your curve fitting.
    I note that you are afraid of taking my comments out of moderation.

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