|
|
Post by af4ex on Feb 5, 2011 14:34:38 GMT
@leif > a count has no statistical significance. If I count 5 people on > the other side of the street, 5 people it is. No error bar or > confidence interval.... But a spot has to be detected before it can be counted, therefore is subject to the usual errors of Signal Detection Theory: en.wikipedia.org/wiki/Detection_theoryEven after detection, there will be counting errors. (That's why they invented double-entry accounting: en.wikipedia.org/wiki/Double-entry_bookkeeping_system ) Isn't that why we have a bunch of different groups counting sunspots? I'll bet their counts are not all the same. Hopefully any random discrepancies will be averaged out. :-| The main problem is that the so-called 'seeing' varies with location and time and the seeing is the real limiting factor, so you get the variance mostly from that. Not from accounting errors. The original Wolf number was based on counting only spots that were big enough that they would not be influenced by variable seeing. Wolf's successor, Wolfer, changed the counting method to count ALL spots no matter how small that are visible. To put the new counts on the original Wolf scale, it is multiplied by 0.6 [except that the NOAA count is not so modified]. As a result of this rather 'messy' state of affairs, whenever I look at sunspot activity reports, I usually can't tell whether they're Fahrenheit or Celsius.  |
|
bradk
Level 3 Rank
Posts: 199
|
Post by bradk on Feb 5, 2011 14:47:35 GMT
The _smoothed_ SC14 distribution looks approximately normally distributed. But the raw data is so spikey that the normal statistical inferencing would be meaningless. Looks more like a harmonic analysis problem. True, but bootstrapping ang modeling are really pretty good these days (partially thanks to gene mapping in humans - a very messy statistical business) and we could likely get a pretty good guess, that said, as Leif points out, even if we did it it wouldn't mean much. |
|
bradk
Level 3 Rank
Posts: 199
|
Post by bradk on Feb 5, 2011 14:50:33 GMT
[quote author=bradk It is too early to tell. On the other hand, there is good evidence that sunspots are harder to see [Livingston] and are thus undercounted right now. making comparison even more dubious. The spots are seeming to get tinier and tinier. Time will tell if Livingston and you are correct, and at what rate they will get so small they will dissappear - I am still betting it is sooner than expected and this is not like solar cycle 14, but...I have no real idea what I am talking about ;D |
|
|
|
Post by lsvalgaard on Feb 5, 2011 15:07:36 GMT
[quote author=bradk It is too early to tell. On the other hand, there is good evidence that sunspots are harder to see [Livingston] and are thus undercounted right now. making comparison even more dubious. The spots are seeming to get tinier and tinier. Time will tell if Livingston and you are correct, and at what rate they will get so small they will dissappear - I am still betting it is sooner than expected and this is not like solar cycle 14, but...I have no real idea what I am talking about ;D I'll agree that SC14 did not suffer from the L&P effect. So F10.7 for SC24 and SC14 are expected to be similar [slide 7 of www.leif.org/research/Eddy-Symp-Poster-1.pdf ], but the SSN for SC24 will progressively deviate. Lots of ground for speculation. |
|
bradk
Level 3 Rank
Posts: 199
|
Post by bradk on Feb 5, 2011 18:23:07 GMT
I'll agree that SC14 did not suffer from the L&P effect. So F10.7 for SC24 and SC14 are expected to be similar [slide 7 of www.leif.org/research/Eddy-Symp-Poster-1.pdf ], but the SSN for SC24 will progressively deviate. Lots of ground for speculation. Wow, maybe I was onto something as that was my point, namely the L & P effect will cause the sunspot numbot number in this cycle to be lower than SC14. As my dad used to say. " Even a blind squirrel finds an acorn every now and then." Intersting times, and thank you for being such a knowledgeable and tolerant tour guide. |
|
bradk
Level 3 Rank
Posts: 199
|
Post by bradk on Feb 5, 2011 18:25:55 GMT
10-15 years of L&P observation is far too short in the Sun’s history to make any serious, let alone scientifically founded, statements about the sunspots disappearance. Here are current monthly non-smoothed SSN compared to the Dalton min: True, but if one does not form a hypothesis one cannot test it and then science does not move forward, right? As for the comparison, the timeframes you have selected for comparison may, or may not, be the correct ones to correlate. Time will tell... |
|
|
|
Post by lsvalgaard on Feb 5, 2011 18:30:54 GMT
10-15 years of L&P observation is far too short in the Sun’s history to make any serious, let alone scientifically founded, statements about the sunspots disappearance. Here are current monthly non-smoothed SSN compared to the Dalton min: The SSN around 1800 is highly uncertain [perhaps to a factor of two] so any numerical comparison may not be too significant. |
|
bradk
Level 3 Rank
Posts: 199
|
Post by bradk on Feb 5, 2011 21:50:52 GMT
vukcevic-
This seems a problem of definition, if I understand everyone's point. If there is a true L & P effect as discussed here, meaning that the magnetism per spot drops to such a level that the spots are invisible then we are sure that did not happen in 1900 or 1950 because we saw spots then. If you are arguing that magnetism per sunspot varies in some way beyond the major change we are seeing now, then you could be right.
|
|
|
|
Post by lsvalgaard on Feb 6, 2011 7:19:22 GMT
What I am implying is that L&P effect either does not exist as an independent variable, or if it does, it is a direct function of a sunspot (SSN) or magnetic (Hale) cycle. It is just not credible to say that L&P is there now but it was not there in 1950’s or 1900's. We have evidence [better than credible] that the L&P effect did not occur in 1900s. The ratio SSN/F10.7 was 'normal' back then, but is abnormal now, so L&P is not a function of the SSN. |
|
bradk
Level 3 Rank
Posts: 199
|
Post by bradk on Feb 6, 2011 12:43:16 GMT
Vukcevic- ...but it is not the count alone. It is the flux versus the sunspot number, not just the number. A poor man's approach might be to replace flux with sunspot area as we have that data going back a long way and area is what actually drops, at least early in the period where the L & P effect matters: solarscience.msfc.nasa.gov/greenwch/daily_area.txtJust eyeballing your period does make it look similar to today, but that might simply show the L & P effect is not that rare and that F10.7/SSN number does change over time and that the L & P effect can reverse from any point in its drop and need not go to "grand minima" lows before it retuns to "normal." It also might show nothing, namely that there is a difference between "normal" minima and grand minima and that there are many reason SSN number can drop for shorter periods, absent an L & P effect. I would think sunspot area would be a good proxy for flux Dr. Svalgaard, is this correct? If I don't drink much and the Super Bowl is boring I might pop some data into Excel and play with it to see if the area argument has real validity. |
|
|
|
Post by lsvalgaard on Feb 6, 2011 15:41:48 GMT
vukcevic- This seems a problem of definition, if I understand everyone's point. If there is a true L & P effect as discussed here, meaning that the magnetism per spot drops to such a level that the spots are invisible then we are sure that did not happen in 1900 or 1950 because we saw spots then. If you are arguing that magnetism per sunspot varies in some way beyond the major change we are seeing now, then you could be right. This is first (major) minimum that L&P has been studied, my previous post shows (1810s) that the periods where spots became invisible is not that unusual. For 1900’s visibility was not as you suggest. Average SSN monthly during 42 consecutive months, 3.5 years (1910-1914) average SSN count was 3.9 , just above half of 7.0 in the current minimum during same length of the time, i.e. in 1910s min visibility of the spots was about half what is in the current minimum. Here I show certain correlation between L&P contrast and SSN up to the mid Sept 2010. If Dr. S. has the data beyond September I could update and follow up to progression. The data does not allow trend lines to be drawn with any confidence other that the linear [minimum assumption] line. The ratio ~SSN/(Flux-66) does not show a solar cycle dependence and is constant 1835-1990 [with the usual noise on top], but has fallen to about half of that value since, suggesting that L&P started about 1990. The SSN alone does not say anything about L&P. There is no evidence of L&P around 1810. L&P was likely present 1645-1715. Here is reconstructed F10.7 since 1835:  or using more stations:  and the divergence:  From the above data one would actually PREDICT the L&P effect and sure enough it is observed. L&P alone does not have enough statistical significance to constitute the extraordinary evidence for the extraordinary claim that L&P make. In combination with F10.7 and its proxy, the argument becomes compelling. Back to 1781 we have Wolf's reconstruction of the daily change of the magnetic needle [proxy for F10.7]:  |
|
|
|
Post by sigurdur on Feb 6, 2011 17:16:58 GMT
Thank you Dr. Svalgaard for your posts here.
|
|
|
|
Post by france on Feb 6, 2011 21:11:40 GMT
This is first (major) minimum that L&P has been studied, my previous post shows (1810s) that the periods where spots became invisible is not that unusual. For 1900’s visibility was not as you suggest. Average SSN monthly during 42 consecutive months, 3.5 years (1910-1914) average SSN count was 3.9 , just above half of 7.0 in the current minimum during same length of the time, i.e. in 1910s min visibility of the spots was about half what is in the current minimum. Here I show certain correlation between L&P contrast and SSN up to the mid Sept 2010. If Dr. S. has the data beyond September I could update and follow up to progression. The data does not allow trend lines to be drawn with any confidence other that the linear [minimum assumption] line. The ratio ~SSN/(Flux-66) does not show a solar cycle dependence and is constant 1835-1990 [with the usual noise on top], but has fallen to about half of that value since, suggesting that L&P started about 1990. The SSN alone does not say anything about L&P. There is no evidence of L&P around 1810. L&P was likely present 1645-1715. Here is reconstructed F10.7 since 1835: or using more stations: and the divergence: From the above data one would actually PREDICT the L&P effect and sure enough it is observed. L&P alone does not have enough statistical significance to constitute the extraordinary evidence for the extraordinary claim that L&P make. In combination with F10.7 and its proxy, the argument becomes compelling. Back to 1781 we have Wolf's reconstruction of the daily change of the magnetic needle [proxy for F10.7]: Thank you dr Svalgaard, just a question : are "magnetic needle" "Geomagnetic East component" and "F10.7 radio flux" "sfu flux" comparable all the times ? |
|
bradk
Level 3 Rank
Posts: 199
|
Post by bradk on Feb 6, 2011 21:31:07 GMT
Dr. Svalgaard-
Thanks! Since both the SSN drops and the sunspot area per spot drops in a period when the L & P effect is in effect, could that be used as aproxy going backward as well? I now understand that the magnetic needle is also a measure, and a better one, but...
|
|
|
|
Post by lsvalgaard on Feb 6, 2011 22:06:04 GMT
Dr. Svalgaard- Thanks! Since both the SSN drops and the sunspot area per spot drops in a period when the L & P effect is in effect, could that be used as aproxy going backward as well? I now understand that the magnetic needle is also a measure, and a better one, but... The sunspot area is also a good [better] measure of solar activity. |
|
