The widely circulated Kepler results, announced yesterday, tell us that over twenty percent of Sun-like stars in the Milky Way have Earth-sized planets in the habitable zone, where liquid water could exist on the surface. Work out the math and it turns out that the nearest Sun-like star with a planet like ours in the habitable zone is probably on the order of twelve light years away, an energizing thought for those of us who ponder future technology and interstellar probes. Imagine: One in five Sun-like stars with a planet the size of Earth in the zone where liquid water can exist.
Image: Analysis of four years of precision measurements from Kepler shows that 22±8% of Sun-like stars have Earth-sized planets in the habitable zone. If these planets are as prevalent locally as they are in the Kepler field, then the distance to the nearest one is around 12 light-years.zone. Credit: Petigura/UC Berkeley, Howard/UH-Manoa, Marcy/UC Berkeley.
But how did we get here? Kepler, which was launched in 2009 to look for planets transiting their stars, examined over 150,000 stars for four years and turned up more than 3000 planet candidates. It’s been a fascinating ride, but finding ‘hot Jupiters’ and Neptune-class worlds and even intriguing super-Earths always reminded us that the primary goal was to learn what fraction of stars have Earth-sized planets at just the right temperatures for life. Ideally, retrieving data for G-class stars like the Sun would have helped us look for close twins of our planet.
Kepler’s malfunctions challenged but did not end that effort. The team involved in the present work includes Geoff Marcy and Erik Petigura (UC-Berkeley) and Andrew Howard (University of Hawaii, Manoa), who have been working with the 10-meter instruments at the Keck Observatory (Mauna Kea) to obtain data from the HIRES spectrograph, focusing on 42,000 stars in the Kepler field that are only slightly smaller and cooler than the Sun. 603 planets turned up, with 10 being between one and two Earths in diameter and orbiting in the habitable zone. Remember that only a small number of systems are oriented so that transits occur as viewed from Earth. The team’s algorithms yielded the estimate of 22 percent of all Sun-like stars with Earth-sized planets in their habitable zones, plus or minus eight percent depending on the habitable zone definition.
On the latter point, Erik Petigura defined the habitable zone for this study as that region where a planet receives between four times the light the Earth receives from the Sun and one-quarter of that amount. Kepler’s stuck reaction wheels have meant that extending the mission to analyze G-class stars like the Sun was not possible. Instead, the potentially habitable planets the team found in its survey all occur around K-class stars (Alpha Centauri B is the nearest example of a K-class star, though not in the Kepler field). The team’s analysis demonstrates that the results for K stars can be extrapolated to G-class stars, and thus we arrive at the 22 percent figure.
Telling us how common potentially habitable planets are around Sun-like stars is prime-time for Kepler, and now we have a reading that’s highly encouraging. As we look forward to missions to characterize exoplanet atmospheres and look for the signatures of life in their spectra, we can now assume that only a few dozen nearby stars will need to be observed before we detect an Earth-sized planet in the habitable zone, a fact that will play into the design of telescopes for such missions. Geoff Marcy, though, is quick to point out that just because a planet is Earth-like in size and in the habitable zone defined here, it isn’t necessarily life-bearing:
“Some may have thick atmospheres, making it so hot at the surface that DNA-like molecules would not survive. Others may have rocky surfaces that could harbor liquid water suitable for living organisms. We don’t know what range of planet types and their environments are suitable for life.”
The caution is understandable, and we are a long way from being able to make the kind of observations that help us pin down the presence of life on an exoplanet. We also need to remember that without information about the mass of these planets, we can’t say anything about their density and thus can’t be sure that they are in fact rocky worlds like our own. The discovery that Kepler-78b has the same density as the Earth, announced just last week, does tell us that at least some of these planets are likely to be rocky.
In any case, the idea that there are tens of billions of potentially habitable worlds in a galaxy of 200 billion stars is exhilarating, as Andrew Howard notes:
“It’s been nearly 20 years since the discovery of the first extrasolar planet around a normal star. Since then, we have learned that most stars have planets of some size orbiting them, and that Earth-size planets are relatively common in close-in orbits that are too hot for life. With this result, we’ve come home, in a sense, by showing that planets like our Earth are relatively common throughout the Milky Way Galaxy.”
We should also put these findings in a broader context. Red dwarf stars are not included in the study, but they represent 75 percent of the stars in the Milky Way. The question of whether life could exist around such a star, given the problems of tidal lock and stellar flare activity, is an open one, but we do know from previous work that 15 percent of these stars are expected to have Earth-sized planets in their own habitable zones (this is based on work by David Charbonneau and Courtney Dressing at the CfA; Ravi Kopparapu at Penn State obtained an even higher estimate). We don’t know yet whether life exists on any of the worlds around any of the stellar classes, but it does appear that the cosmos is stuffed with planets where the great natural experiments that lead to life can be run again and again.
The paper is Petigura et al., “Prevalence of Earth-size planets orbiting Sun-like stars,” Proceedings of the National Academy of Sciences, published online 4 November 2013 (abstract). Dennis Overbye’s report in the New York Times is well worth reading. I love this quote in Overbye’s article from Geoff Marcy: “This is the most important work I’ve ever been involved with. This is it. Are there inhabitable Earths out there? I’m feeling a little tingly.” Me too.
It’s also strange that Kepler should have been able to find Earth-sized (<1.2 R_E) planets up to a period of 240 days, but did not find any candidates above 50 days (acc. to the paper); among the confirmed (sub-)Earths none even has a period above Kepler-20f's 20 days.
This makes the extrapolation to the habitable zones in the paper rather dubious. It actually does not bode well for the existence of Earth-sized planets in the HZ’s. I think the true conclusion of the paper is quite opposite of its purported conclusion. Both the researchers themselves as well as the science journalists covering it seem to be afflicted by wishful thinking.
@Holger,
All I said about RV is that, where the data overlap with Kepler’s, they do not contradict it, they show the same empty area. So, HARPS, observing for ~10 years, capable of detecting mini-neptunes, has not found them for longer periods.
Nothing can be said about RV for earth size planets because it is not sensitive enough (i.e. it doesn’t overlap with Kepler’s in that particular area).
HARPS has increased precision since late 2012 and it can now detect “habitable super-earths”, whatever that means (<5Me ?) :
http://en.wikipedia.org/wiki/High_Accuracy_Radial_Velocity_Planet_Searcher
The overlap with Kepler will increase as results emerge in the future.
Another quirk on which both Kepler and RV seem to agree is the existence of warm jupiters (near the HZ).
There is another possible explanation for the absence of Earth Analogues in the Kepeler data. I admit Keplers’ sensitivity limits is propability ONE by far but:
The mechanics of solar system assemby from a solar nebula may make
it less likely to have smaller planets around the HZ of Most stars. Maybe there is something in the chemistry, orbits, solar flux in combination that make HZ’s less likely to contain Earth anlogues. Actually if you think about it Kepler found few Planets of ANY TYPE around the HZ of stars
from the K-G solar types.
I am late in this fascinating discussion, been very busy, so I am sorry I may be repeating things, but I cannot help it because this is my greatest fascination, the search for terrestrial planets in the HZ of solar type stars.
I went through the publication.
Ok, several commenters have already said things to the same effect, but I hope I can maybe add a few cents.
As mentioned by others, I also noticed in the publication (“Prevalence of Earth-size planets orbiting Sun-like stars”, by Petigura et al.) that the definitions used for ‘Earthlike’ and ‘Habitable Zone’ (HZ, Goldilocks zone) are VERY tolerant and quite different from the usual ones, even from Kepler’s own usual definitions;
– Earth size is here defined as 1 – 2 Re, which corresponds with about 1 – 8 earth mass (Me, at earthlike densities), whereas the usual Kepler definition of earth size is 0.8 – 1.25 Re (about 0.5 – 2 Me). This surprises me, in particular the very high upper limit, this is the usual Kepler definition for the upper limit of ‘super earths’, i.e. those were included in the total.
– The Habitable Zone (HZ) is defined here as from 0.25 – 4 times earthlike insolation (light). That corresponds to an HZ from 0.5 AU (way within the orbit of Venus at 0.7 AU and almost near Mercury at 0.4 AU) to 2 AU (way beyond Mars at 1.5 AU, almost at the Asteroid Belt). This, particularly the inner edge, is outrageous. Most researchers, and especially the recent update of the HZ by Kasting et al., put the HZ from about 0.95 to 1.7 AU (at the most).
Allow me to make a very quick back-of-the-envelope correction for both:
– For HZ, if the width is reduced from 1.5 AU to 0.75 AU, this cuts the number of candidates roughly in half (assuming more or less even distribution across that range).
– For mass, tricky, if we assume, just for the sake of ease, more or less even mass distribution across the whole mass range for terrestrials, we have to reduce the number of candidates again by a factor of 4 to 5.
Altogether, using the stricter and more reasonable definitions for earthlike and the HZ, we have to reduce the total number of candidate planets by a factor of about 8 to 10, leaving not 22%, but a more modest 2 – 3 % of sunlike stars with an earthlike planet in the HZ. Still impressive, but much more in line with previous research.
It would have been much more useful to publish an updated planet mass distribution, planet orbital AU distribution, and the combination of both: how planets of various mass classes are distributed across orbital (AU) distance.
I am so looking forward to the publication of the complete (4 year) observation results of Kepler, and the latest HARPS results. In fact, I expect more now the HARPS spectrograph and near future high precision spectrographs such as the VLT-ESPRESSO combination, and even more so: the future ELT-CODEX combo.
http://en.wikipedia.org/wiki/ESPRESSO
http://en.wikipedia.org/wiki/CODEX#Instrumentation
Great comments and discussion thread!
In particular, I think Holger, November 8, 2013 at 15:27, sums matters up well:
“All these figures are already corrected for observational bias. If the decreases of occurrence rates continue for higher orbital periods and slightly smaller sizes, the number of planets with 0.8-1.2 Earth radii around G-stars in the more conservative habitable zone (orbital period 300-600 days) will be below 1%.”
That is actually what I expect: we are not unique, but we are rare.
The main cause for this, though of course still uncertain, I *suspect* (for what that is worth) lies in the extreme commonness of the compact systems and medium-sized planets (Neptunes, mini-Neptunes/super-earths), which in turn, I *suspect*, lies in the absence of an inward and then again outward migrating gas giant which absorbs most planetary material in the inner system.
Question remains then, whether an earthlike planet in the HZ can only exist within a very specific and narrow set of chemical abundances (a kind of ‘chemical HZ’), or that such earthlike planets can also exist in compact systems among the (mini)Neptunes (e.g. there is a large gap around Tau Ceti in the HZ, between super-earth/mini-Neptune planets e and f, where a small undetected planet could exist), and/or that such earthlike planets can also exist in low-metallicity systems without any larger planets.
Enzo:
I thought you were talking about the “empty area” around Earth and Venus. If you’re talking about the region of “habitable zone mini-Neptunes” instead, it is also rather empty in Systemic’s diagram, but that’s from a year ago. And the HARPS upgrade you mention happened only then, so I don’t think there’s any “proven” emptiness there. As Greg says himself,
“The lower-right portion of the above diagram is incomplete, and there are a whole slew of observational biases at work, but nevertheless, the relatively depopulated divisions between the superEarth/subNeptunes, the hot Jupiters and the eccentric giants are real features of the planet distribution.”
So there’s no “real” depopulated region for smaller HZ planets due to lack of data in that diagram.
Of course Kepler and RV must agree on all data they both can measure (if they’re not really “broken”) – they’re each looking at “random” Milky Way stars, after all. Hot Jupiters, warm Jupiters and hot super-Earths are indeed proven features that our Solar System does not have.
Question: as I understand it, HARPS detects planets by a statistical fitting technique. They try fitting different orbits and masses until the remaining variance is minimised. I imagine this means that close-orbiting planets with higher signal get fitted first. The remaining variance will have more statistical noise, and as you continue fitting planets the remainng noise gets progressively larger relative to the signal you are trying to find.
That being the case, close-orbiting planets must interfere with finding those planets further out. So although the technique could detect Earths in isolation, when there are close-in super Earths, it cannot? Is this a valid arguement?
Holger:
Isn’t this just because the unexpected noise and shortened observation time has reduced the range of sensitivity? In other word, does the should have simply reflect unfulfilled wishful thinking?
This is true. However, because these are exactly those planets that produce the strongest signals, they are probably rarer than it now appears. Sure, attempts have been made to account for the observational bias, but it is not easy to do this completely and accurately.
kzb:
Good point. I am not an expert, but this makes sense to me.
Enzo: Kepler and RV have very similar detection characteristics: Both bigger planets and those closer to their stars produce higher signals. The methods agree in not showing Earth sized planets in the HZ, but not because there aren’t any. They have to agree because such planets are beyond detection limits of both methods. Of course, technically there could be a sudden drop-off just beyond, but I do not think we have any reason to think so.
@Ronald:
Thanks for the praise ;-)
“That is actually what I expect: we are not unique, but we are rare.”
Yes, me too.
“The main cause for this, though of course still uncertain, I *suspect* (for what that is worth) lies in the extreme commonness of the compact systems and medium-sized planets (Neptunes, mini-Neptunes/super-earths), which in turn, I *suspect*, lies in the absence of an inward and then again outward migrating gas giant which absorbs most planetary material in the inner system.”
I agree with the first suspicion, but not with the second: Jupiter never moved closer than 1.5 AU to the Sun, so this doesn’t explain the lack of hot (super-)Earths in our system. But maybe we had a hot Jupiter (or hot Neptune) originally, which dropped into the Sun before the inner planets formed from the little debris it left behind in the inner system?
kzb:
“That being the case, close-orbiting planets must interfere with finding those planets further out. So although the technique could detect Earths in isolation, when there are close-in super Earths, it cannot? Is this a valid arguement?”
That sounds plausible for RV discoveries. The endless dispute about Gliese 581g is good evidence for it…
Eniac:
“Isn’t this just because the unexpected noise and shortened observation time has reduced the range of sensitivity? In other word, does the should have simply reflect unfulfilled wishful thinking?”
No, unless Kepler is even less sensitive than they themselves reported in 2011. After noticing the unexpected noise, they said they’d need ~6 transits instead of 3 for Earth-sized planets. And 3 years, 11.x months observation time divided by 6 transits is about 240 days.
“This is true. However, because these are exactly those planets that produce the strongest signals, they are probably rarer than it now appears. Sure, attempts have been made to account for the observational bias, but it is not easy to do this completely and accurately.”
Then why do you trust this paper that claims to compute unbiased occurrence rates? ;-)
In fact, both this paper and previous studies found that the unbiased proportion of hot Jupiters is indeed low at ~1%, but that hot super-Earths occur at >50% of stars. So the Solar System is already known to be somewhat uncommon, though hot Jupiters may be rarer still.
Holger: “Then why do you trust this paper that claims to compute unbiased occurrence rates?” My trust is quite limited, sorry if I haven’t made that clear. It is just that their assertion also agrees with the default expectation that one should have in the absence of data, making their conclusions at least plausible.
I can live with >50% of stars being different from ours. At 95% I would begin to get concerned, and typical Rare Earth arguments require more like >99.9% ….
I am fully on board with the idea that nature favours planetary systems with close-orbiting planets, and the absence of such planets in our own system makes our system a rarer type.
However, the idea that such systems do not possess Earths in the HZ – I am not sure this is proven.
Several people agree that close-in planets interfer with the detection of planets further out by HARPS. But the same thing must also be true of Kepler.
Planetary orbital planes have a large degree of statistical correlation. Earths orbiting in the HZ will tend to be in same plane as close-orbiting super-Earths. So there is a greater than random chance that the Kepler signal from an Earth is superimposed on a much larger signal from a super-earth.
Eniac:
“My trust is quite limited, sorry if I haven’t made that clear. It is just that their assertion also agrees with the default expectation that one should have in the absence of data, making their conclusions at least plausible.”
It was tongue-in-cheek, but I inferred your trust from your statement (as I understood it) that there “cannot be any serious doubt” of the commonness of Earth analogues.
I don’t see how there can be a “default” expectation in the absence of data (other than eta_Earth>0). The Copernican Principle already breaks down when looking at the stellar type – our Sun is not an M-dwarf, but a much rarer G-star.
“I can live with >50% of stars being different from ours. At 95% I would begin to get concerned, and typical Rare Earth arguments require more like >99.9% ….”
I’m not advocating the “Rare Earth” argument. As I said above, I’d expect close Earth analogues at only ~1% of GK-stars based on Petigura’s data (so you may “get concerned”), but that’s still hundreds of millions of Earth analogues in the Milky Way…
kzb:
” So there is a greater than random chance that the Kepler signal from an Earth is superimposed on a much larger signal from a super-earth.”
With the transit method, “superimposition” can only happen if there are simultaneous transits of the two planets, can’t it? I think double transits should be very rare, except for the very hottest super-Earths like Kepler-80b which transit every day…
Holger:
The default expectation would be that planets are equally common in places where we can’t see them as they are in places where we can.
Statements like “[frequency of] close Earth analogues at only ~1%” are not very meaningful, because the number depends more on what you mean by “close analogue” than on anything real. To be more meaningful, numbers should be given as densities, i.e. number per orbital distance interval per planetary size interval. If you pick a wide range, like the authors of this paper, you get plenty of planets. If you pick a narrow range, you get very few. Neither is cause for either celebration or alarm.
Eniac:
“The default expectation would be that planets are equally common in places where we can’t see them as they are in places where we can.”
That’s only reasonable if we exclude our own system, because it’s not “randomly chosen” – if Earth wasn’t habitable, we wouldn’t be here to discuss other planets. By default we don’t know anything about the commonness of other Earths…
“Statements like “[frequency of] close Earth analogues at only ~1%” are not very meaningful, because the number depends more on what you mean by “close analogue” than on anything real.”
I did define what I meant by it in my previous post, didn’t want to repeat myself: I was talking about “planets with 0.8-1.2 Earth radii around G-stars in the more conservative habitable zone (orbital period 300-600 days)”.
(BTW, you didn’t define what you meant by systems “different from ours” for your stated frequencies (95%, 99,9%) either.)
To have cause for “celebration or alarm” you need to calculate the “density” of planets AND choose reasonable intervals for size and orbit range of potentially habitable planets; that’s what I was trying to do.
Good news:
Kepler seems to have found a small habitable-zone planet candidate after all! KOI-571-05 is a 1.1 R_E planet candidate around a late K-dwarf (the outermost of 5 super-Earth candidates) with orbital period 130 days; its equilibrium temperature is similar to Mars’. (See https://sites.google.com/a/upr.edu/planetary-habitability-laboratory-upra/projects/habitable-exoplanets-catalog/data .)
I wonder why it’s not yet confirmed after ~11 transits (and 6 transits already 2 years ago).
Since closer-in planets should be easier to detect, the lack of “warmer” candidates probably means that there are no close Earth analogues in the “warmer half” of the HZ around the latest K-dwarfs and M dwarfs in Kepler’s FOV.
Holger -but how do we know double transits are rare? That is the key question. The signal from an Earth in the HZ will be a small amplitude long-lasting but infrequent event. It is superimposed on much stronger signals from the close-orbiting super Earths.
Those stronger signals do however have their own uncertainties and add to the noise of weaker signals. The detection limit of the weaker signals is increased by the presence of the stronger signals.
But then you have to consider the co-planarity of the planets. If the planets were in random planes, then the statistics would not be biased. But we have good evidence, a priori, that orbital planes are correlated, hence there should be a bias.
As to how large that bias is, I don’t know I must admit. Maybe it is negligible. But I still wonder if ideas are being biased by the fact that the majority of systems have close-in planets, which then raises the detection limit for planets further out.
“Holger -but how do we know double transits are rare?”
By dividing the transit time by the orbital period, I’d say: typical hot super-Earths only transit ~1-10% of the total time, so the majority of HZ Earths’ transits will happen without added statistical noise due to double transits, even when the system is perfectly coplanar (unless you have many hot super-Earths in one system, which is rather rare).
You may need a little (~10%) more observing time so you can get an extra single transit to “replace” the occasional double transit’s bad data, but it shouldn’t be a show-stopper for discovering Earth analogs IMO.
Holger, OK thanks. Another thing is probably that being coplanar enough to have double transits could be unlikely. I must admit I don’t know what the tolerance is. I still find the empty bottom right corner of the diagram strange and await further data.
Finally I read the full paper last night (highly recommended strategy before commenting :). It actually says that only the highest signal to noise planet in a system was counted in their assessment.
Another point is the numbers are focussed on the 1-2 R-Earth range of size. What about the sub-Earths that are large enough to be habitable? The number of those does not seem to have been estimated.
The next point is subjective and unscientific. Nevertheless, stare at Fig. 1. Notice how the planet number density increases as the mass decreases across the plot. It increases down the plot until you get to the shaded area, where it starts to drop again. Inferences about the density in that area are purely that: inferences and extrapolations. Every single planet in our own system is deep in the low-detection shaded area, with virtually no probability of detection.
I think there is still a lot of work that can be done with this data. That’s not meant as a criticism because this paper is a good start, but that’s what it is, just a start.
kzb:
“Finally I read the full paper last night (highly recommended strategy before commenting :). It actually says that only the highest signal to noise planet in a system was counted in their assessment.”
I agree :-) I read most of the paper before commenting but missed this detail. So the paper (unlike Kepler itself) ignores any HZ Earths when there are hot super-Earths, which might a priori have drastically underestimated HZ frequencies…
I suppose the paper didn’t estimate sub-Earths because this would have required even more extrapolation. According to Kepler (http://kepler.nasa.gov/Mission/faq/#a13), planets need to be >0.8 earth radii to have a “habitable” atmosphere. Even planets only slightly smaller than Earth may not be habitable due to lack of plate tectonics (see Venus at 0.95 Earth radii), so habitable sub-Earths might be negligibly rare.
If you follow the paper’s “result/assumption” that planet densities are (logarithmically) the same for all (sub-Neptune) size ranges, the number of planets with 0.8-1 Earth radii should be about the same as the number of planets with 1-1.25 Earth radii, for every orbital range. (That’s essentially what I used to give my above estimate for the frequency of 0.8-1.2 Earth radii planets.)
I don’t see anything “unscientific” with your “next point”; the authors themselves admit the extrapolation due to lack of data for this region. Maybe that’s why they extended the “Earth-sized” size range to 2 R_E: so they had actual data points at least for part of the range.
I agree that the paper contains lots of interesting information. IMO, Figure 2 of the paper is much more informative than the publicized “22%” headline.
Exoplanet Habitable Zone Around Sunlike Stars Bigger Than Thought
SPACE.com By Miriam Kramer, Staff Writer
Earth’s place in the solar system is just right. It’s not too hot, like Venus, and it’s not too cold, like Mars, and this “Goldilocks zone” of habitability around other stars like the sun just might be bigger than thought, scientists say.
A new study, unveiled today (Dec. 11), expands the habitable zone — the sweet spot in a solar system where liquid water and therefore life could potentially exist — surrounding stars like the sun.
Previous studies on the habitability zone around sunlike stars have placed the innermost edge of so-called Goldilocks zoneat about 0.99 AU (1 AU, or astronomical unit, is the average distance from Earth to the sun, about 93 million miles, or 150 million kilometers). But a new computer model study pushes that border closer to its parent star, to a distance of about 0.95 AU (about 88 million miles, or 142 million kilometers).
The study in the journal Nature, led by Jeremy Leconte, now a postdoc at the Canadian Institute for Theoretical Astrophysics of the University of Toronto, used 3D computer modeling to find that the runaway greenhouse effect isn’t an issue unless the planet is less than 0.95 AU from its star.
The new inner boundary for habitable zones might not make a big difference for scientists trying to determine if an alien planet is habitable, but it does make a big difference for future life on Earth, Leconte said.
Full article here:
http://news.yahoo.com/exoplanet-habitable-zone-around-sunlike-stars-bigger-thought-190606278.html
ljk:
“Exoplanet Habitable Zone Around Sunlike Stars Bigger Than Thought”
It’s ironic that you post this as a comment here – Petigura et al. “thought” that the habitable zone extends to 0.5 AU, not just to 0.95 AU.
(But Leconte’s minor extension seems much more reasonable to me.)