Support the Arctic Sea Ice Forum and Blog

Author Topic: Solar cycle  (Read 25363 times)

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 720
  • Likes Given: 35
Re: Solar cycle
« Reply #150 on: March 01, 2021, 05:35:02 PM »
Quote
“First they ignore you. Then they ridicule you. And then they attack you. And then they build a monument to you.” — from a 1914 speech by a union organizer for the Amalgamated Clothing Workers. Mahatma Gandhi never described satyagraha this way. Nasa/Noaa has been ignoring NCAR predictions to date.
Solar Cycle 25 is well underway though it is too soon to say if NCAR will outperform the NOAA/NASA ‘expert panel’. The first sunspots are being counted, tracked and characterized in near real-time (along with flares, coronal mass emissions and geomagnetic storms) by various institutions and talented citizen-scientists who use the free JHelioviewer app to produce remarkable graphics (that make clever use of the jpeg2000 spec) out of the many daily terabytes of incoming solar satellite data.

https://www.jhelioviewer.org/
https://doi.org/10.1051/0004-6361/201730893
https://twitter.com/hfsolar1
http://jsoc.stanford.edu/HMI/HARPS.html

Solar is light-years ahead of Arctic sea ice in terms of Big Data organization and dissemination, though it wouldn’t take much to adapt JHelioviewer to Arctic satellites because like AstroImageJ, it has already been forked to cell biology. Astronomy journals embed advanced graphics and use Dexter format suited to inline analysis; in the cryosphere, we are still taking page screenshots. Solar research is not driven so much by questions in fundamental physics as by practical applications, notably predicting catastrophic ‘space weather’ events affecting the terrestrial electric grid, orbiting satellite electronics and astronaut safety; its followers hope for big X-class flares just Arctic watchers want record melt seasons.

https://www.astro.louisville.edu/software/astroimagej/
https://doi.org/10.1051/0004-6361/201730893
http://cdsads.u-strasbg.fr/abs_doc/help_pages/dexter.html

Sunspots are currently catalogued as active regions by NOAA SWPC (reaching AR2807 on 02 Mar 2021, HARP 6285) though the nomenclature itself does not capture initial latitude or hemisphere much less birth, growth and decay processes. Watching high resolution movies of the photosphere quickly reveals the utter inadequacy of historic characterization (sunspot counts, areas, groups) not fully mitigated by centuries of dedicated observation and tabulation by designated official record keepers (today, the Royal Observatory of Belgium). Advanced satellite instruments collect vast additional detail that goes unutilized in 1D graphs of sunspot numbers.

The sun’s rotation can hide sunspot emergence and evolution from telescopes for half of each month. A major revision of sunspot counting methodology in 2015 did not elicit consensus (for good reason); integration of newly discovered observational records throw off thousands of prior publications; faux controversy continues over a missing cycle within the Dalton grand minimum which largely repairs even-odd Hale cycle pairing. So with the sunspot record not only an inherently noisy proxy for magnetohydrodynamics but that compounded by a record-keeping shambles, harmonic analysis can’t be expected to work miracles.

A Solar Cycle Lost in 1793-1800: Early Sunspot Observations Resolve the Old Mystery
L Usoskin K Mursula R Arlt G Kovaltsov
Astrophysical Journal Letters. 700: L154
https://iopscience.iop.org/article/10.1088/0004-637X/700/2/L154  (2009)

Sunspots are not that fundamental but instead one of many manifestations of solar dynamics taking place at deeper levels. Sunspot numbers are not the best benchmark for predicting duration, intensity or downwind impacts of the coming solar cycle yet they get inordinate attention because they have by far the longest observational record — all the way back to the invention of the telescope. The question is, how relevant are long-ago cycles? Auroras may have a longer written record as do Be-10 and C-14 in ice cores, tree rings and stalactites but these have calibration issues, only track limited aspects of solar activity and those indirectly, and have no prospects for prediction.

Sunspot records are perhaps too accessible — anyone with a phone can download data, find fancied periodicities, promote eccentric theories on social media, associate peaks with lost Roman battles, troughs with little ice ages and invent convenient global cooling conspiracies. No science background needed; it’s not too distant from the daily horoscope. What’s needed is a much broader range of trackables and full integration with the remarkable incoming satellite data.

The NCAR model of solar cycles will again use sunspot records because that alone picks up 25 eleven year cycles, yet a Hilbert transform cannot escape underlying mediocracy: the benefit of length is offset by worsening quality at these earlier times (though that record does reveal unexpected range of variation). So here it might be better to re-calibrate the time scale with F10.7 and other time series, possibly with cross-correlation optimization. No question, the sun has some  memory of the state that it’s been in (hysteresis), yet predictive relevance falls off rapidly after a Hale cycle or two, limiting predictive utility for SC25.

While Belgium’s SILSO provides updates of predictions once a month, those are in-house predictions that don’t include those of NCAR or NOAA/NASA. The latter has its own update site that does not include SILSO or NCAR. While the SW McIntosh group could update their terminator dates and prediction methods on twitter or in arXiv drafts that’s not been happening: the focus is on merits of the forecast. Predictions end when the cycle begins.

Thus it is worthwhile here, while retaining overall NCAR innovations, to apply better cross-correlated methods of harmonic analysis to wider choices of proxies, perhaps refining terminator dates and adding other internal benchmarks to allow accurate synchronization of all the observational records and thus improve the estimate of  total solar irradiance arriving from SC25 (along with solar wind effects on the stratosphere) and so the synergety with greenhouse gases and conventional global warming on Arctic Amplification in view of the increasing area of seasonally low albedo open water.

It’s almost fair to say that all 25 journal articles (of 35 page length!) from the NCAR group could be condensed into a single stack of many co-registered line graphs plus a good explanatory caption. As an example, the He abundance paper plots its 46 year time series vs solar wind speeds, SSN, solar minima, F10.7 and Lyman α to define five internal solar cycle time points termed ‘AHe Shutoffs’. These are not quite evenly spaced in time and have slightly variable offsets relative to (ill-defined) solar minima. Despite the complexity, this co-plots only 6 of the 32 commonly used solar cycle observables though it might be leverageable into a GIS-like stack.

However the very next graphic adds extreme ultraviolet bright-point for the central solar meridian as a 2D graph over latitude. These butterfly diagrams are more informative than whole-disk integrations even if the data only extends over fewer solar cycles because the differential latitudinal rotation of the solar atmosphere (fastest at the equator) underlies downstream proxies such as sunspot numerology.

The solar wind component might have been done better as a 2D surface to provide a more compelling visual for double-peaked Gnevyshev gaps in the odd cycles. The F10.7 proxy for solar EUV outperforms the seemingly redundant Lyman α but itself might be improved using the full set of wavelengths measured at the Nobeyama Radio Observatory. It is not the custom in solar physics to attach the actual csv data used to make the graphs as supplemental; instead the reader has to replicate the cut-out from large institutional database pointers which few will do.

When averaged over cycles, this approach is termed epoch analysis. It’s all about consistently remapping the time axis using internal markers such as terminators so each cycle is rescaled to the same length. The underlying assumption is that quasi-periodic solar behavior has an underlying genuine periodicity whose modulation by secular variation can be removed, improving predictability of the next cycle (after inversion back to conventional time).

An 11-year solar cycle makes no more sense than half a sine wave. The NCAR papers provide overwhelming evidence for the more fundamental 22-year Hale cycle, reviving a view already taken by Hale (fig.1) who first observed the solar magnetic polarity cycle via the Zeeman effect. Numerous consistent distinctions (eg Gnevyshev-Ohl rule and Gnevyshev Gaps) exist between even and odd numbered cycles that require two consecutive 11-year cycles to be combined for a return to the initial state. Even/odd pairs too has long been known. NCAR also revives Wilson’s 1985 idea of ‘extended solar cycle’ in which 11-year cycles don’t fully end at 11 years but rather are interdigitated for several years into the next.

On the 22-year cycle of solar activity
MN Gnevishev AI Ohl
Astronomicheskii Zhurnal 25 (1): 18 in Russian (1948)
oft-cited but not online even at sci-hub

It’s a bit mysterious why all this lost traction — the short answer is sunspots have cast too long a shadow. The early NCAR papers involve authentic rediscovery; more recent ones show increasing awareness of long-forgotten papers. Solar science goes back centuries; pre-WWII papers are not all digitized and are seldom google-searchable. There is nothing new under the sun — that phrase was invoked 29 times in Ecclesiastes 1:9 over two millenia ago.   

Here even/odd have nothing to do with intrinsic parity but merely on how sunspot cycles were sequentially numbered in the 19th century. Adding small Maunder cycles or including the extra Dalton cycle throw off the system. There’s no way to change the numbering frozen into thousands of already-published papers.

Given the numerous and consistent physical distinctions common to even cycles, is the correct pairing even-odd, odd-even or neither (daisy chain)? Since aspects of even cycles prepare the ground for the next odd cycle but not vice versa, that favors the even-odd partitioning. As a consequence, SC25 completes the 24-25 pair so itself has the properties of an odd cycle which must factor into the prediction. Note that both epochal and modified epochal averaging (solar climatology) need to respect the even-odd distinction, ie averaging all cycles degrades the product as even and odd, being different, shouldn’t be mixed.

A similar question comes up for the north and south hemispheres, physically defined by the sun’s rotation axis. Although some features develop more or less in symmetry across the equator and indeed show coupling of polarity, there’s also a fair degree of distinctive and persistent asymmetry possibly related to differential latitudinal rotation speeds, large scale convection cell disorder or relic magnetic field history. The known periodicity is that of one Hale cycle.

There’s been talk of a double Hale cycle based on sunspot diameter periodicity. And in view of the better supported 88-year Gleissberg cycle, one wonders if four pairs of Hale cycles should be grouped, especially if the phase of the Gleissberg cycle begins and ends synchronously (or with fixed offset) with the first and last Hale cycle. The most recent NCAR paper uncovers this approximate periodicity as the leftover trend from the Hilbert transform. It’s not clear if the Gleissberg cycle will significantly moderate SC25 nor how much it matters overall to space or earth weather.

Some 32 solar observables have online databases providing multiple-cycle time series of variable length, quality and applicability. The NCAR group has assimilated many of them into their terminator-based scheme, some subject to their own Hilbert transforms and others just going along for the ride. In subsequent posts, the table below of 33 potential solar cycle markers will be completed to list all the co-plotting of variables in all the papers to see if the full set can be reduced to a tall stack of single variable plots from which tthe highest resolution combinations can be tiled.

aa and Kp indices
aurorae borealis
bright point extreme ultraviolet
coronal holes
coronal mass ejections
coronal rotation variation
cosmogenic isotopes
ENSO ocean
equatorial magnetic cancellation
extreme ultraviolet coronal bright points
F10.7 microwave emission
g-nodes large scale granulation
galactic cosmic rays
Gleissberg cycle
green corona line Fe XIV
heliospheric current sheet
helium-to-hydrogen abundance
hemispheric asymmetry
interplanetary magnetic field
latitudinal drift butterfly diagrams
Lyman-α
photosphere surface magnatism
polar faculae
polar magnetic reversal
quasi-biennial oscillation
recurrent high-speed solar wind streams
solar cycle maxima and minima
solar filaments
solar prominences
solar wind speed
sunspot area
sunspot group numbers GSN
sunspot number SSN

NCAR papers of SW McIntosh group:
— use ‘Bulk URL Opener’ plugin to open new web browser window with all 23
— read final published version when available
— see also: https://www.researchgate.net/profile/Scott-Mcintosh-2/publications
— see also: https://twitter.com/swmcintosh
— see also: https://www.youtube.com/watch?v=lRNJPkQPo_g&feature=youtu.be

2021.1   https://arxiv.org/pdf/2012.15186.pdf
2021.2   https://arxiv.org/pdf/2101.02569.pdf
2021.3   https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2020EA001223
2020.0   https://tinyurl.com/p4tndz5u 
2020.1   https://link.springer.com/article/10.1007/s11207-020-01723-y
2020.2   https://iopscience.iop.org/article/10.3847/1538-4357/abb9a1
2020.3   https://sci-hub.se/10.1007/s11207-020-01731-y
2020.4   https://arxiv.org/pdf/2010.06048.pdf
2020.5   https://arxiv.org/pdf/1909.06603.pdf
2020.6   https://arxiv.org/pdf/2006.04669.pd Helium Solar Wind SSN F10.7 Lyman-α EUV
2020.7   https://doi.org/10.1029/2020GL087795
2020.8   https://iopscience.iop.org/article/10.3847/1538-4357/abb9a1/pdf
2020.9   https://iopscience.iop.org/article/10.3847/2041-8213/abce69
2019.1   https://doi.org/10.1007/s11207-019-1474-y
2019.2   https://iopscience.iop.org/article/10.3847/2041-8213/ab0e0e
2018.1   https://www.frontiersin.org/articles/10.3389/fspas.2018.00038/full
2017.1   https://sci-hub.se/10.1038/s41550-017-0086
2017.2   https://www.frontiersin.org/articles/10.3389/fspas.2017.00004/full
2015.1   https://www.nature.com/articles/ncomms7491.pdf
2014.1   https://iopscience.iop.org/article/10.1088/2041-8205/784/2/L32
2014.2   https://iopscience.iop.org/article/10.1088/0004-637X/792/1/12/
2013.1   https://iopscience.iop.org/article/10.1088/0004-637X/765/2/146
2011.1   https://iopscience.iop.org/article/10.1088/2041-8205/740/1/L23/meta

Best twitter sites:

https://twitter.com/hfsolar1   citizen-scientist
https://twitter.com/nenecallas   citizen-scientist
https://twitter.com/halocme   scientist
https://twitter.com/swmcintosh   scientist
https://twitter.com/TamithaSkov   scientist
https://twitter.com/erikapal   scientist
https://twitter.com/WolfDotSolar   solar aggregator
https://twitter.com/_SpaceWeather_   solar and auroral aggregator
https://twitter.com/swmcintosh/status/1341126781059948544
https://twitter.com/ESASolarOrbiter/status/1359403953792909316
« Last Edit: March 03, 2021, 03:08:25 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 720
  • Likes Given: 35
Re: Solar cycle
« Reply #151 on: March 06, 2021, 09:28:22 PM »
It’s worth a closer look at the solar wind helium marker that heralds the end of a sunspot cycle ahead of its minimum because it potentially pushes the prediction prospects for solar cycles back earlier. The NCAR paper, while still an unreviewed arXiv preprint (with no alerting system for updates or journal final production), illustrates many of the co-plotting issues in this series of 25 papers. https://arxiv.org/pdf/2006.04669.pdf

Note the first two authors specialize in complex solar wind studies using ‘in situ’ instrumental data coming from close approaches of the Parker Solar Probe. This satellite can actually count helium nuclei (aka He-4++ alpha particles) arriving at the onboard faraday cup as normalized to background H+ protons in addition to looking at spectral lines and measuring solar wind speeds and structures. Note though PSP has only been operational for a fraction of a solar cycle and in close approach to the sun for only a fraction of that (green bar in fig.1).

The rationale for measuring the ratio of He to H isn’t provided but presumably has to do with solar wind energetics, notably distinguishing fast from slow regions of origin (coronal hole?) so here as potential proxy for state of solar cycle. Helium has 4 baryons but only two positive charges vs one:one for hydrogen so the former’s greater mass:charge ratio means it only attains photosphere escape velocity of 617 km/sec in more energetic events (though actual mechanisms of acceleration are far more complicated). The utility of other ratios to the solar cycle, such as those deriving from uncommon CNO atoms, isn’t known.

Although He-4 is a principal product of core nuclear fusion, what’s detected at the satellite mostly arose very early in the history of the universe. The core is not being studied here; that can only be done with helioseismetry (so far ineffectual) or via rare events at massively shielded neutrino detectors on earth. Far less abundant He-3 is not being studied here either; it (or Li-7) are used to count measuring neutrons generated by galactic cosmic rays that hit the earth’s atmosphere during periods when solar winds have knocked back he protective geomagnetic field.

Deuterium or H-2, initially present at 26 atoms per million H-1, is produced but consumed overall in solar core fusion. It’s present in solar wind but like He-3 not studied much. Borexino studies of CNO neutrinos have not yet been able to distinguish between high and low surface ’metallicity’ models of solar composition. The solar core where neutrinos are a byproduct of fusion cycles is still an observational black box, not varying much over mega-year time scales nor in its coupling with the next up convective layer.

Solar physics in the 2020s: DKIST, Parker Solar Probe, and Solar Orbiter
https://arxiv.org/pdf/2004.08632.pdf

Streamer blow-out coronal mass ejections at PSP
https://iopscience.iop.org/article/10.3847/1538-4365/ab6ff9/meta

Advanced image re-processing of ‘stealth’ CME via MGN, NRGF and DST filters
https://iopscience.iop.org/article/10.3847/1538-4357/aa6caa

Observational synergies of PSP + DKIST + Solar Orbiter
https://arxiv.org/pdf/2004.08632.pdf

Nice introductory overview
https://www.swpc.noaa.gov/sites/default/files/images/u2/Chapter_2_mod2-FINAL.pdf

Borexino neutrino highlights
http://borex.lngs.infn.it/

Fig.1 below combines the two main graphics of the paper by linearly up-scaling the time axis in the shorter cycle series to match the longer term one. The overlap now co-plots, sunspot cycle numbers 23-25 and minima, smoothed sunspot numbers, He/H excess as a function of solar wind speed and minima, F10.7, Lyman alpha and extreme ultraviolet bright points.

It’s great to have 8 variables co-displayed synchronously but with another 25 coming, there’s a strong need to isolate non-redundant variables into separate layers (and indeed attach as supplemental the cvs and grayscales needed to make them, for correlation analysis and error statistics). Oddly, terminators (NCAR’s previous emphasis) are not included nor has a Hilbert transform time adjustment been performed, even though the authors have published those several times previously.

Nothing is gained by including the Lyman-alpha data because it only mimics the far higher resolution, longer-term line of F10.7. The same can be said for the solar wind speed bins — the 330 line seems to suffice. However these aren’t known in advance so constitute contributions of the paper important to subsequent simplification. Note the Lyman line can be edited out of the graphic but it is about impossible to remove the extraneous wind speed bins — another reason to attach figures as a csv or gif stack.

In the bigger picture, this is similar to the Landsat-8 story: its green and blue channels are so highly correlated that no real purpose is served by retaining the blue. Instead some other channel, either infrared or microwave can be stubbed in before reassembly to a three channel RGB semi-false color with greater information content. With solar variables having phase offsets, the method could be lagged cross-correlation analysis as is done for sunspots and ever-changing total solar irradiance ie cross-wavelet transform (XWT) or wavelet coherence (WTC). There’s too much data saying the same thing and not enough getting at what underlies the Hale cycle.

https://iopscience.iop.org/article/10.3847/1538-4357/aa9bda
https://www.hindawi.com/journals/aa/2019/3641204/

Another variant is principal component analysis which can be done on any number of channels, here line graphs of all the proxies listed in the previous post. The method basically finds the optimal linear combinatorial mix of channels optimally describing the initial variance. One has to wonder what would result from this applied to all the concurrent full disk images at a given time, mixing and matching types and instruments. It would lead to yet more ‘indices’ which would not have immediate interpretability in terms of solar dynamo constituents.

http://jsoc.stanford.edu/data/hmi/images/latest/

On the extreme UV bright point display, the authors made five common graphic errors (hopefully to be remedied during peer review): first in blowing out the higher ranges of EB latitudinal false color scale to white 255, second in dithering to 9274 colors instead of restricting to the 10 of their palette, third in not having a recoverable grayscale upon desaturation, and fourth in applying a data-destroying one-sigma gaussian convolution kernel (described by a dead link but salvaged below) despite observing that it created a misleading edge artifact. Image enhancement options, including grayscale posterization, coloration and false coloring, are better left open so that downstream users with a wider range of sophisticated tools in (Astro)ImageJ and Gimp can make improved analyses. We see these exact same mistakes in Arctic sea ice journal imagery though not so often in the Antarctic.

https://docs.scipy.org/doc/scipy/reference/generated/scipy.ndimage.gaussian_filter.html

The inset (pixel-perfect, added in Gimp from the pdf) illustrates the key issue with the helium marker: the elapsed time between it and sunspot minimum is not quite constant, notably it’s shorter for cycle 23 but longer for cycle 24 though practically identical for 20-22. Two alignments are shown, one left-justified to emphasize variation in minimum and the other right-justified to compare helium heralding variation. The authors examine this closely in fig.2-3 of the paper, focusing on error analysis in how the helium cycle shutoff accuracy is undermined.

Many solar proxies are rather blurred out over time. There do not seem to be abrupt turning points or sharp transitions. Yet that is what is needed for precise re-callibration of Hale cycles so that all their internal markers align. This can be done in a piecewise linear fashion (one per Hale cycle) based on terminator cycle lengths. If relaible internal markers can be found, internal piecewise linear recalibration can be added. If enough of these can be found, they would accurately approximate any ‘holistic’ one-step non-linear adjustment such as Hilbert transform (induced by the left over Gleissberg trend).

The sunspot count itself is over-rated; its internal markers such as solar cycle minimum and maximum cannot be timed accurately. In some cycles, the sun’s quiet period extends for years, creating a shallow almost flat dip in its graph. Estimate of the bottom is then problematic and highly dependent on the choice of smoothing kernel width and weighting. Too often in solar, a simple-minded ‘trapezoidal’ convolution is used. That could be done better but it doesn’t get around the noisy nature of the primary sunspot data or its many variations.

The article here favors the helium shutoff as a relatively better method for anticipating the end of a cycle. The shutoff has a physical basis in the cancellation of opposing polarity magnetic bands as they near the solar equator, making helium a more direct proxy. However other markers based on other proxies such as EUV bright pints or HARP magnetograms may be better still. The need really is for interior markers all along the Hale cycle, not a lot of similar markers clustered near the end. Without a good way of adjusting each Hale cycle to a globally consistent time scale, (ie converting quasi-periodic to periodic with internal controls), it won’t be feasible to average over cycle epochs as needed for ‘solar climatology’ and next-cycle prediction.

The authors twice mention an important conclusion relevant to cycle 25 prediction, namely that according to the helium shutoff, cycle 24 ended significantly earlier than officially pronounced by SSN minimum methodology but the shortened date is not specifically provided as of 06 Mar 2021. Since length of the preceding (even) cycle is critical to predicting the length and intensity of the next (which we need for total solar irradiance effects on Arctic ice), this omission in the draft needs a fix because retrodiction is far less persuasive than prediction.

The authors have an earlier helium abundance paper published on 21 Sep 2011 in Astrophysical Journal Letters but a seemingly identical arxiv draft by the same authors time-stamped as last revised on 16 Nov 2018. The original letter has been referenced just 13 times in ten years, 4 of them self-citations. This paper addresses evolution of the solar wind, or rather how to get back to what it looked like closer to the time and place of its formation. That’s proven incredibly difficult to unravel but has been accomplished in a March 2021 Science paper from Hinode and ISSI scientists.

Part of the problem is that ‘in situ’ instruments on satellites only measure atomic abundances, ionization states and velocities far from the original sites of solar energetic processes. Solar plasma changes considerably in passaging from the lower solar atmosphere to an eventual open magnetic field line that can export it out to the more remote heliosphere (eg earth orbit).

This paper focuses on compositional evolution during the long activity minimum between solar cycles 23 and 24 to take advantage of reduced magnetic activity exposing more basal processes, notably from a reduced supergranular network length scale co-plotted earlier against solar wind speed, high speeds being associated with coronal holes. SOHO, Ulysses, and the stupidly named WIND satellite provided data on the (OMNI alpha/proto) helium abundance ratio, Fe charge state and Fe/O ratio in the solar wind at the L1 Lagrange point.

Because the data covers only 1970-2010 for helium vs fast/slow solar wind vs smoothed SSN, it cannot weigh in on solar climatology or predictions of Solar Cycle 25 without updating, ie the newer helium abundance article (20 Oct 2020) subsumes this piece. However the discussion, notably of the dominant contribution of the small-scale photospheric magnetic field and sensitivity of helium first ionization energy requirements (high FIP) to later remotely observed abundances in fast solar wind, seems to being holding up, there being two loosely coupled stages of heating and acceleration out. However the 2020 paper describes a distinct additional mechanism arising from the overlap of adjacent solar half cycles, namely equatorial flux cancelation of the older extended solar cycle during early in its solar minima.

https://iopscience.iop.org/article/10.1088/2041-8205/740/1/L23/meta published
https://arxiv.org/pdf/1109.1408.pdf active draft of already published
https://arxiv.org/pdf/2006.04669.pdf new draft not peer-reviewed
https://advances.sciencemag.org/content/7/10/eabf0068 major modern analysis

Subsequent posts will revisit projected solar outputs during Solar Cycle 25 and if notably higher, estimate those incremental effects on the earth’s atmosphere, clouds, ocean and land/ice surface. ‘Space weather’ per se, weakly tied to solar cycle specifics, is only indirectly relevant because it funds the incredible new instruments, now approaching closer to the sun and higher above the ecliptic line of sight than ever before. These will refine understanding of the underlying solar physics and so the expectations for the ongoing odd Hale cycle.

Ironically, the high-cadence flood of higher resolution satellite data has had little impact so far on Solar Cycle 25 predictions. That's because multi-cycle parameters have been preferred (for averaging) over current cycle observations, decades vs days being a poor match. However once key cycle markers have been determined, they could be measured much more accurately for cycles 23-24 as well as fixing the exact current state of cycle 25.

Solar inputs relevant to earth climate consist of 3 components: the familiar TSI (total solar irradiance @ top of atmosphere @ mean earth-sun distance), TSSI (total solar spectral irradiance, variation in wavelengths capable of stratosphere chemistry) and enhanced energetic galactic cosmic ray showers when the sun manage to disable the earth’s protective magnetic field. Thus the drama of X-flares, coronal mass ejecta, high and low solar wind radio interference and aurorae is secondary to cycle inferences that can be made from them.

We can already see this will amount to a dozen dueling statistical cross-correlation methods and fourier transform variants.

The text attachment below provides explanations of 120 essential solar science acronyms, the 10 best daily twitter sites (tested 06 Mar 2020), the 31 best online solar databases, and direct links to full text of the 23 most recent SW McIntosh/NCAR papers.
« Last Edit: March 07, 2021, 02:04:00 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 720
  • Likes Given: 35
Re: Solar cycle
« Reply #152 on: March 15, 2021, 12:39:54 PM »
Looking now at the journal situation publishing solar physics, sunspots and solar weather, it’s fair to say none of them offer drafts or open peer review (cf Arctic research at Copernicus), most don’t require attaching supplemental data allowing third party reproduction, updating or reanalysis of results, many seem to have loose standards on contribution merit, some don’t seem too concerned about verbosity, excessive self-citation and non-citation of conflicting or earlier results, and all have outrageous pay-per-pdf and heavy charges for open access discouraging access and sharing.

If peer review can’t cut the clutter, why not just stop at a preprint server? With so many recent papers inevitably dead wrong about Solar Cycle 25, one wonders how many will be retracted. Most likely none, followup papers will instead tweak or de-emphasize previously favored solar observable analysis while praising early pioneering efforts (their own). IOP journals are moving to double-blind peer review (both authors and reviewers masked) which might help.

On the plus side, most use arxiv as a central preprint server (unlike Arctic researchers) but not many have researchGate pages. It’s possible to sign up for an arxiv email alerting system, taking care to pick ‘sun’ AND subject class ‘astro-ph.SR’ within the broader topic of ‘astro-ph’. Yet even narrowed, who can keep up with a thousand solar papers/year volume — and how are we any the wiser for it? In the last twelve months, 58 papers addressed sunspots out of 714 on the sun. Some 60 papers in the last ten years corrected the historical record of sunspot numbers — maybe that should have been fixed first before writing so many papers using a flawed record. https://arxiv.org/search/advanced

Astrophysics uses a graphics-and-numerics bundling format called FITS. Rather like netCDF and HDF, the concept is to replace a folder full of files with a single immensely complicated master file in machine-readable binary. ImageJ and GIMP readily open FITS graphics, even at 32-bit. However cross-disciplinary climate research, for example total solar irradiance, is otherwise blocked by diverse packaging. Some journals offer FITS graphs in DEXTER format allowing data retrieval though this approach; it has not caught on elsewhere in scientific publishing. MathJax allows scalable display of equation-heavy text in browsers, a substantial improvement over fixed bitmaps.

Elsevier’s notion of Living Reviews is an excellent one — authors of the Solar Physics section can regularly update articles (though it’s not clear how many do nor how they’re further credited). https://www.springer.com/journal/41116)

The social media component, mainly Twitter, has fairly good coverage of daily solar developments and new research developments. It’s fair to say a handful of citizen-scientists, aurora enthusiasts and solar dot.com sites monitor incoming solar data far more regularly — and sometimes more knowledgeably — than solar physicists themselves. That’s reminiscent of Arctic satellites, buoys and extent/volume.

Curious as to how seriously the affected research community takes the NCAR group’s ideas behind the extended 22-year even-odd Hale cycle and consequent bold Solar Cycle 25 prediction, outside citations to the 23 articles can be tracked subsequent to the key 2014 paper, itself an energy-sapping two year struggle to get published. Frankly, there are not a lot in the expected places — the silly sunspot cycle is still sacred. For example, non-contentious scientific papers of yours truly have been cited 8,960 times, more than the 7,500 for those of SW McIntosh, deputy director of a major federal research facility and breakthrough solar analyst.

However this is rapidly changing as NCAR ideas get better articulated and extended to additional astronomical observables (ie co-plotted), with 3 AGU posters and 9 papers in 2020 alone and eminent outside authorities now joining in as co-authors. Scientists who pitched rival theories while ignoring or dismissing NCAR now seem keen to get on board since the SC25 prediction may be working out.

However that’s hard to track. The pace is more quarterly than monthly or daily. As of 10 Mar 2021, the sunspot for SC25 has reached 40, pushing outside the safety zone of the NOAA/NASA expert panel prediction (fig.1). NCAR itself had to revise its prediction due to a later-than-expected solar minimum for SC24 which lengthened that cycle so increased the delta for the next (relevant per a 2015 Hathaway empirical rule). However the predicted SC25 still remains short at 9.8 years, peaking early in July 2023 with sunspot number only slightly lowered from 234 to 222 (fig.2).

If so, total solar irradiance will still be quite high but not dramatically so. By way of contrast, alternative predictions called for a second Maunder-like minimum and an end to global warming if not the start of another ice age. Erratic sunspot academics and hired climate deniers won’t ‘move on’ in the face of an active cycle — they will simply pivot to blaming higher TSI instead of greenhouse gas emissions for worsening global warming.

http://www.realclimate.org/index.php/archives/2020/03/why-are-so-many-solar-climate-papers-flawed/
http://www.realclimate.org/index.php/archives/2006/09/the-trouble-with-sunspots/
http://www.realclimate.org/index.php/archives/2021/02/laschamps-ing-at-the-bit/
https://www.nature.com/articles/ncomms9611 extreme 100MeV solar proton events of July 774 CE and May 993 CE

NCAR vs the panel may be fully distinguishable on SSN by the end of 2021 but other aspects such as the special flux change at the 55º helio-latitude must wait for the arrival in Mar 2027 of the Solar Orbiter at 25º above the ecliptic. The end of major solar activity (arrival of pre-terminator) is now set for the first half of 2027 so obviously cannot be vetted until then, despite windowing implications for astronaut and grid safety.

More to the point, NCAR could hardly anticipate significant developments to be found in late 2020 and early 2021 major-journal papers from other research groups. These may provide ways and meanings of refining NCAR predictions or more accurate interpretation of the implied underlying solar physics that makes the NCAR program work. For example, June and Dec 2020 helioseismology papers significantly update understanding of convective cells underpinning solar meridional and torsional transport and a Jan 2021 dendrochronology advance extends solar cycle interval timing back to 969 CE via C-14 accelerator mass spectroscopy. Neither result could have been assimilated.

The 11-year cycle shows up clearly in the thousand-year tree ring data though only a wavelet transform has been done on it so far. It’s a much longer record than sunspots will ever be and not dependent on hand-drawn observational histories. Tree data cadence is seasonal rather than Carrington rotations; it’s attached as a supplemental file so ready for a Hilbert transform and epochal averaging after exceptional events are removed. There are plans to extend the tree ring data to the entire Holocene which could help with the longer Gleissberg cycle history. It’s not clear yet whether C-14 data fully supports an extreme Maunder minimum or the extra small cycle proposed for the Dalton.
 
https://science.sciencemag.org/content/368/6498/1469 June 2020
https://arxiv.org/pdf/2008.09347.pdf 30 Dec 2020
https://presentations.copernicus.org/EGU2020/EGU2020-11118_presentation.pdf poster
https://tinyurl.com/2ctp5jd2 the last thousand years of solar cycles from tree ring C-14 mass spec

Here it must be said that the SW McIntosh program goes well beyond descriptive phenomenology. Hints have been dropped to the effect that N and S hemisphere active regions are somehow coupled in simultaneous meridional outbreaks requiring revision of Babcock-Leighton magnetohydrodynamics. Further, 55º begs for an underlying explanation as critical solar latitude, presumably because it lies at a meeting place of differential solar advection cells.

The main two issues intertwined here are differential rotation of a gaseous body (fastest at the equator) wrapping a conventional (but perhaps off-center) axial magnetic field into toroidal and convective advection of magnetic features, equator-ward for one surface band and poleward for another, with return flow at depth providing conservation of matter.

There may be an NCAR paper in the works positing 4 convective cells (in a single layer) with two controlling feature migration about the equator and two the ‘rush to the poles’. Cell pairs are not quite identical in the two hemispheres, accounting for asymmetry in lead times to polar magnetic reversals and equatorial cancellation offsets. The origin of differences between consecutive even/odd cycles requiring their pairing into a Hale cycle has already been explained.

However certain chaotic aspects of solar behavior may remain forever inexplicable, diluting prospects for solar weather prediction, yet these extreme events can nonetheless be edited out of the historical record to refine predictive parameters applicable to assumed normal times for the current cycle.

NCAR is doing just that for the late-cycle Halloween Storms of 2003 (manifested as an abrupt slope change or 'knee' in Hilbert transform phase) which happened unpredictably during “declining sunspots in another quiet month in the unremarkable waning phase of an average solar cycle” as well as the remarkable Nov 1960 proton/alpha flares and consequent ground level neutron decays. The Bastille Day solar flare of 14 July 2000 was mid-cycle 23 rather than a quiet period outlier.

http://adsabs.harvard.edu/full/1962ApJ...136..534T
http://umich.edu/~lowbrows/history/mcmath-hulbert.html McMath Plage 5925
https://arxiv.org/pdf/1909.06603.pdf timing terminators 13 mentions of knees
https://arxiv.org/pdf/1502.07020.pdf DH Hathaway Living Review of solar cycle
https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005JA011137 Lynch 2005 helicity SC23

The extended 2003 event extracted so much magnetic flux and helicity that the next cycle was cheated out of two years length (Lynch 2005). In other words, knees could serve as detectors of these events in past centuries prior to invention and deployment of detectors, though the C-14 tree ring record can pick them up directly and indeed itself might manifest additional 'out-of-cycle' kinks (beyond the 1052 and 1279 CE events already discovered) if the more sensitive Hilbert transform were applied to its record.

Alternatively for purposes of forecasting, removal or deprecation of extreme events would have the effect of regularizing some of mysterious variation in past solar cycle hemispheric and length differences (ie, what does it take to make the knee go away or more broadly, to idealize or groom the cycle overall). This would have the benefit of improving Hathaway's empirical relation between consecutive cycles currently used for SC25 prediction. While 'reverse cherry-picking' of data (grooming) seems inappropriate at first, the bottom line is improved accuracy.

Note unpredictable extreme events within SC25 would throw off the forecast, though aid later with SC26. The sun is a long-running hydrogen bomb; its steadiness has some exceptions.

The New York Railroad Storm of May 1921 (coronal mass ejection) and the Carrington Event (solar flare) of 1859 could also be effectively removed, sharpening residual cycle homogeneity. Extreme events are not all that rare considering that relatively few are aimed effectively enough at the earth to leave terrestrial proxies behind.

The same might be attempted for the far more extreme events of 774, 993, 1052 and 1279 CE evident in tree rings though only slight improvements could result for Gleissberg or Hale cycle length stats.

The colossal 774 CE event is attributable to an extraordinary flux of high energy solar protons that enhanced ionospheric production of C-14, Be-10 and Cl-36 evidenced later in Greenland ice cores, rather than more galactic cosmic rays entering during a solar-driven diminution of the earth’s protective geomagnetic field.

https://www.essoar.org/doi/pdf/10.1002/essoar.10501832.1 Jan 2020 upper right corner
https://www.weather.gov/media/publications/assessments/SWstorms_assessment.pdf
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2019SW002250
https://arxiv.org/pdf/1908.10326.pdf
https://www.nature.com/articles/s41467-018-06036-0 thousand year history of cycles

With a long tradition of entrenched misconceptions, sunspots are an altogether inadequate metric for inferring solar structure and cyclic evolution. Active Regions provide a far better framework for solar magnetic phenomena that encompasses sunspots. Limiting context to white light — eyeball appearance in ordinary telescope — makes no sense. There’s a need to bring in all observables at all wavelengths preceding, accompanying and following sunspot birth and decay which include gamma, X-ray, extreme ultraviolet plus outgoing proton, alpha and electron fluxes. This complexity is explained in 31 evolutionary stages in an extraordinarily thorough Living Review (by the editor of Solar Physics):

Evolution of Active Regions
L Driel-Gesztelyi LM Green
https://link.springer.com/article/10.1007/lrsp-2015-1

Active regions don’t emerge at random locations but more commonly nest within a pre-existing magnetic environment formed by previous active regions, a determining factor of lifetime. Repeated episodes of ad hoc flux emergence within an evolving active region leads to increased magnetic field complexity, polarity cancellation and yet more magnetic activity. Thus active regions in quiet regions don’t tell the whole story as hysteresis must be factored in.

So forget sunspots, let’s track active regions instead! The daily Solar Region Summary is officially compiled in a totally half-@ssed way by NOAA/NWS at its Space Weather Prediction Center SWPC. Information is primitively presented in small separate text files of courier, one for each day back to 1996 in a graphics-free layout that hasn’t changed since the days of telnet. There’s no key to inexplicable acronyms.

However the many hundreds of text files can quickly be downloaded by ftp, concatenated sequentially in BBedit, with junk removed by an Excel sort. As of 10 Mar 2021, active region AR2809 is the latest to get assigned a tracking number. I AR2809 is 16th named active region of 2021. Active regions may or may not ever have had sunspots; they may consist solely of plage (defined by H-alpha images). Sunspots are tagged for one of seven magnetic classes by greek letter combinations and also receive one of the 60 valid Zurich-McI descriptive codes.

https://www.swpc.noaa.gov/products/solar-region-summary
https://scholar.afit.edu/cgi/viewcontent.cgi?article=1348&context=etd Zpc code explained on page 10
ftp://ftp.swpc.noaa.gov/pub/warehouse/2020/

Since the declared start of SC25 was Dec 2019, it is necessary to get those 31 + 366 + 69 to retrieve a full set. These begin with AR2753 and end at AR2809 and so consist of jsut 57 active regions for this quiet part of the early cycle. These generate 583 records because each AR is tracked each day it is visible from one of the four reporting terrestrial observatories.

Only alpha, beta and beta-gamma sunspot types of the seven possible have been seen so far in SC25 along with 19 of the 60 Zurich-PMcIntosh Zpc codes. Some 286 entries concern plage which may/may not develop sunspots at the next Carrington rotation (there are no satellites for the back side of the sun). Acronyms include Lo for Carrington longitude, area in millionths of a solar hemisphere, Z for Zurich-PMcIntosh system and so on.

Nmbr  Location  Lo  Area  Z   LL  NN  MagType
2808  N19E42  036 0050 Hsx 02  01 Alpha
2809  S21E13  065 0010 Bxo 03  02 Beta
2807  S18W78  156 [plage]
2804  N18  324 [due date for return]

There’s got to be a better way of displaying the daily status and full-spectrum/full disk properties and indeed various dashboards are maintained for that purpose. The most advanced of these, HARPS, extends the concept past NOAA’s visible-light to include the very complex data flow of the 45s cadence Helioseismic and Magnetic Imager (HMI) instrument on SDO. The future lies not with decades of daily hand-drawn maps and labor-intensive expert annotation in view of the incredible incoming volume of satellite data and need for automated processing.
 
https://suntoday.lmsal.com/suntoday/?suntoday_date=2021-03-14
https://observethesun.com/?current=2021-03-09&objects=2f&past=2020-03-14
https://www.solarmonitor.org/?date=20210314s
https://www.spaceweatherlive.com/en/solar-activity/region/12807.html
http://jsoc.stanford.edu/HMI/HARPS.html HMI Active Region Patches
 
Some of these dashboards are so complex that they hardly can load; data is not information. A better way to go is illustrated by a simple yet compelling design restricted to synoptic maps and their coronal holes (marked up by convolutional neural net AI). There’s no log-in, no fumbling with ftp or funky file formats — just buttons for each (or every) Carrington rotation 2098-2240 back to 2010 either as a classical JPEG showing CH boundaries overlaid on a synoptic strip map or a 32-bit FITS grayscale heatmap of the CH regions. The 143 synoptic maps butt up perfectly in ImageJ at 720x360 pixels allowing easy subsequent analysis of cycles and hemispheric asymmetry.

https://sun.njit.edu/#/coronal_holes
https://arxiv.org/pdf/2006.08529.pdf

Back in post #150, a list of 34 co-plottable solar cycle observables was compiled from published graphics. These have time (or modular time) as common abscissa. That list has now grown to 60 (some of these are just smoothing variations, subsequent posts). Given 23 papers and posters from the NCAR group with up to 22 graphs per paper, database organization is needed.

Which solar observables are used how often in the 23 NCAR papers, have any new developments gone unused, which are graphics vs line graphs, which have extractable data (vs data obliterated by over-plotting), which have been phased by a Hilbert or related harmonic transform allowing epochal averaging, which internal cycle markers can best be time-sharpened, and above all, which could provide mid-stream refinement of Solar Cycle 25 and consequent impacts on earth climate?

Frequency of use of solar cycle observables in graphics in last 3 NCAR papers:
11 Term,9 PTerm,7 F10.7,5 CycMin,4 PFS,3 EUVBP,3 SSNnh,3 SSNsh,2 CGL,2 Cycmax,2 Hal2003,2 LymanA,2 MAGdip,2 MAGqua,2 PFSav,2 PFSsmo,2 SSNday,2 SSNnhHilPhase,2 SSNshHilPhase,2 SSNsmo,2 Xflare,1 Ca2K,1 CR,1 EITcor,1 F10.7HilPhase,1 F10.7sm,1 F10.7smo,1 Fe X,1 Fe XVI,1 GNode,1 He II,1 HeA,1 HeAsdev,1 HeAShut,1 HeAsmo,1 HILamp,1 HILphase,1 MAGoct,1 PFR,1 SLWvel ,1 SSA,1 SSAave,1 SSAkurt,1 SSAskew ,1 SSN,1 SSN.NHsm,1 SSN.SHsm,1 SSNarea,1 SSNdayHil,1 SSNdayHilPhase,1 SSNhilEdge,1 SSNlat,1 SSNmon,1 SSNmonHil,1 SSNmonHilPhase,1 SSNnHil,1 SSNnhSmo,1 SSNshSmo,1 SSNsiz,1 SSNsm,total 110
« Last Edit: March 16, 2021, 03:11:16 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 720
  • Likes Given: 35
Re: Solar cycle
« Reply #153 on: March 22, 2021, 08:46:40 PM »
The previous post #152 looked briefly at whether the observational record could be ridded of one-off extreme solar storms to give a ‘better’ (straighter) Hilbert transform yielding more accurate solar cycle parameter estimation and ultimately a better prediction of Solar Cycle 25, one that is perhaps idealized but still correctable in near real time for unfolding exceptional events.

Hale cycles are only quasi-periodic; they vary perhaps 15% in length about a mean. The question is why, or rather how to best regularize the time series. The main outcome of the Hilbert transform of discrete observational variables such as F10.7 or sunspot numbers is a not-necessarily-linear map on time — the horizontal axis in common to all data plots — that brings all cycles to a common length (presented as frequency: -2π to 2π phase in some NCAR papers) allowing them to be stacked in register determining parameter distribution statistics, modifying a process called epochal superpositioning analysis in astronomy.

The Hilbert transform, an obscure choice given more transparent options such as expansion in derivatives, cross-wavelet transform or wavelet coherence (post #151), has been applied so far only to sunspot numbers/areas (± hemispheric asymmetry) and F10.7 cm radio emissions. These have the long curated cycle records needed (tree ring C-14, auroras, and ice cores have even longer) but are so predictably correlated that SSN can even extend the F10.7 back in time or assign its signal to hemispheres. Nothing is accomplished by doing both since outcomes are interconvertible via a low order polynomial. Indeed SSN plotted vs F10.7 calls for double regression (https://en.wikipedia.org/wiki/Total_least_squares open algo: scipy.odr).

Is the F10.7cm – Sunspot Number relation linear and stable?
F Clette  SILSO, Royal Observatory of Belgium
https://www.swsc-journal.org/articles/swsc/full_html/2021/01/swsc200021/swsc200021.html

The proprietary scientific software package IDL provides the key hilbert.pro command used in the NCAR papers to regularize quasi-periodic solar cycle lengths and align internal markers. IDL was developed in the 1970’s at the nearby Atmospheric and Space Physics lab but was soon privatized, sold and resold many times since by geospatial companies. IDL is used widely in astrophysics and signal processing. The package provides minimal documentation of the Hilbert transform; the 2020 NCAR paper on timing terminators just copies scrambled highlights from an esoteric wikipedia article written by and for mathematicians specializing in harmonic analysis. It’s better explained in later articles co-authored with S. Chapman who first suggested it.

A comparable command is available with a Matlab license or its open source version GNU Octave as well as Scipy python. There’s a helpful discussion of a practical example at StackExchange (https://tinyurl.com/awdbfku9). HT capability is offered online as a command line paste (https://octave-online.net/) though it’s not easy to feed the text boxes since the NCAR group doesn’t provide its time series data as supplemental and it is time-consuming and problematic to regenerate them from central depository urls, much less keep results near-real time post-publication.

The Hilbert transform surely improves on simple linear expansion or contraction of individual cycles based on elapsed time between loosely determined solar minima and maxima in sunspot numbers, in part by using better internal markers such as terminators, though so far no direct comparison has been made. SSN numbers are especially goofy around cycle minima (where the Wolf formula jumps from 0 to 11 with the first sunspot) with yet another revision announced for mid-2021 as additional older observations are digitized and the hand-off between Zurich and Brussels better scrutinized.

The F10.7 record is better but has issues of its own, not entirely resolved. These involve the retirement of radioastronomy pioneer Covington, moving the instrument from noisy 1947 Ottawa to defunded Algonquin to DRAO in backwoods Penicton BC, a brief era of bungled automation, the hand-off and later retirement of curator KF Tapping, and absolute calibration issues (resolved by 1966) relative to radio telescopes around the globe monitoring at the same wavelength. 

The history here is quite interesting. The choice of F10.7 cm seems eminently logical today but was hit upon entirely by accident in the early days of waveguide manufacture. Its variation over the solar cycle came as a surprise — the thinking back then was that solar emission at centimeter wavelengths would simply amount to black body emission from a ball of constant-temperature hot gas. It is not. Total solar irradiance too departs from Planck radiation over a Hale cycle especially at extreme ultraviolet wavelengths affecting the earth’s ionosphere.

https://en.wikipedia.org/wiki/Arthur_Covington
https://en.wikipedia.org/wiki/Dominion_Radio_Astrophysical_Observatory Penicton
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/swe.20064 KF Tapping 197 cites
https://www.swsc-journal.org/articles/swsc/pdf/2017/01/swsc170049.pdf  Nobeyama F30

Covington designed an instrument in 1951 that resolved portions of the sun's disk allowing directly measurement of coronal flux and inference of temperature above individual sunspots. However the F10.7 archive used in NCAR papers is full-disk, 3-hr cadence of 1-hr average, not resolving hemispheres or latitudes but adjusted to 1AU. (A second archive more appropriated for ionosphere studies does not adjust for earth-sun orbital distance variation.) It is more common today to measure a half-dozen measured wavelengths near F10.7 to monitoring different radiative processes; these records are perhaps too short for cycle predictive purposes.

The basic drawback of the F10.7 record, seemingly not addressed in NCAR papers, is that its numbers represent the sum of very different continuum emission physics, brehmstrahlung (free-free) and gyroresonance. Both arise from acceleration of electric charge, the former as electrons brake (decelerate) approaching each other and or as they move helically along magnetic tube field curves. These vary in different ways over different solar features and the Hale cycle and so need decomposition and separate treatment (the Hilbert transform is a linear operator).

Solar cycles have overall active and quiet periods sometimes punctuated stochastically by extreme electromagnetic flares and coronal mass ejections. These result in a lurch (‘knee’) in the transformed time map putting that particular cycle out of synch with the others. It’s normal near the maximum of a solar cycle to have notable peaks in active regions (corresponding to magnetic bands advected in solar latitude prior to cancellation of hemispheric matching polarities at the equator and poles) but not so normal to have them just before, during or after a quiet period minimum.

It may be the case that only extreme storms during quiet periods do the damage to the Hilbert transform. Thus as a first step it might be appropriate to ’take out’ cycle oddities such as one-off flares, coronal mass ejections and ground level neutron events (tabulated at wikipedia for 1582-2005) as most are near cycle maximum. If out-of-cycle events really contribute the lion’s share of distortion to next cycle’s length and amplitude, grooming the record prior to applying the HT would improve the product and prediction. It’s necessary here respect the even-odd pairing of two half cycles into one Hale cycle.

Of course, if all downstream effects on next-cycle length and severity could be removed, there wouldn’t even be any need for a Hilbert transform. The prediction would still be flawed to the extent rare but extreme stochastic events cannot be anticipated. Note though that in this scenario, since SC24 has completed and SC25 begun, there’s no benefit to a many-cycle record, meaning that the vastly better data on SC24 from Stereo-Ahead, SDO, Parker and SOHO can be brought to bear.

The table below shows 25 extreme solar storms listed at wikipedia (not complete or definitive) supplemented by decimal date of onset, solar cycle number, phase, latest article, whether discussed in NCAR as a Hilbert transform knee, and a brief descriptor. (The Dst index measures geomagnetic ring current energy, not solar parameters; most events are extended series of bursts; event characterization varies in quality.) Most but not all events occur near the solar maximum or adjacent declining phase, as can also be seen on the X-flare map in fig.2 of the 2020.h NCAR paper.

Individual events can be quite complex, extending over several days or even a month which complicates display over their solar cycle (or knee in Hilbert transform). Characterization varies according to the observers and scientific instruments of the day, often improved later by reanalysis that determines the underlying solar physics and ascribes comparative strength indices. Satellite-era events have far better data but only go back to SC19. For these reasons, a dedicated open-source paper, the more recent the better, has been assigned to each storm. Full texts are conveniently read as web browser tabs using a list-opening plugin.

1582.27  SC-3 dec 2019 - ICMEs aurora to 29∞ Dst -595
1770.79  SC02 max 2017 - red auroras Dst -1100   
1859.74  SC10 max 2013 + Carrington event X45 Dst -1760 
1872.16  SC11 dec 2018 - auroral oval 24º SSN max+2
1882.95  SC12 asc 2018 - Dst -386 aa 214 spot 2417 ºHem
1903.93  SC14 asc 2020 - Dst -531 500km/s CME aurora 44º 
1909.81  SC14 asc 2018 - Dst -595     
1921.46  SC15 dec 2019 + NY Railroad CME Dst -907 
1938.15  SC17 max 2020 - Fatima red aurora +19∞N 
1941.26  SC17 dec 2020 - Three storms including Jul Sep 
1956.22  SC19 asc 2020 - GLE#5 50x galactic cosmic rays 
1957.78  SC19 max 2018 - geomagnetic storm     
1958.18  SC19 max 2018 - SSN 172 F10.7 246   
1959.64  SC19 dec 2018 - geomagnetic storm     
1960.94  SC19 dec 1962 + McMath Plage 5925 cycle max+3
1967.47  SC20 asc 2016 - McMath 8818 Dst -387 ionizing X rays
1972.69  SC20 dec 2018 - McMath 11976 fast CME harbor mines
1989.29  SC22 max 2018 - Hydro-Quebec Dst -589 X6.5 flare   
1989.72  SC22 max 2015 - geomagnetic storm     
2000.35  SC23 max 2017 - plasma bubbles Dst -200   
2000.64  SC23 max 2011 - Bastille Day X5.7 flare DST -301
2001.36  SC23 max 2017 - plasma bubbles Dst -200   
2003.93  SC23 dec 2004 + Halloween storm 17 X47 flare Nov 4
2003.96  SC23 dec 2018 - intense SSN = 55, F10.7 = 118
2005.14  SC23 dec 2017 - 50x GLE
2012.56  SC24 asc 2013 - Dst −1182 nT double CME just missed earth
2015.20  SC24 max 2019 - St. Patrick fast CME C9.1 flare AR12297


1582  https://arxiv.org/pdf/1905.08017.pdf 
1770  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2017SW001690 
1859  https://www.swsc-journal.org/articles/swsc/pdf/2013/01/swsc130015.pdf
1872  https://iopscience.iop.org/article/10.3847/1538-4357/aaca40/pdf
1882  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2017SW001795
1903  https://arxiv.org/pdf/2001.04575.pdf
1909  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018SW002079
1921  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2019SW002195
1938  https://arxiv.org/pdf/2010.15762.pdf
1941  https://arxiv.org/pdf/2010.00452.pdf
1956  https://agupubs.onlinelibrary.wiley.com/doi/pdfdirect/10.1029/2020JA027921
1957  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018SW001945
1958  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018SW001945
1959  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018SW001945
1960  https://apps.dtic.mil/dtic/tr/fulltext/u2/281948.pdf
1967  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2016SW0014231972
1972  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018SW002024
1989  https://sxi.ngdc.noaa.gov/sxi_greatest.html
2000  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2017SW001674
2000  https://www.space.com/12278-bastille-day-solar-storm-anatomy.html
2001  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2017SW001674
2003  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018SW001945
2003  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2005JA011268
2005  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2017JA024125
2012  https://agupubs.onlinelibrary.wiley.com/doi/10.1002/swe.20097
2015  https://iopscience.iop.org/article/10.3847/1538-4357/ab06ff

NOAA rates space weather by Solar radiation events, Geomagnetic storms, and Radio blackouts using flux of 10+MeV ions, the Kp 3-hr planetary magnetic disturbance index (post #149), and GOES X-ray peak brightness respectively. The latter has a four solar cycle year record but needs to be presented along with extreme ultraviolet (for which F10.7 is a proxy). The EUV cadence is practically continuous, excessive relative to a 4090-day Hale cycle but better suited to a Hilbert transform than a dodgy integer sunspot count.

From NOAA’s table, it would come as no surprise if SC25 had 4 S4-5 radiation events (Kp8-9), several dozen geomagnetic storms near G5 and 9 R4-5 x-ray flares greater than X10 which pencils out to 2-3 events/yr around the predicted strong maximum in April 2025. Weaker A, B, and C-class flares precede M and X on the log scale, with SC25 already having had a B5.6 flare on 22 Mar 2021 followed by a slow CME that took 4 days to reach the earth. Note the distinction between events during SC25 affecting SC26, events detected by the current fleet of solar-orbiting satellites, and the far fewer events impacting earth systems. Thus the truly massive CME of July 2012 hit Stereo-Ahead but largely missed the earth.

https://www.swpc.noaa.gov/sites/default/files/images/NOAAscales.pdf
https://www.nasa.gov/mission_pages/sunearth/news/X-class-flares.html
https://iopscience.iop.org/article/10.3847/1538-4357/abb9a1  739 event ICME catalog 2007-2021
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2011SW000734  12% odds of another Carrington in ten years
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2019GL086524  odds studied in 2020 via aa index

S5 (Kp9) Extreme   <1 per cycle   G5 (10^5)    4 per cycle   R5 (X20)  <1 per cycle
S4 (Kp8) Severe     3 per cycle   G4 (10^4)  100 per cycle   R4 (X10)   8 per cycle
S3 (Kp7) Strong    10 per cycle   G3 (10^3)  200 per cycle   R3 (X1)  175 per cycle
S2 (Kp6) Moderate  25 per cycle   G2 (10^2)  600 per cycle   R2 (M5)  350 per cycle
S1 (Kp7) Minor     50 per cycle   G1 (10^1) 1700 per cycle   R1 (M1) 2000 per cycle


Time and instrument coverage of solar event observables needed for graph mark-ups:
 0567-2021: aurora lowest latitude and extreme color
 0929-1935: tree ring C-14 Hale cycles extendable to entire Holocene
 1755-2021: sunspots, areas, groups hand-drawn maps
 1925-2021: magnetic polarities by E Hale
 1932-2021: Kp planetary mid-latitude magnetic index
 1951-2021: Nobeyama NRO 3.2-10.7-30 cm
 1957-2021: AE auroral electrojet index
 1957-2021: Dst ring current index
 1957-2021: satellite era
 1990-2021: GOES X-flares
 1990-2009: Ulysses episodic solar data
 1994-2021: Wind at L1 ecliptic
 1995-2021: SOHO/LASCO C3 coronagraph
 1997-2016: ACE solar wind data
 2006-2021: Stereo-A
 2006-2014: Stereo-B
 2006-2014: Vex orbiting about Venus
 2010-2021: SDO extreme ultraviolet
 2011-2015: Messenger orbiting about Mecury
 2013-2021: Maven solar wind at Mars
 2016-2021: DSCOVR replaces ACE

A striking new consensus emerges from these papers: the celebrated Carrington Event was actually not all that unusual.  By 1859, an affectable infrastructure existed — telegraph wires — that wasn’t available earlier. Numerous events before and since have been proven [b}just as strong and even stronger[/b]. Very solid new data has been located in Asian auroral records; the lowest latitude recorded along with aurora color (notably red), unusual features and long duration can now reliably indicate storm type and strength.

It follows from high solar storm frequency, given the massive deployment of satellites and electric grids today, that space weather is not grant gimmickry but rather a legitimate alerting priority. Since major solar events can erupt and reach the earth in minutes (days for billion-ton clouds of magnetized plasma CMEs), damage-mitigating automatic responses have to be written into the devices; solar cycle predictions can only give a broad multi-year window of peaking. Case in point: the 23 July 2012 was far worse than the Carrington Event but occurred one week too late to hit earth.

The most common mistake is to think of these as super-sized solar events. In analyzable cases, they have not been that out of the ordinary. Rather, they are compound events in which early moderate CMEs sweep out the intervening sun-earth space and push back the protective magnetic shield somewhat, clearing the way for a next moderate CME — originating from the same active region — to have an outsized effect. So it’s about the frequency and physical basis of persistently active active regions.

The earth as terrestrial receiver only catches solar eruptions headed its way: short-lived events at higher solar latitudes on the dark side of the sun are not necessarily noticed. However these storms could still have a major effect on the solar cycle whether detected or not. Consequently, unavoidable limitations exist to the benefits of grooming observables for purposes of regularizing the cycles. However most major events are extended in time and space; some that don’t hit earth have been studied by satellites ahead, behind or over earth orbit. Again, extreme events are much more common than previously believed.

To the extent epochal average is helpful, a more nuanced understanding of what phenomena strongly perturb cycle length is needed to pick the optimal (combination?) of proxy observables, back to when that record starts, followed by a second sub-optimal combination with variables having longer tracking records, and so on back to really sub-optimal things like auroras and sunspot counts.

The last two graphics represent a DIY determination of the transition between SC24 and SC25, an even-odd pair Hale Cycle. Here 100x27 days Carrington synoptic maps of extreme ultraviolet AIA 193A courtesy of IDSEAR were analyzed in ImageJ's 2D plot tool. A 7th degree polynomial fit was used to find the very shallow minimum.
« Last Edit: March 24, 2021, 04:26:35 PM by A-Team »

vox_mundi

  • First-year ice
  • Posts: 5220
    • View Profile
  • Liked: 2639
  • Likes Given: 403
Re: Solar cycle
« Reply #154 on: March 27, 2021, 05:26:20 PM »
A Powerful Solar Storm Hit Earth Back In 1582
https://phys.org/news/2021-03-powerful-solar-storm-earth.html

"A great fire appeared in the sky to the North, and lasted three nights," wrote a Portuguese scribe in early March, 1582. Across the globe in feudal Japan, observers in Kyoto noted the same fiery red display in their skies, too. Similar accounts of strange nighttime lights were recorded in Leipzig, Germany; Yecheon, South Korea; and a dozen other cities across Europe and East Asia.

It was a stunning event. While people living at high latitudes were well aware of auroras in 1582, most people living closer to the equator were not. The solar storm that year was unlike anything in living memory, and it was so strong it brought the aurora to latitudes as low as 28 degrees (in line with Florida, Egypt, and southern Japan). People this close to the equator had no frame of reference for such dazzling nighttime displays, and many took it as a religious portent.

"All that part of the sky appeared burning in fiery flames; it seemed that the sky was burning," wrote Pero Ruiz Soares, an eyewitness in Lisbon, and the author of a 16th-century Portuguese chronicle. "Nobody remembered having seen something like that… At midnight, great fire rays arose above the castle which were dreadful and fearful. The following day, it happened the same at the same hour but it was not so great and terrifying. Everybody went to the countryside to see this great sign."

... The historical record seems to suggest that major storms like the one in 1582 are, at minimum, a once-in-a-century occurrence, and so we should expect one or more of them to hit Earth in the 21st century.

Hattori, Hayakawa, and Ebihara, "Occurrence of Great Magnetic Storms on 6–8 March 1582." ArXiv Preprint, 2019. (see page 22 for an awesome 16th-century illustration of the Aurora)
https://arxiv.org/abs/1905.08017

Carrasco and Vaquero, "Portuguese eyewitness accounts of the great space weather event of 1582." ArXiv Preprint, 2021.
https://arxiv.org/abs/2103.10941
“There are three classes of people: those who see. Those who see when they are shown. Those who do not see.” ― Leonardo da Vinci

Insensible before the wave so soon released by callous fate. Affected most, they understand the least, and understanding, when it comes, invariably arrives too late

vox_mundi

  • First-year ice
  • Posts: 5220
    • View Profile
  • Liked: 2639
  • Likes Given: 403
Re: Solar cycle
« Reply #155 on: April 05, 2021, 03:18:08 PM »
New Study Ties Solar Magnetic Variability to the Onset of Decadal La Nina Events
https://phys.org/news/2021-04-ties-solar-variability-onset-decadal.html

A new study shows a correlation between the end of solar cycles and a switch from El Nino to La Nina conditions in the Pacific Ocean, suggesting that solar variability can drive seasonal weather variability on Earth.

If the connection outlined in the journal Earth and Space Science holds up, it could significantly improve the predictability of the largest El Nino and La Nina events, which have a number of seasonal climate effects over land. For example, the southern United States tends to be warmer and drier during a La Nina, while the northern U.S. tends to be colder and wetter.

The study was led by Robert Leamon at the University of Maryland-Baltimore County, and it is also co-authored by Daniel Marsh at NCAR.

... In the new study, the researchers rely on a more precise 22-year "clock" for solar activity derived from the Sun's magnetic polarity cycle, which they outlined as a more regular alternative to the 11-year solar cycle in several companion studies published recently in peer-reviewed journals.

The 22-year cycle begins when oppositely charged magnetic bands that wrap the Sun appear near the star's polar latitudes, according to their recent studies. Over the cycle, these bands migrate toward the equator—causing sunspots to appear as they travel across the mid-latitudes. The cycle ends when the bands meet in the middle, mutually annihilating one another in what the research team calls a terminator event. These terminators provide precise guideposts for the end of one cycle and the beginning of the next.

The researchers imposed these terminator events over sea surface temperatures in the tropical Pacific stretching back to 1960. They found that the five terminator events that occurred between that time and 2010-11 all coincided with a flip from an El Nino (when sea surface temperatures are warmer than average) to a La Nina (when the sea surface temperatures are cooler than average). The end of the most recent solar cycle—which is unfolding now—is also coinciding with the beginning of a La Nina event.

"We are not the first scientists to study how solar variability may drive changes to the Earth system," Leamon said. "But we are the first to apply the 22-year solar clock. The result—five consecutive terminators lining up with a switch in the El Nino oscillation—is not likely to be a coincidence."

In fact, the researchers did a number of statistical analyses to determine the likelihood that the correlation was just a fluke. They found there was only a 1 in 5,000 chance or less (depending on the statistical test) that all five terminator events included in the study would randomly coincide with the flip in ocean temperatures. Now that a sixth terminator event—and the corresponding start of a new solar cycle in 2020—has also coincided with an La Nina event, the chance of a random occurrence is even more remote, the authors said. ...



Robert J. Leamon et al, Termination of Solar Cycles and Correlated Tropospheric Variability, Earth and Space Science (2021)
https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2020EA001223
“There are three classes of people: those who see. Those who see when they are shown. Those who do not see.” ― Leonardo da Vinci

Insensible before the wave so soon released by callous fate. Affected most, they understand the least, and understanding, when it comes, invariably arrives too late