≡ Menu

Red Dwarf Planets: Weeding Out the False Positives

For those of you who, like me, are fascinated with red dwarf stars and the prospects for life around them, I want to mention David Stevenson’s Under a Crimson Sun (Springer, 2013), with the caveat that although it’s on my reading list, I haven’t gotten to it yet. More about this title after I’ve gone through it, but for now, notice that the interesting planet news around stars like Gliese 581 and GJ 667C is catching the eye of publishers and awakening interest in the public. It’s easy to see why. Planets in the habitable zone of such stars would be exotic places, far different from Earth, but possibly bearing life.

At the same time, we’re learning a good deal more about both the above-mentioned stars. A new paper by Paul Robertson and Suvrath Mahadevan (both at Pennsylvania State) looks at GJ 667C with encouraging — and cautionary — results. The encouraging news is that GJ 667Cc, a super-Earth in the habitable zone of the star, is confirmed by their work. The cautionary note is that stellar activity can mimic signals that we can interpret, wrongly, as exoplanets, and not every planet thought to be in this system may actually be there.


Image: The view from GJ 667Cc as presented in an artist’s impression. Note the distant binary to the right of the parent red dwarf. New work confirms the existence of this interesting world in the habitable zone. Credit: ESO/L. Calçada.

Remember that this is a system that was originally thought to have two super-Earths: GJ 667Cb and GJ 667Cc. The complicated designation is forced by the fact that the red dwarf in question, GJ 667C, is part of a triple star system, a distant companion to the binary pair GJ 667AB. It was just last year that the first two planets around GJ 667C were announced, followed by results from a different team showing five more super-Earths around the same star. See Gliese 667C: Three Habitable Zone Planets for my discussion of the apparent result, which at the time seemed spectacular.

The new paper from Robertson and Mahadevan takes a critical look at this system, examining the amount of stellar activity found in the host star and finding ways to study the average width of the star’s spectral absorption lines, which should flag changes to the spectrum being produced by magnetic features like starspots. Using these methods the team was able to remove the stellar activity component from the observed signals, allowing the signature of the real planets to remain while suggesting problems with the other candidates.

GJ 667Cc survives the test, a happy outcome for those interested in the astrobiological prospects here. GJ 667Cb also makes the cut, but Robertson and Mahadevan believe that planet d in this system, originally thought to be near the outer edge of the habitable zone, is a false positive created by stellar activity and the rotation of the star. As for the other planet candidates in this system, Paul Robertson has this to say in an online post:

The signals associated with them are so small that they cannot be seen with “industry-standard” analysis techniques, regardless of whether we have corrected for activity. However, considering how successful our activity correction has been at boosting the signals of real planets, the fact that we see no sign of any of these planet candidates after the activity correction leads us to strongly doubt their existence.

All of this should remind us not to jump too swiftly to conclusions about planet candidates, particularly given the sensitivity involved with the spectrographs used in our planet-finding work. Radial velocity data that looks strong can actually be the result of magnetic events on the surface of the star being observed. When Robertson and Mahadevan looked at Gliese 581, another highly interesting system because of planets possibly in the habitable zone, they found no sign of Gliese 581g, a controversial candidate whose existence is still being debated. In Gliese 581 and the Stellar Activity Problem, Robertson has this to say:

With an orbital period of 33 days, the controversial “planet g” also lies at an integer ratio of the stellar rotation period. Sure enough, no sign of g remains after our activity correction, revealing that it too was an artifact of magnetic activity. While this outcome is certainly disappointing for anyone hoping to find signs of life in the GJ 581 system, it is heartening to finally put the confusion and dispute surrounding this system to rest.

Moreover, another candidate potentially in the habitable zone, Gl 581d, falls back into the measurement noise, meaning that it was another signature of stellar activity rather than an actual planet. The red dwarf Gliese 581 is thus reduced to three planets in its system, and we’ve lost the best candidates for life. That may sound discouraging, but I think we can take heart from the fact that work like this shows we’re getting much better at eliminating false positives. Using these methods, real planets stand out in the data, which means we can more readily identify habitable zone planets as our spectrographic instrumentation improves.

The paper is Robertson and Mahadevan, “Disentangling Planets and Stellar Activity for Gliese 667C,” accepted for publication at Astrophysical Journal Letters (preprint). For Gliese 581, see Robertson and Mahadevan, “Stellar Activity Masquerading as Planets in the Habitable Zone of the M dwarf Gliese 581,” published in Science Express (3 July 2014).


Comments on this entry are closed.

  • Andrew LePage September 3, 2014, 10:43

    While the paper by Robertson and Mahadavan eliminates GJ 667Cd as a bona fide planet and casts further doubt on the existence of planets e through g, the simple fact of the matter is that the potential habitability of planets c, e and f (assuming the latter two even exist) has been wildly overstated. Given their rather substantial msini values (and unknown orbit inclination), recent analysis of Kepler data strongly suggests that it is more likely that these planets are NOT terrestrial planets at all never mind potentially habitable planets. It is more probable that they are mini-Neptunes or gas dwarfs:


    For those who are interested in Robertson’s earlier work on GJ 581, I’d also suggest the following:


  • Alex Tolley September 3, 2014, 10:43

    I don’t understand the nuances of the technique, but it seems to me that it may generate false negatives. It would be nice if this technique could be validated against more systems where planets have been confirmed.

  • FrankH September 3, 2014, 14:44

    With a mass > 4 Earths and a radius > 1.7 Earths, GJ 667Cc is definitely not Earth-like and may not even be a rocky planet.

    There have been very, very few truly Earth-like planets discovered by Kepler. Probably less than a handful; this planet is not even in the game.

    Leslie A. Rogers, “Most 1.6 Earth-Radius Planets are not Rocky”

  • Brian Swiderski September 3, 2014, 15:09

    “Under a crimson sun.”

    Sheer poetic license, of course: Red Dwarf stars aren’t actually red, and certainly not “crimson” – the “red” is just a spectral classification, not a description of how they look. Red dwarf stars are orange. Just as our “yellow dwarf” sun is white.

    Being cold enough to actually be red, a star would have to barely be hot enough to sustain fusion. In fact, brown dwarfs would be red, if I’m not mistaken.

  • Richard Black September 4, 2014, 4:16

    Yes, there are questions about the “true” actual existence of Gliese 581 d and g, but we are still in the infant stages of our search for “goldilocks zone” planets; particularly around m-type stars. Just being able to detect these planets at this stage of our technological prowess is a true testament to our abilities. Barely 20 years ago we couldn’t even imagine what we may see or find beyond our own solar system. Now we are finding new candidates for exoplanets practically, if not literally daily. There are going to be “bumps in the road, yes, and some planets we find may truly turn out to be false echos caused by H-alpha emissions, or micro-gravitational lensing, or some other reason; but it’s still practically trial and error to get to where we are at this stage, and so many professionals as well as amatures are giving their life to this cause to show that we are not alone in the cosmos. I salute them and say: while you may think no one cares there are at LEAST equally as many the do, if not more, as do not. I for one will hold on to the hope that Gliese 581g is still there and one of the top candidates for life elsewhere until further in depth study shows that there are truly only 3 or 5 planets in that system.

  • andy September 4, 2014, 5:20

    On a similar subject of controversial/disappearing planets, the latest orbital solution to HD 10180 reduces the system to six planets. Unfortunately they don’t much go into why the other planet candidates (b, i and j) vanished, only noting that the new model lacked their presence.

  • Andrew LePage September 4, 2014, 10:04

    Richard Black September 4, 2014 at 4:16

    “I for one will hold on to the hope that Gliese 581g is still there and one of the top candidates for life elsewhere until further in depth study shows that there are truly only 3 or 5 planets in that system”

    I wouldn’t waste too much effort hoping that GJ 581g actually exists. In addition to the work by Robertson et al. mentioned here, there have been others who have already reexamined the RV data and can find no evidence for its existence including the European HARPS team who have used their own independent set of RV measurements. Only the Lick-Carnegie Exoplanet Survey team that made the “discovery” have continued to insist it exists (although I believe they have recently changed their minds). This is not to say that habitable planets might not exist orbiting GJ 581, it just not going to be GJ 581g and it is unlikely it will be discovered using precision RV measurements. For a summary of the searches performed and all the analyses results of GJ 581, I’d suggest that your read the following:


  • ljk September 4, 2014, 11:29

    One Planet, Two Stars: A System More Common Than Previously Thought

    by SHANNON HALL on SEPTEMBER 4, 2014

    There are few environments more hostile than a planet circling two stars. Powerful tidal forces from the stars can easily destroy the rocky building blocks of planets or grind a newly formed planet to dust. But astronomers have spotted a handful of these hostile worlds.

    A new study is even suggesting that these extreme systems exist in abundance, with roughly half of all exoplanets orbiting binary stars.

    NASA’s crippled Kepler space telescope is arguably the world’s most successful planet hunter, despite the sudden end to its main mission last May. For nearly four years, Kepler continuously monitored 150,000 stars searching for tiny dips in their light when planets crossed in front of them.

    As of today, astronomers have confirmed nearly 1,500 exoplanets using Kepler data alone. But Kepler’s database is immense. And according to the exoplanet archive there are over 7,000 “Kepler Objects of Interest,” dubbed KOIs, that might also be exoplanets.

    There are a seeming endless number of questions waiting to be answered. But one stands out: how many exoplanets circle two stars? Binary stars have long been known to be commonplace — about half of the stars in the Milky Way are thought to exist in binary systems.

    A team of astronomers, led by Elliott Horch from Southern Connecticut State University, has shown that stars with exoplanets are just as likely to have a binary companion. In other words, 40 to 50 percent of the host stars are actually binary stars.

    “It’s interesting and exciting that exoplanet systems with stellar companions turn out to be much more common than was believed even just a few years ago,” said Horch in a news release.

    Full article here:


  • ashley baldwin September 4, 2014, 13:34

    I have read “under a crimson sun” and would recomend it, although it is on the fuzzy intephase between advanced pop science and an undergraduate text book. In terms of Rodgers’ arxiv paper, the figure of 1.6 radius Earth as being the level at which most planets are no longer terrestial, this is at odds with Geoff Marcy’s review of the Kepler data which showed a cut off point for Terrestrial and mini Neptune “ice dwarfs” as 1.7 Earth radius. It should be remembered also that the data from Kepler is based on further away stars with consequently less of the M dwarfs like GJ 667. M dwarf protostars are as much or even greater producers of UV radiation that chases off the lightweight gases that form deep atmospheres around rocky cores, so it is likely the Gj planet is terrestrial. Good. Big planets are tectonicly active longer, helping maintain a secondary atmosphere against the M stars’ wild outbursts with volcanic outbursts and also helping hold on to those atmospheres with their greater gravity, and often forgotten fact. Everyone is obsessed with dynamo-produced protective magnetic fields. That said, as I recall it the HBZ of the 667 planet is at .4 AU, which is on the outer edge of the HBZ zone, where its greater mass will help as well as the greater absorption of the infrared from the parent star helping to reduce the ice albedo instability effect (whereby ice is maintained by its reflection of the stars insolation back into space, especially at visible wavelengths) and helping mitigate the greater distance. The distance might even put the planet outside the tidal locking zone too which will allow some planetary rotation and some sort of induced magnetic field. All in all this is exactly the type of planet we might just find life on, and just what missions like TESS are looking for.

  • ashley baldwin September 4, 2014, 13:54

    Indeed, to anyone interested in the habitability of M dwarfs, or in general, and with TESS drawing nearer, I would recommend Jim Kasting’s outstanding book “How to find a habitable planet” which includes an interesting section on the M stars’ suitability. I would also recommend the seminal arxiv papers “in situ models for planet assembly around cool stars” Hansen March 2014, “The habitable zone of Earth-like planets with different levels of atmospheric pressure” Vladilo Feb 2013 and “geodynamics and rate of volcanism on massive Earth-like planets”. Kite May 2009

  • andy September 4, 2014, 15:56

    It’s also interesting to note here that both the Gliese 581 and the Gliese 667 C results were both preceded by studies of correlated (red) noise that came to the same conclusion about the planet inventory of the systems concerned as the studies of the effect of stellar activity on the radial velocities.

    For Gliese 581, the noise study was Baluev (2013), and the results confirmed by the stellar activity study of Robertson et al. (2014). Likewise, this result for Gliese 667 C was previously found in the study of Feroz & Hobson (2014) of correlated noise in the radial velocity data.

  • Ron S September 4, 2014, 19:57

    From the paper:

    “The results of this study imply an urgent mandate for
    the acquisition of additional data for GJ 667C. In the
    case of GJ 581, time-series photometry revealed very
    low variability, indicating the magnetic activity creat-
    ing the RV signals of false-positive planets d and g may
    be caused by localized active regions without optically
    dark starspots. We are unaware of any such photome-
    try for GJ 667C, and therefore cannot conclude whether
    the activity-induced RV shifts are caused by traditional
    starspots or spotless active regions. Finally, more RV ob-
    servations are required to determine whether additional
    low-mass planets truly exist in this system.”

    There’s only so far one can keep going over the same data sets but with different tools and methods. So they’re right, we need new data to nail this one. Presumably the purported stellar features are gone now, unless they are similar to Jupiter’s red spot. Of course there may be new spots now but it should be possible to distinguish these if the period of the previously-measured features is sufficiently accurate.

  • ashley baldwin September 5, 2014, 6:33

    Thats the trouble with M dwarfs and also why so many respected authorities doubt their ability to host life baring planets. Their high envelope of convection combined with early rapid rotation rates gives rise to strong msgnetic fields, the flux lines of whjh trsp energy wbich is eventually released in high energy UV and X ray flares. Same as the sun but on a much greater scale. RV calculation of exoplanets is entirely dependent on stellar spectra soalready you can see why doubt occurs so often. Its for this reasons that as with and good experiment the more independent verification the better. As the spectral reading is determined by gravitational effects of the orbiting body too, the fact that in this case GJ 667 is part of a trinary system further complicates the results, which to the credit of the original team are largely vindicated. Its worth noting here that one of the biggest causes of “false positives” in RV discovery is so called “eclipsing binaries” , close stellar companion stars that alter the stellar spectrum exactly lime a potential planet. Most stars exist in psirs as is righly pointed out above so this confounding situation is common. Its the reason why no planets have yet been definetely disvovered even in the nearby alpha centauri system. That said, multiple star systens shoukdnt be ruled out as hosts for life, its the distznce thstvthe constuents are apart that determines whether they can have plsnets in the habitsble zone. Omicron or 40 Eridani is an eary K dwarf sbout 16 light years from Earth . It is part of a trinary system with a white dwarf and a small M4 ” flare ” star” . However, it is 400 AU from tbese two, so more tban capable , and suitsble for HBZ planets. For Trekky fans, its the system that Mr Spock’s beloved Vulcan was said to exist! Bigger telescopes pull in more light and allow greater spectroscopic resolution , the ESPRESSO RV spectroscope is due to see first light at the VLT in 2016 and will drive precision to a new level of accuracy . Worth reading about in wikipedia as it explains the wider aspect of RV science too. Time on big telescopes like the VLT and Keck is at a premium however so astronomers are often obliged to use smaller more accessible scopes and its this that can lead to variation, but not so nearly often as to make tbeir hard work not worthwhile. Famous planet hunter Paul Butler uses the old 4m Anglo Australian telescope in the outback for just this reason, acceping he may be wrong occasionally , but with the peer review system , knowing any glitches will be put right when time becomes available on bigger scopes.

  • ashley baldwin September 5, 2014, 14:18

    All this transit sciencee would much less important if we could direct imsgd exoplanets. This is easil possible from space. NASA ard currently reviewing the use of their two donated ” spy satellite” mirrors for use in astronomy . NASA are considering using one in a pre existing mission, WFIRST, along with a chronagraph to block out blinding starlight , for just this purpose. Budget $1.6 billion .The NRC, National Research Council, are worried this will add to cost escalation as with the JWST and are being obstructive. The WFIRST mission will discover Earh like planets and maybe even signs of life as it has a spectrograph to look for astmospheric
    biosignature gases. NASA have been forced to look at two cheaper
    alternatives ( budget $1 billion) with a coronagraph and a star shade but much smaller mirrors. In exoplanet l
    sciencee , mirror size is everything, bigger examples collecting more light . The two cheaper options are for proof of concept only really other than
    discovering new planets, that RV and Transit studies do already!
    Spectography is the key to characterisation. The two mirrors in the
    cheaper missions arent large enough for this and afterwards nothing else
    is planned for years. Get lobbying Congress guys. Budget decision 2017.

  • Andrew LePage September 7, 2014, 10:59

    A new article published today examines the history of claims made about the extrasolar planetary system of GJ 667C and presents the latest analysis of the data. Instead of there being as many of seven planets, there would appear instead to be only two in orbits that vary cyclically over just six months. And instead of there being as many as three potentially habitable planets, GJ 667C now appears to have no realistic habitable planet candidates.


  • Aleksandar Volta October 6, 2014, 22:30

    I think there is a lot of assumptions flying around about the planetary candidates and this new theory about mini-Neptunes and such. Until we can study the atmosphere in detail, or image directly, we can’t be sure about their composition.

    We have Mars, but then we have Titan. How similar are they in terms of their geology, atmospheric cover, and composition? We have Earth, and we have Venus. Practically same-sized worlds. But one is habitable, the other is not.

    There is nothing suggesting say .. Kepler-22 is a smaller version of Neptune, rather than a water world. We don’t know yet. It could go either way. And that counts for every other exoplanetary candidate. My point is, maybe a substantial fraction of them are gas dwarfs/mini–Neptunes, but a large fraction of them may very well be terrestrial, or even Earth-like.

    Just my 2 cents.