From NOAA news: NOAA Scientist Finds Clue to Predicting Solar Flares
Forecasters at NOAA’s Space Weather Prediction Center in Boulder, Colorado.
High resolution (Credit: NOAA)
For decades, experts have searched for signs in the sun that could lead to more accurate forecasts of solar flares — powerful blasts of energy that can supercharge Earth’s upper atmosphere and disrupt satellites and the land-based technologies on which modern societies depend. Now a scientist at NOAA’s Space Weather Prediction Center and her colleagues have found a technique for predicting solar flares two to three days in advance with unprecedented accuracy.
The long-sought clue to prediction lies in changes in twisting magnetic fields beneath the surface of the sun in the days leading up to a flare, according to the authors. The findings will be published in Astrophysical Journal Letters next month.
“For the first time, we can tell two to three days in advance when and where a solar flare will occur and how large it will be,” said lead author Alysha Reinard, a solar physicist at NOAA’s Space Weather Prediction Center and the Cooperative Institute for Research in the Environmental Sciences, a partnership between NOAA and the University of Colorado.
Twisting magnetic fields beneath the surface of the sun erupt into a large solar flare, as shown above.
High resolution (Credit: NSF)
The new technique is already twice as accurate as current methods, according to the authors, and that number is expected to improve as they refine their work over the next few years. With this technique, reliable watches and warnings should be possible before the next solar sunspot maximum, predicted to occur in 2013. Currently, forecasters see complex sunspot regions and issue alerts that a large flare may erupt, but the when-and-where eludes them.
Solar flares are sudden bursts of energy and light from sunspots’ magnetic fields. During a flare, photons travel at the speed of light in all directions through space, arriving at Earth’s upper atmosphere—93 million miles from the sun—in just eight minutes.
Almost instantly the photons can affect the high-orbiting satellites of the Global Positioning System, or GPS, creating timing delays and skewing positioning signals by as much as half a football field, risking high-precision agriculture, oil drilling, military and airline operations, financial transactions, navigation, disaster warnings, and other critical functions relying on GPS accuracy.
“Two or three days lead time can make the difference between safeguarding the advanced technologies we depend on every day for our livelihood and security, and the catastrophic loss of these capabilities and trillions of dollars in disrupted commerce,” said Thomas Bogdan, director of NOAA’s Space Weather Prediction Center.
Reinard and NOAA intern Justin Henthorn of Ohio University pored over detailed maps of more than 1,000 sunspot groups, called active regions. The maps were constructed from solar sound-wave data from the National Science Foundation’s Global Oscillation Network Group.
Reinard and Henthorn found the same pattern in region after region: magnetic twisting that tightened to the breaking point, burst into a large flare, and vanished. They established that the pattern could be used as a reliable tool for predicting a solar flare.
“These recurring motions of the magnetic field, playing out unseen beneath the solar surface, are the clue we’ve needed to know that a large flare is coming—and when,” said Reinard.
Rudi Komm and Frank Hill of the National Solar Observatory contributed to the research.
NOAA understands and predicts changes in the Earth’s environment, from the depths of the ocean to the surface of the sun, and conserves and manages our coastal and marine resources. Visit us on Facebook.
Note to Editors: The paper has been accepted for publication in Astrophysical Journal Letters in February: “Evidence that temporal changes in solar subsurface helicity precede active region flaring,” by Alysha Reinard, Justin Henthorn, Rudi Komm, and Frank Hill.
Leif Svalgaard (12:12:18) :
Hung’s analysis seems less that kosher. For example, for the Nov 4th 2003 event he omits the X-flare on Oct. 22, omits one of the X-flares on Oct. 26, and omits two X-flares on Nov. 03. For his Apr. 15, 2001 event he omits the X-flares on Mar. 29, and April 2. Includes the M-flare on Apr. 9, but omits the large M-flares on Apr. 5. This wanton selection of what fits and omission of what does not, completely destroys his argument.
Were the magnitudes of these missing flares M9.0 or less?
tallbloke (12:20:31) :
Make your mind up. A couple of posts ago you said it was the peak at one year that was spurious. I’m talking about the separate peak at 400 days.
Since there is noise in the process unless you have a really strong signal [like the peak at 4000 days] you’ll usually get a series of close peaks around a ‘real’ peak. These are not individually significant, just shows that there is power around one year. That the peak is spurious is also seen from the fact that does not show up in the SORCE data.
As I said, pseudo-scientists are all too willing to ignore simple error considerations and ascribe reality to spurious things if these fit their agenda.
Paul Vaughan (12:35:35) :
Surely you don’t assume stationarity.
This whole thing started with Vuk suggesting a stationary signal at 27 days and 43 minutes extending over 38+ years. If so, that would have shown as a strong peak in the FFT, and it doesn’t.
tallbloke (12:40:16) :
Were the magnitudes of these missing flares M9.0 or less?
Of course not. They were X-flares. Except the ones in 2001 of higher magnitude that the M-flare mentioned by Hung. He lists four M-flares down to M3 [out of 19 total for which we have X-ray data] among his ‘successful’ predictions.
Leif Svalgaard (12:12:18) :
Hung’s analysis seems less that kosher. For example, for the Nov 4th 2003 event he omits…
Did you read the title of table III?
TABLE III.—PLANET AND SOLAR FLARE POSITIONS FOR MAJOR FLARE EVENTS WHERE THE LARGEST FINAL FLARE WAS APPROXIMATELY 30° FROM NEAREST HIGH-TIDE POSITION
Leif, I think you are confused.
the prediction is made in Appendix D.
tallbloke (12:20:31) :
“Interesting little peak around 400 days… (E+J synodic period)”
Leif Svalgaard (11:27:07) :
“The peak is completely artificial. Only shows in PMOD’s data. Not in SORCE that does compute the distance correctly.”
400 days has more significance than being spurious or artificial.
This graph shows filtered 400 day cycle (blue line, red line is 400 day COS wave). within SC17& SC23
http://www.vukcevic.talktalk.net/LFC5.htm
Some 2 years ago I analysed all 23 cycles my ideas and results can be found here:
http://www.vukcevic.co.uk/
follow link for ‘Solar subcycle’.
tallbloke (13:00:21) :
Did you read the title of table III?
So what? the flares he omit still belong to the same group he selects. Yet another example of how you’ll believe anything as long as it fits, and ignore the rest.
People are actually by Evolution conditioned to do this. We are likely to accept false positives because the penalty for doing so is so small. Every time we interpret some shadows in the grass as a tiger and flee, is better than to ignore one [because it is not statistically significant] that turns out to actually be a tiger.
In fact, his title and selection only weaken his case as no explanation [that I could find] is there for why the 30 degree difference is important. He even says: “In all five events, at the time of the final major flares, the event positions were about 30° away from the nearest planets. Whether these similarities were
the results of coincidence remains to be determined.”
tallbloke (13:27:27) :
the prediction is made in Appendix D.
That is where I looked. Nowhere there is a ‘prediction’ made. In my book it is a prediction if it is made public [either when made or by a sealed and notarized document], otherwise not.
vukcevic (13:29:22) :
400 days has more significance than being spurious or artificial.
The 400 ‘peak’ in the PMOD [the other TSIs don’t have it] is artificial because of known [very small] errors in the calculation of the solar distance. As I pointed out to tallbloke [and you nicely confirm], pseudo-scientists will willing accept erroneous data if it just fits.
Leif Svalgaard (13:38:08) :
“vukcevic (13:29:22) :
400 days has more significance than being spurious or artificial.
The 400 ‘peak’ in the PMOD [the other TSIs don’t have it] is artificial because of known [very small] errors in the calculation of the solar distance. As I pointed out to tallbloke [and you nicely confirm], pseudo-scientists will willing accept erroneous data if it just fits.”
This is nothing to do with TSI, it is the Sunspot numbers record, and as far as I know SSN has NOT “errors in the calculation of the solar distance”
Obviously you do not read beyond the first few words: here it is again so need no search:
This graph shows filtered 400 day cycle ( ! Sunspot numbers, not TSI !) (blue line, red line is 400 day COS wave). within SC17& SC23
http://www.vukcevic.talktalk.net/LFC5.htm
Some 2 years ago I analysed all 23 cycles my ideas and results can be found here:
http://www.vukcevic.co.uk/
follow link for Solar subcycle.
Leif Svalgaard (12:49:26) :
As I said, pseudo-scientists
Again, and again, and again.
Bye Leif.
Vuk mentions a NASA report on a 27 day period in the solar wind, this might be different to the 20-30 day observation in TSI fluctuations outside of sunspot activity.
I remember a story on WUWT dealing with a 27 day frequency picked up on earth via grids used to measure lightning. They were suggesting it was a solar signal.
Has there been a comparison of both data sets?
vukcevic (13:58:23) :
This is nothing to do with TSI
Of course, it has something to do with TSI. I showed a power spectrum of TSI, and you suggested that the 400-day peak was real or significant because it ‘matched’ your speculation about SSNs. Anyway, as long as you know that the TSI-peak is artificial everything is cool. Just don’t take that peak as ‘confirmation’ of your SSN speculation.
tallbloke (14:10:04) :
Bye Leif.
were it only true… 🙂
Clive E Burkland (16:55:16) :
Has there been a comparison of both data sets?
I don’t think so. There are lots of physical parameters that vary with an approximate 27-day quasi-period, so any number of plausible reasons can be put forward, but would be of little practical significance, except on the very occasions where the current is really huge. Amusingly, one of the foremost customers of predictions of geomagnetic activity is pigeon fanciers, who won’t do races if the geomagnetic Kp-index exceeds 4 [on the 0 to 9 scale used] as the pigeons get confused and can’t find their way home [it is claimed].
It has been known since the 1840s [not a typo] that geomagnetic [and therefore solar] activity causes electric currents in wires [e.g. telegraphs wires (in the 1840-90s) and now power lines].
Leif Svalgaard (17:51:08) “tallbloke (14:10:04) : “Bye Leif.”
were it only true… :-)”
When tallbloke, vukcevic, & others ‘challenge’ Dr. Svalgaard, Dr. Svalgaard divulges a LOT more info about various phenomena – useful for note-taking (…but there probably is an ‘optimal’ length to each round).
—
Updated in response to questions (in tallbloke’s & other forums) about calculations:
http://www.sfu.ca/~plv/SunspotsJEV.htm
Very interesting info on helical currents and toroidal structures.
Thanks all.
For an on-the-earth version, take a peek at the Focus Fusion
Reactor. focusfusion.org.
Paul Vaughan (20:51:08) :
When tallbloke, vukcevic, & others ‘challenge’ Dr. Svalgaard,
I thought it was the other way around 🙂
Dr. Svalgaard divulges a LOT more info about various phenomena
The info I provide is an attempt to educate our beloved pseudo-scientists. I realize that this is probably futile [having brought up four children], but such educational effort is my wont.
Updated in response to questions (in tallbloke’s & other forums) about calculations
In spite of the many decimals and the heavy smoothing [some approximation to running means] which btw enormously decreases the number of degrees of freedom [you might enlighten us with how many the resulting curves have], the eventual correlation with JVE is extremely poor, both in phase and amplitude, e.g. [showing one of your Figures – it would be a good idea to number them]: http://www.leif.org/research/Vaughan1.png
If the correlations were good, I would be the first to try to make sense of them physically [in preparation for my Nobel Prize which would surely be the result of a successful outcome], but they are just too weak to arouse interest.
Paul Vaughan (20:51:08) :
Updated in response to questions (in tallbloke’s & other forums) about calculations
In one of your figures you try to make the mismatch between the amplitudes smaller by plotting Log2(R+1). There is, of course, no justification for adding ‘1’, rather than 2.71828183 or 3.14159265328979 or any thing else not smaller than 1 [to make the argument positive]. This is what I call numerology.
Oliver K. Manuel (21:15:09) :
Did Leif forget that the camera used 171 Å filters to see emissions from iron ions, Fe (IX) and Fe (X).
No, but that has nothing to do with what the dominant ions are. Each ion will radiate in a given temperature range [1 million K for 171 Å, 2 million K for 284 Å], and the choice of those ions was to see structures having temperatures of a certain value [thus at different heights in the corona]. Other filters are used for other temperatures, e.g. to see the He(II) ions at 304 Å. The latter gives us a view at a temperature of ~70,000 K, but does not prove that the Sun consists entirely of Helium. I have a hard time believing that you do not know this [but I could be wrong], so shall characterize your statement as a cheap attempt of deception [with suitable apologies if the statement was based on ignorance].
Leif Svalgaard (21:41:01) “In one of your figures you try to make the mismatch between the amplitudes smaller by plotting Log2(R+1). There is, of course, no justification for adding ‘1′ […] This is what I call numerology.”
I’ve had those graphs of Log2(R+1)’ on file for many months – there’s no deceitful scheming going on here.
Elaboration:
The +1 is to:
a) avoid the singularity at Log(0).
b) Make 0 map to 0.
I discussed this with an academic statistician who agreed that 1 is a sensible choice. In my files, the “1” is adjustable, so I’m well-aware of the effect of adjustments (…but there are already enough graphs in the public summary).
The log is because the distribution is skewed (which corrupts analyses by violating statistical assumptions, regardless of what is or isn’t physical).
I prefer Log2(R+1)’, but I presented R’ because that is what most seem to prefer. I have other adjustable graphs on file for other positions on the “ladder of powers” (e.g. square root, cubed-root, etc.).
The preceding notes are not specific to sunspot numbers, but rather standard procedure for variables with skewed distributions.
Leif Svalgaard (21:41:01) “In one of your figures you try to make the mismatch between the amplitudes smaller by plotting Log2(R+1). There is, of course, no justification for adding ‘1′ […] This is what I call numerology.”
I’ve had those graphs of Log2(R+1)’ on file for many months – there’s no deceitful scheming going on here.
Elaboration:
The +1 is to:
a) avoid the singularity at Log(0).
b) Make 0 map to 0.
I discussed this with an academic statistician who agreed that 1 is a sensible choice. In my files, the “1” is adjustable, so I’m well-aware of the effect of adjustments (…but there are already enough graphs in the public summary).
The log is because the distribution is skewed (which corrupts analyses by violating statistical assumptions, regardless of what is or isn’t physical).
I prefer Log2(R+1)’, but I presented R’ because that is what most seem to prefer. I have other adjustable graphs on file for other positions on the “ladder of powers” (e.g. square root, cubed-root, etc.).
The preceding notes are not specific to sunspot numbers, but rather standard procedure for variables with skewed distributions.
–
Leif Svalgaard (21:27:32) “[…] decimals […] smoothing […] degrees of freedom […] poor, both in phase and amplitude […] Figures […] number them […] If the correlations were good, I would be the first to try to make sense of them physically [in preparation for my Nobel Prize which would surely be the result of a successful outcome], but they are just too weak to arouse interest.”
The phase-contrast is actually fairly stable for some of the indices (using cross-wavelet methods).
As indicated, the smoothing is to assist people who don’t understand cross-wavelet methods. Cross-wavelet methods work on RAW data — the repeat-smoothing over harmonics is just a way to make insights from wavelet methods more intuitive for a mainstream audience.
I’m not trying to sell ideas about JEV-R – just devising indices that capture things claimed by others to assist with assessment. My current interest is in phase-aware data analysis – not necessarily physics …but I appreciate your notes on physics.
As for your valid points about decimals & figure numbering: Rather than investing more in formalizing casual JEV-R notes (that are subject to change), I’d prefer to get on to cross-wavelet QBO/LOD/GLAAM/SOI analyses.
Leif Svalgaard (21:27:32) “[…] decimals […] smoothing […] degrees of freedom […] poor, both in phase and amplitude […] Figures […] number them […] If the correlations were good, I would be the first to try to make sense of them physically [in preparation for my Nobel Prize which would surely be the result of a successful outcome], but they are just too weak to arouse interest.”
The phase-contrast is actually fairly stable for some of the indices (using cross-wavelet methods).
As indicated, the smoothing is to assist people who don’t understand cross-wavelet methods. Cross-wavelet methods work on RAW data — the repeat-smoothing over harmonics is just a way to make insights from wavelet methods more intuitive for a mainstream audience.
I’m not trying to sell ideas about JEV-R – just devising indices that capture things claimed by others to assist with assessment. My current interest is in phase-aware data analysis – not necessarily physics …but I appreciate your notes on physics.
As for your valid points about decimals & figure numbering: Rather than investing more in formalizing casual JEV-R notes (that are subject to change), I’d prefer to get on to cross-wavelet QBO/LOD/GLAAM/SOI analyses.
Leif Svalgaard (21:41:01) “In one of your figures you try to make the mismatch between the amplitudes smaller by plotting Log2(R+1). There is, of course, no justification for adding ‘1′ […] This is what I call numerology.”
I’ve had those graphs of Log2(R+1)’ on file for many months – there’s no deceitful scheming going on here.
Elaboration:
The “+1” is to:
a) avoid the singularity at Log(0).
b) Make 0 map to 0.
I discussed this with an academic statistician who agreed that 1 is a sensible choice. In my files, the “1” is adjustable, so I’m well-aware of the effect of adjustments (…but there are already enough graphs in the public summary).
The log is because the distribution is skewed (which corrupts analyses by violating statistical assumptions, regardless of what is or isn’t physical).
I prefer Log2(R+1)’, but I presented R’ because that is what most seem to prefer. I have other adjustable graphs on file for other positions on the “ladder of powers” (e.g. square root, cubed-root, etc.).
The preceding notes are not specific to sunspot numbers, but rather standard procedure for variables with skewed distributions.
Leif Svalgaard (21:41:01) “In one of your figures you try to make the mismatch between the amplitudes smaller by plotting Log2(R+1). There is, of course, no justification for adding ‘1′ […]”
The +1 is to:
a) avoid the singularity at Log(0).
b) Make 0 map to 0.
I discussed this with an academic statistician who agreed that 1 is a sensible choice. In my files, the “1” is adjustable, so I’m well-aware of the effect of adjustments (…but there are already enough graphs in the public summary).
The log is because the distribution is skewed (which corrupts analyses by violating statistical assumptions, regardless of what is or isn’t physical).
I prefer Log2(R+1)’, but I presented R’ because that is what most seem to prefer. I have other adjustable graphs on file for other positions on the “ladder of powers” (e.g. square root, cubed-root, etc.).
The preceding notes are not specific to sunspot numbers, but rather standard procedure for variables with skewed distributions.