Superflares are very strong explosions observed on solar-like stars with energies up to ten thousand times that of typical solar flares. The stars in this class satisfy conditions which should make them solar analogues, and would be expected to be stable over very long time scales.
A Superflare star is not the same as a Flare star, which usually refers to a very late spectral type red dwarf. The term is restricted to large transient events on stars that satisfy the following conditions:
- The star is in spectral class F8 to G8
- It is on or near the main sequence
- It is single or part of a very wide binary
- It is not a rapid rotator
- It is not exceedingly young
Essentially such stars may be regarded as solar analogues. Originally nine superflare stars were found, some of them similar to the Sun.
Original superflare candidates
The original paper  identified nine candidate objects from a literature search:
|Star||Type||Detector||V (mag)||Flare Amplitude||Duration||Energy (erg)|
|Groombridge 1830||G8 V||Photography||6.45||ΔB = 0.62 mag||18 min||EB ~ 1035|
|Kappa1 Ceti||G5 V||Spectroscopy||4.83||EW(He) = 0.13Å||~ 40 min||E ~ 2 × 1034|
|MT Tauri||G5 V||Photography||16.8||ΔU = 0.7 mag||~ 10 min||EU ~ 1035|
|Pi1 Ursae Majoris||G1.5 Vb||X-ray||5.64||LX = 1029 erg/s||>~35 min||EX = 2 × 1033|
|S Fornacis||G1 V||Visual||8.64||ΔV ~ 3 mag||17 - 367 min||EV ~ 2 × 1038|
|BD +10°2783||G0 V||X-ray||10.0||LX = 2 × 1031 erg/s||~ 49 min||EX >> 3 × 1034|
|Omicron Aquilae||F8 V||Photometry||5.11||ΔV = 0.09 mag||~ 5 - 15 day||EBV ~ 9 × 1037|
|5 Serpentis||F8 IV-V||Photometry||5.06||ΔV = 0.09 mag||~ 3 - 25 day||EBV ~ 7 × 1037|
|UU Coronae Borealis||F8 V||Photometry||8.86||ΔI = 0.30 mag||>~ 57 min||Eopt ~ 7 × 1035|
Type gives the spectral classification including spectral type and luminosity class.
V (mag) means the normal apparent visual magnitude of the star.
EW(He) is the equivalent width of the 5875.6Å He I D3 line seen in emission.
The observations vary for each object. Some are X-ray measurements, others are visual, photographic, spectroscopic or photometric. The energies for the events vary from 2 × 1033 to 2 × 1038 ergs.
The Kepler spacecraft is a space observatory designed to find planets by the method of transits. A photometer continually monitors the brightness of 150,000 stars in a fixed area of the sky (in the constellations of Cygnus, Lyra and Draco) to detect changes in brightness caused by planets passing in front of the stellar disc. More than 90,000 are G-type stars (similar to the Sun) on or near the main sequence. The observed area corresponds to about 0.25% of the entire sky. The photometer is sensitive to wavelengths of 400–865 nm: the entire visible spectrum and part of the infrared. The photometric accuracy achieved by Kepler is typically 0.01% (0.1 mmag) for 30 minute integration times of 12th magnitude stars.
The high accuracy, the large number of stars observed and the long period of observation make Kepler ideal for detecting superflares. Studies published in 2012 and 2013 involved 83,000 stars over a period of 500 days. The stars were selected from the Kepler Input Catalog to have Teff, the effective temperature, between 5100 and 6000K (the solar value is 5750K) to find stars of similar spectral class to the Sun, and the surface gravity log g > 4.0 to eliminate sub-giants and giants. The spectral classes range from F8 to G8. The integration time was 30 min in the original study. 1547 superflares were found on 279 solar-type stars.The most intense events increased the brightness of the stars by 30% and had an energy of 1036 ergs. White-light flares on the Sun change the brightness by about 0.01%, and the strongest flares have a visible-light energy of about 1032 ergs. (All energies quoted are in the optical bandpass and so are lower limits since some energy is emitted at other wavelengths.) Most events were much less energetic than this: flare amplitudes below 0.1% of the stellar value and energies of 2 × 1033 ergs were detectable with the 30 minute integration. The flares had a rapid rise followed by an exponential decay on a time scale of 1–3 hours. The most powerful events corresponded to energies ten thousand greater than the largest flares observed on the Sun. Some stars flared very frequently: one star showed 57 events in 500 days, a rate of one every nine days. For the statistics of flares, the number of flares decreased with energy E roughly as E−2, a similar behaviour to solar flares. The duration of the flare increased with its energy, again in accordance with the solar behaviour.
Some Kepler data is taken at one minute sampling, though inevitably with lower accuracy. Using this data, on a smaller sample of stars, reveals flares that are too brief for reliable detection with 30-min integrations, allowing detection of events as low as 1032 ergs, comparable with the brightest flares on the Sun. The occurrence frequency as a function of energy remains a power law E-n when extended to lower energies, with n around 1.5. At this time resolution some superflares show multiple peaks with separations of 100 to 1000 seconds, again comparable to the pulsations in solar flares. These observations suggest that superflares are different only in scale and not in type to solar flares.
Superflare stars show a quasi-periodic brightness variation, which is interpreted as evidence of starspots carried around by solar rotation. This allows an estimate of the rotation period of the star; values range from less than one day up to tens of days (the value for the Sun is 25 days). On the Sun, radiometer monitoring from satellites shows that large sunspots can reduce the brightness by up to 0.2%. In superflare stars the most common brightness variations are 1-2%, though they can be as great as 7-8%, suggesting that the area of the starspots can be very much larger than anything found on the Sun. In some cases the brightness variations can be modelled by only one or two large starspots, though not all cases are so simple. The starspots could be groups of smaller spots or single giant spots.
Flares are more common in stars with short periods. However, the energy of the largest flares is not related to the period of rotation. Stars with larger variations also have much more frequent flares; there is as well a tendency for them to have more energetic flares. Large variations can be found on even the most slowly rotating stars: one star had a rotation period of 22.7 days and variations implying spot coverage of 2.5% of the surface, over ten times greater the maximum solar value. By estimating the size of the starspots from the amplitude variation, and assuming solar values for the magnetic fields in the spots (1000 G), it is possible to estimate the energy available: in all cases there is enough energy in the field to power even the largest flares observed. This suggests that superflares and solar flares have essentially the same mechanism.
In order to determine whether superflares can occur on the Sun, it is important to narrow the definition of Sun-like stars. When the temperature range is divided into stars with Teff above and below 5600K (early and late G-type stars), stars of lower temperature are about twice as likely to show superflare activity as those in the solar range and those that do so have more flares: the occurrence frequency of flares (number per star per year) is about five times as great in the late-type stars. It is well known that both the rotation rate and the magnetic activity of a star decrease with age in G-type stars. When flare stars are divided into fast and slow rotators, using the rotation period estimated from brightness variations, there is a general tendency for the fastest-rotating (and presumably youngest) stars to show a greater probability of activity: in particular, stars rotating in less than 10 days are 20-30 times more likely to have activity. Nevertheless, 44 superflares were found on 19 stars with similar temperatures to the Sun and periods greater than 10 days (out of 14000 such stars examined); four superflares with energies in the range 1-5 × 1033 ergs were detected on stars rotating more slowly than the Sun (of about 5000 in the sample). The distribution of flares with energy has the same shape for all classes of star: although Sun-like stars are less likely to flare, they have the same proportion of very energetic flares as younger and cooler stars.
K and M type stars
Kepler data has also been used to search for flares on stars of later spectral types than G. A sample of 23,253 stars with effective temperature Teff less than 5150K and surface gravity log g > 4.2, corresponding to main sequence stars later than K0V, was examined for flares over a time period of 33.5 days. 373 stars were identified as having obvious flares. Some stars had only one flare, while others showed as many as fifteen. The strongest events increased the brightness of the star by 7-8%. This is not radically different from the peak brightness of flares on G-type stars; however, since K and M stars are less luminous than type G, this suggests that flares on these stars are less energetic. Comparing the two classes of stars studied, it seems that M stars flare more frequently than K stars but the duration of each flare tends to be shorter. It is not possible to draw any conclusions about the relative proportion of G and K type stars showing superflares, or about the frequency of flares on those stars that do show such activity, since the flare detection algorithms and criteria in the two studies are quite different.
Most (though not all) of the K and M stars show the same quasi-periodic brightness variations as the G stars. There is a tendency for more energetic flares to occur on more variable stars; however flare frequency is only weakly related to variability.
Hot Jupiters as an explanation
When superflares were originally discovered on solar-type stars it was suggested that these eruptions may be produced by the interaction of the star's magnetic field with the magnetic field of a gas-giant planet orbiting so close to the primary that the magnetic fields were linked. Rotation or orbital motion would wind up the magnetic fields until a reconfiguration of the fields would cause an explosive release of energy. The RS Canum Venaticorum variables are close binaries, with orbital periods between 1 and 14 days, in which the primary is an F- or G-type main sequence star, and with strong chromospheric activity at all orbital phases. These systems have brightness variations attributed to large starspots on the primary; some show large flares thought to be caused by magnetic reconnection. The companion is close enough to spin up the star by tidal interactions.
A gas giant however would not be massive enough to do this, leaving the various measurable properties of the star (rotation speed, chromospheric activity) unchanged. If the giant and the primary were close enough for the magnetic fields to be linked, the orbit of the planet would wrap the field lines until the configuration became unstable followed by a violent release of energy in the form of a flare. Kepler discovered a number of closely orbiting gas giants, known as hot Jupiters; studies of two such systems showed periodic variations of the chromospheric activity of the primary synchronised to the period of the companion.
Not all planetary transits can be detected by Kepler, since the planetary orbit may be out of the line of sight to Earth. However, the hot Jupiters orbit so close to the primary that the chance of a transit is about 10%. If superflares were caused by close planets the 279 flare stars discovered should have about 28 transiting companions; none of them actually showed evidence of transits, effectively excluding this explanation.
Spectroscopic observations of superflare stars
Spectroscopic studies of superflares allow their properties to be determined in more detail, in the hope of detecting the cause of the flares. Most studies have been made using the high dispersion spectrograph on the Subaru telescope in Hawaii. Some 50 apparently solar-type stars, known from the Kepler observations to show superflare activity, have been examined in detail. Of these, only 16 showed evidence of being visual or spectroscopic binaries; these were excluded since close binaries are frequently active, while in the case of visual binaries there is the chance of activity taking place on the companion. Spectroscopy allows accurate determinations of the effective temperature, the surface gravity and the abundance of elements beyond helium ('metallicity'); most of the 34 single stars proved to be main sequence stars of spectral type G and similar composition to the Sun. Since properties such as temperature and surface gravity change over the lifetime of a star, stellar evolution theory allows an estimate of the age of a star: in most cases the age appeared to be above several hundred million years. This is important since very young stars are known to be much more active. Nine of the stars conformed to the narrower definition of solar-type given above, with temperatures greater than 5600K and rotation periods longer than 10 days; some had periods above 20 or even 30 days. Only five of the 34 could be described as fast rotators.
All the stars showed the quasi-periodic brightness variations, ranging from 0.1% to nearly 10%, interpreted as the rotation of large starspots. When large spots exist on a star, the activity level of the chromosphere becomes high; in particular, large chromospheric plages form around sunspot groups. The intensities of certain solar and stellar lines generated in the chromosphere, particularly the lines of ionised calcium (Ca II) and the Hα line of hydrogen, are known to be indicators of magnetic activity. Observations of the Ca lines in stars of similar age to the Sun even show cyclic variations reminiscent of the 11 year solar cycle. By observing certain infrared lines of Ca II for the 34 superflare stars it was possible to estimate their chromospheric activity. Measurements of the same lines at points within an active region on the Sun, together with simultaneous measurements of the local magnetic field, show that there is a general relation between field and activity.
Although the stars show a clear correlation between rotational speed and activity, this does not exclude activity on slowly rotating stars: even stars as slow as the Sun can have high activity. All the superflare stars observed had more activity than the Sun, implying larger magnetic fields. There is also a correlation between the activity of a star and its brightness variations (and therefore the starspot coverage): all stars with large amplitude variations showed high activity.
Knowing the approximate area covered by starspots from the size of the variations, and the field strength estimated from the chromospheric activity, allows an estimate of the total energy stored in the magnetic field; in all cases there was enough energy stored in the field to account for even the largest superflares. Both the photometric and the spectroscopic observations are consistent with the theory that superflares are different only in scale from solar flares, and can be accounted for by the release of magnetic energy in active regions very much larger than those on the Sun. Nevertheless, these regions can appear on stars with masses, temperatures, compositions, rotation speeds and ages similar to the Sun.
Effects of a hypothetical superflare
Superflares increase the brightness of the star by up to 20 times its normal brightness and the luminosity by 1,000 times. They may last from a few hours to a week. The ozone layer of the Earth might be destroyed by the intense flow of charged particles produced by such a flare, and surface ice would melt on the daylight side of moons as distant from the sun as those of Jupiter, freezing again after the flare faded away. There is no evidence of superflares ever having occurred in the Solar System.
The 1965 story The Testament of Andros by James Blish contains scenes in which the world is devastated by a series of violent solar flares; these are revealed however to be part of the narrator's psychotic fantasies. Although Blish was an enthusiastic amateur astronomer, the publication date is before the first publication on superflares and he seems to have invented the concept himself.
In the film, Knowing (2009), the world was destroyed by a superflare in a story that was loosely based on parts of the Bible.
In the 2015 TV miniseries Heroes Reborn, a superflare-like event is part of the main plot.
- Schaefer, Bradley E.; King, Jeremy R.; Deliyannis, Constantine P. (1 February 2000). "Superflares on ordinary solar-type stars" (PDF). Astrophysical Journal 529: 1026–1030. doi:10.1086/308325.
- Maehara, Hiroyuki; Shibayama, Takuya; Notsu, Shota; Notsu, Yuta; Nagao, Takashi; Kusaba, Satoshi; Honda, Satoshi; Nogami, Daisaku; Shibata, Kazunari (24 May 2012). "Superflares on solar-type stars". Nature 485: 478–481. doi:10.1038/nature11063.
- Shibayama, Takuya; Maehara, Hiroyuki; Notsu, Shota; Notsu, Yuta; Nagao, Takashi; Honda, Satoshi; Ishii, Takako T.; Nogami, Daisaku; Shibata, Kazunari (November 2013). "Superflares on solar-type stars observed with Kepler I. Statistical properties of superflares.". Astrophysical Journal Supplement Series 209: 5. doi:10.1088/0067-0049/209/1/5.
- Notsu, Yuta; Shibayama, Takuya; Maehara, Hiroyuki; Notsu, Shota; Nagao, Takashi; Honda, Satoshi; Ishii, Takako T.; Nogami, Daisaku; Shibata, Kazunari (25 June 2013). "Superflares on solar-type stars observed with Kepler II. Photometric variability of superflare-generating stars: a signature of stellar rotation and starspots.". Astrophysical Journal 771: 127. doi:10.1088/0004-637X/771/2/127.
- Walkowicz, Lucianne M.; et al. (13 January 2011). "White-light flares on cool stars in the Kepler Quarter 1 data" (PDF). The Astronomical Journal 141 (2). doi:10.1088/0004-6256/141/2/50.
- Rubenstein, Eric P.; Schaefer, Bradley E. (February 2000). "Are Superflares on Solar Analogues Caused by Extrasolar Planets?". The Astrophysical Journal (http://adsabs.harvard.edu/cgi-bin/nph-bib_query?bibcode=2000ApJ...529.1031R&db_key=AST&high=38e0b7728728235: American Astronomical Society) 529 (2): 1031–1033. arXiv:astro-ph/9909187. Bibcode:2000ApJ...529.1031R. doi:10.1086/308326. Lay summary – Groombridge 1830. putative