≡ Menu

A Wary Look at Habitable Worlds

The confirmation of a planet circling two stars, recounted in these pages yesterday, is actually the result of a long process. Jean Schneider (CNRS/LUTH – Paris Observatory) noted in a follow-up comment to the Kepler-16b story that investigation of such systems dates back to 1990 (see citation below), while Alex Tolley has pointed out that the great space artist Chesley Bonestell was painting imaginary planets orbiting binary stars fully sixty years ago. So the idea isn’t new, but the confirmation was obviously useful, and in more ways than we might have expected.

For one thing emerging from the Kepler-16b paper is that the smaller of the two stars in this binary system, an M-dwarf, is now the smallest low-mass star to have both its mass and radius measured at such precision. The question of stellar mass and M-dwarfs is significant because a new paper by Philip Muirhead (Cornell University) and colleagues goes to work on the parameters of low-temperature Kepler planetary host stars and finds stellar radii that are roughly half the values reported in the Kepler Input Catalogue. The authors believe these values correlate better with the estimated effective temperatures (Teff) of these stars and suggest a striking possibility:

The effective temperatures, radii and masses of the KOIs imply different planet-candidate equilibrium temperature estimates, such that 6 planet-candidates are terrestrial-sized and have equilibrium temperatures which may permit liquid water to reside on the planet surface, assuming Earth-like albedos and re-radiation fractions. Scaling the Earth’s equilibrium temperature of 255 K by the orbital semi-major axis, stellar Teff and stellar radius of the KOIs in this letter, we find that KOIs 463.01, 1422.02, 947.01, 812.03, 448.02 and 1361.01 all have equilibrium temperatures between 217 K and 261 K: the limits of the habitable zone as described in Kasting et al. (1993).

This one has struck a nerve and it’s easy to see why, as we are suddenly looking at six Earth-like planets in the habitable zone of their stars. I’ve received quite a few links to the paper (and thanks to all who sent them, as this is often how I find interesting work!), but we first have to note a few qualifiers. The authors point out, for example, that this work assumes “the same albedo, re-radiation fraction and greenhouse effect” as are found in our own system, an assumption that may well be challenged for a terrestrial planet orbiting a red dwarf star.

I’m also cautious because the physical parameters of exoplanet-hosting stars are so crucial to our understanding of the detected exoplanets themselves. Here we run into issues, and the authors are quick to point this out. We have detailed information about the Sun, for example, that helps us calibrate models for Sun-like stars, so our analyses of mass, effective temperature, radius and other values seems logical and well-founded. But M-dwarfs are a different story because few such stars are both bright enough and close enough for us to obtain accurate parallaxes and direct measurement of their radius. The authors also note a discrepancy between radii as measured in eclipsing binaries and the predictions of at least some stellar evolution models.

The authors go on to say this:

Although there remains a monotonic correspondence between spectral type (the observational parameter) and effective temperature, Teff, the calibration of this relationship is not as advanced as it is for solar-type stars. M dwarf atmospheres are fully convective, rich in molecular absorption features and depart substantially from blackbody emission at all wavelengths…, so the empirical effective temperature scale is particularly challenging.

Muirhead and team went at their work using the TripleSpec Spectrograph at Palomar, observing 84 Kepler ‘objects of interest’ (KOIs) with effective temperatures (as described by the Kepler Input Catalogue) of less than 4400 K. The resultant mass and radius estimates derived in this paper reduce the size of the planet candidates to the Earth-analogue worlds reported here. This would obviously be a significant finding, but I think we have to wait for a response from the Kepler team, and in particular those involved with the Kepler Input Catalogue, to put the work into perspective. An error of this size would be extreme and, as at least one commenter has noted here, such an error should have shown up in the work on Kepler-16b, yet evidently did not.

I’m a writer, not an astrophysicist, so I’m intrigued but waiting for follow-up work to sort this out. This is, after all, how science works, an interplay of data and analysis that is adjusted as new data emerge. We’ll soon learn whether we have to modify our views of other Kepler candidates to match this result. In the meantime, I’m interested to learn what readers think of the Muirhead team’s analysis.

The paper is Muirhead et al., “Near-Infrared Spectroscopy of Low-Mass Kepler Planet-Candidate Host Stars: Effective Temperatures, Metallicities, Masses and Radii,” submitted to Astrophysical Journal Letters (preprint). On early work on circumbinary planets, see Schneider & Chevreton, “The Photometric Search for Earth-sized Extrasolar Planets by Occultation in Binary Systems,” Astronomy & Astrophysics 232, pp. 251-257 (1990). Abstract available.


Comments on this entry are closed.

  • tesh September 20, 2011, 10:50

    If eclipsing binaries are required for accurate measure of the Kepler stars, then might it not be a good idea to use all of the binaries observable (both within the Kepler field and when observed using other telescopes) to gauge the “average” radii of stars? There probably exists many combinations of eclipsing binaries that are observable by both Kepler and other telescopes.

    Qualifier… need less to to say I’m not an astronomer and this may already have been done…

  • tesh September 20, 2011, 10:51

    By combination of eclipsing binaries, I mean K-M or K-K or V-G and so on…

  • Erik Anderson September 20, 2011, 11:04

    Paul wrote: “This is, after all, how science works, an interplay of data and analysis that is adjusted as new data emerge.”

    With ultrahigh-precision parallax data, most of the confusion surrounding stellar luminosities can be sorted out. We can look forward to that with ESA’s Gaia mission.

  • Tony Sims September 20, 2011, 12:21

    With regards to the Planet orbiting two stars, is it not true that when three bodies are in orbit one of them can suddenly be ejected by the other two or is that not the case here because the planet has negligible mass when compared to its Suns?

  • Martin J Sallberg September 20, 2011, 12:54

    If the “habitable zone” is miscalculated it does not mean there is no habitable planets, it just means that the habitable planets are not the planets we had expected to be habitable. The chance of life remains virtually unchanged.

  • ljk September 20, 2011, 13:14

    Is the Gaia mission still on track? What is its current status?

  • Michael September 20, 2011, 14:53

    It might still be dangerous too orbit around a close binary even if their combined light output is favourable because their magnetic fields may interact with each other to increase solar storm activity

  • Enzo September 20, 2011, 17:10


    Launch appears to be delayed until March 2013 :

  • Enzo September 20, 2011, 17:22

    There seem to be a bit more information about Kepler-16b on its Wikipedia page than I’ve read in news articles :

    For example, the transit will stop being visible from 2018.
    Also, there is a very accurate diameter for the planet : even though it has the mass of Saturn, it is a lot smaller than Neptune.
    There might be a mistake there as the density given does not agree with mass and radius.

  • Enzo September 20, 2011, 18:04

    There is definitely a mistake in the Wikipedia page : it reports the planet radius as a fraction of Jupiter’s radius but this is the same of the red dwarf’s radius ratio with the Sun’s. It is very unlikely that both numbers are .22623 .

  • kurt9 September 21, 2011, 12:17

    My impression is that the Kepler data was showing about 1-2% of G stars having Earth-sized planets in the habitable zones around them. The problem is that there are lots of planets in the habitable zones around stars, but that most of them are Neptune-sized and likely Neptune-like. This kind of puts the kibosh on availability of habitable planets.

  • Bob September 21, 2011, 15:32


    How did you arrive at that rather pessimistic number? The mission is far from over and data on earth sized planets will likely come in more towards the end of the mission.

  • spaceman September 21, 2011, 20:28


    The 1-2% is an extrapolation based on the Kepler data released in February 2011 and is almost certainly subject to change. For all we know the number of earth-sized planets in the habitable zone of solar types could end up being an order or magnitude higher or lower than this estimate.

    One issue that I have been pondering since the February 2011 data was released is how the stars Kepler is obserrving seem to be noiser than expected. This means that Kepler will have to observe more transits to be sure that what it is seeing is an actually candidate planet instead of a product of stellar activity. In fact, one estimate out there is that Kepler will need 6 to 8 years of observations before it could confidently claim a detection of an earth-sized planet in the habitable zone of Solar-type star. So, we are a ways off from knowing the answer to the very important astrobiological question “what fraction of solar-type stars have earth-sized planets in their habitable zones?”

    Yet another issue I’ve been wondering about is related to the fact that, so far, Kepler seems to be finding many Neptune-sized planets and comparatively fewer earth-sized worlds. Some say this is an observational artifact that is worsened by the higher intrinsic stellar noise of the Kepler stars, but the longer the mission progresses and the more the pattern of fewer Earths and lots of Neptunes persists, the harder it will be to dismiss the trend as being a result of observational bias and/or noise. Here is an interesting article summing up the recent Wyoming based exoplanet conference from the following article:

    “Along with the discoveries came some sobering news. Rocky, Earth-like planets may be less common than many hoped, and unexpectedly ‘noisy’ stars are slowing the hunt.”

    Followed up by Geoff Marcy later in the article:

    “”Is Kepler having trouble detecting truly Earth-sized planets?””Or are they rare? We don’t know.”

    I’ll be curious to see how this plays out in light of future Kepler data releases. After all, why would nature decide to make more Neptunes than Jupiters, but not more Earths than Neptune?

    Here is the url for the aforementioned article: http://www.nature.com/news/2011/110919/full/477383a.html

  • Ronald September 23, 2011, 4:59

    @spaceman: “For all we know the number of earth-sized planets in the habitable zone of solar types could end up being an order or magnitude higher or lower than this estimate.”

    I would be surprised if it were an order of magnitude (10), I would rather expect within an order of 2, so from 0.5 – 2%.
    Exactly because it is an extrapolation, correcting for observational bias.

    Unless, of course, it is indeed due to an observational artifact with Kepler, noise etc.

  • ljk September 24, 2011, 21:44

    Update on potentially habitable exoplanets and the Habitable Exoplanet Catalog

    Posted on September 24, 2011 by Paul Scott Anderson

    There is a new report from the Planetary Habitability Laboratory (PHL), part of the University of Puerto Rico at Arecibo, regarding the number of currently known exoplanets which are potentially habitable. The findings are limited to confirmed exoplanets and exoplanet candidates, from the available Kepler data, which are within the habitable zone of their stars and with either a radius less than two Earth radii or a mass less than 10 Earth masses.

    As of now, the number stands at 16, two from the confirmed list of 687 and 14 from the candidates list of 1,235. That may not sound like a lot, but if, as now thought by Kepler scientists, that the number of exoplanets in our galaxy alone is in the millions if not billions, then you extrapolate from there…

    Full article here:


  • ljk September 27, 2011, 10:57


    One-Third of Sun-Like Stars Have Earth-Like Planets In Habitable Zone

    Astronomers have calculated the likelihood of finding Earth-like planets around other stars using the latest data from the Kepler mission.

    kfc 09/27/2011

    The Kepler orbiting observatory is specifically designed to find Earth-like planets around nearby stars.

    Earlier this year, the Kepler team released the mission’s first 136 days of data and it has turned out to be a veritable jackpot. In that time Kepler looked at some 150,000 target stars and found evidence for 1,235 potential exoplanets. That’s quite a haul.

    Since then, most of the work on this database has been to identify the characteristics of all these exoplanets. But such a large dataset also allows for statistical analyses too, from which various projections can be made.

    Today, Wesley Traub at the California Institute of Technology in Pasadena, reveals the results of just such a study. Traub has looked only at the stars that are most similar to the Sun, namely those with the classification F, G or K and worked out often various types of planets occur.

    The results are straightforward to state. Traub says that mid-size planets are just as likely to be found around faint stars and bright ones. By contrast, far fewer small planets show up around faint stars. That’s almost certainly because small planets are more difficult for Kepler to see.

    It’s also easier for Kepler to see planets that are closer to their stars because it looks for the tiny changes in brightness that these transits cause. That’s why almost a third of all Kepler’s detections orbit their star in less than 42 days. For the most part, these planets orbit too closely to be in the habitable zone.

    What interests most astronomers is how many exoplanets orbit at a greater distance, inside the habitable zone. Most of these planets are too far away from their stars to have been picked up by Kepler yet. But Traub says his data analysis provides a way to work out how many their ought to be.

    That’s because he’s found a power law that describes how the number of stars with a given orbital period. So all he has to do is assume a longer orbital period equivalent to being in the habitable zone to work out how many planets there ought to be at this distance.

    Here’s the answer: “About one-third of FGK stars are predicted to have at least one terrestrial, habitable-zone planet,” he says.

    So by this measure, there are plenty of other Earths out there.

    Ref: http://arxiv.org/abs/1109.4682: Terrestrial, Habitable-Zone Exoplanet Frequency from Kepler

  • Ronald October 12, 2011, 3:58

    Further to the previous comment by ljk on Traub’s paper ‘Terrestrial, Habitable-Zone Exoplanet Frequency from Kepler’, since Paul is still so busy with the 100 SS conference, I feel free to add some interesting highlights from the paper here;

    It concerns data from (only) Kepler’s first 136 days of operation and planetary periods < 42 days (in other words, innermost planets), extrapolating from those to the total population.

    ‘the frequency of all planets in the population with periods < 42 days is 29%, broken down as terrestrials 9%, ice giants 18%, and gas giants 3%’

    ‘the frequency distribution in terms of radius is independent of the frequency distribution in terms of period’, ‘size distribution of planets is independent of orbital period’; in other words any size planet can occur at any size orbit.

    ‘extrapolation to longer periods gives the frequency of terrestrial planets in the habitable zones of FGK stars as 34 ± 14%. Thus about one-third of FGK stars are predicted to have at least one terrestrial, habitable-zone planet’

    But: ‘Terrestrial planets are taken to be those with 0.5 ≤ r ≤ 2.0, corresponding to roughly 0.1-10 Earth masses’
    I find that a rather wide mass range, it also includes the super-earths and the smallest terrestrial planets down to about Mars size, too small to hold on long enough to a significant atmosphere.

    ‘the fraction of stars with terrestrial-radius planets is approximately the same for F, G, and K stars’

    But: ‘the fraction of ice giant planets varies by nearly a factor of two, being about 14% for F and K stars, but 24% for G stars’
    That is remarkable, there is a peak of ice (sub)giants, the Neptune class, at G, falling off both toward the hotter/brighter and cooler/dimmer sides.

    ‘the fraction of stars with (gas giants) is a rapidly-dropping function of spectral type, going from 5% around F stars to 2% around K stars;’
    This is entirely in line with earlier research: small/dim stars have a paucity of giant planets. But for larger stars there is a strong correlation with metallicity, which is not addressed in this present paper.

    With regard to period (orbit) distribution: ‘the population has a planet frequency with respect to period which follows a power-law relation dN/dP ∼ P^0.71’.

    In other words (also see Table 4 and Fig. 5), the number of (all) planets increases linearly with increasing period in a double-logarithmic graph, so I expect this number to remain more or less constant or even slightly increasing (?) per orbital (AU) range.

    How come this estimate of terrestrial planets in the HZ of solar type stars is so much higher (34%) than the previous estimates (about 1%, roughly varying from 0.5 – 3%)?

    The reasons I can find are:

    – Terrestrial planet is defined quite broadly as 0.1 – 10 Mearth (rather than 0.3 – 3 or so, as I would suggest).

    – Habitable Zone (Sections 10 and 11): the very high estimate of 34% is only true when the HZ in our solar system is taken as the average of a “wide” HZ, 0.72 to 2.00 AU, a “nominal” HZ, 0.80 to 1.80 AU, and a “narrow” HZ, 0.95 to 1.67 AU. If the HZ is defined (I think more realistically) as only the narrow one, from 0.95 – 1.67 AU, the % of solar type stars with terrestrial planets in the HZ drops off to 22% (Section 11), still quite high.

    – Probably the single most important reason for the huge discrepancy with earlier estimates: the earlier studies considered the results from larger orbits as valid as the smaller ones, not compensating for observational bias (larger orbits, smaller detection chance and very incomplete data). This paper does compensate and extrapolate for the larger orbits. This difference is a whopping factor 30 (Section 12, Fig. 5)! If applying the same assumption as the earlier studies this paper also comes to a low 1.1% solar type stars with a terrestrial planet in the HZ. However, the present assumptions seem much more realistic.

    If we reduce the HZ to 0.95 – 1.5 AU and the terrestrial planet definition to 0.3 – 3 Mearth, I guesstimate that the % of solar type stars with terrestrial planets in the HZ drops to about 5%, still higher than previous estimates.