750 Planetary Candidates from Kepler

by Paul Gilster on June 15, 2010

The release of the first 43 days of Kepler data has demonstrated just how powerful a planet-hunting technology we’ve put into space. Listen to principal investigator William Borucki (NASA Ames) in a video released yesterday by NASA television:

“We’re releasing data on 156,000 stars that we’ve been monitoring with the Kepler mission for 43 days, the first dataset. In these data are some 750 planetary candidates. Some of those are actual planets, some are false positives. Our science team is looking at 400 of those candidates with ground-based telescopes, to figure out which are planets, which aren’t.”

Borucki assumes about fifty percent of the candidates will be false positives, eclipsing binary stars, starspots, or other misleading signals. Now it’s in the hands of ground-based telescopes in the Canary Islands, Texas, Arizona and Hawaii to comb through these findings to make the call. The team is also releasing the data for the remaining 350 candidates to the world community of astronomers to assist with the analysis.

Addendum: The video mentioned above has suddenly gone private. I don’t know why, but will hope to post it again when it reverts to public access.

Ponder this: The number of planet candidates here is actually greater than the number of all the planets that have been discovered in the last 15 years. We could conceivably double the list of known exoplanets with these 43 days of data alone. And, obviously, Kepler is still on the job.

Dennis Overbye discusses Kepler’s data release policy, a subject we’ve examined here before, in a new article for the New York Times, from which this overview of Kepler:

…the treasure hunt for the end to cosmic loneliness continues.“The public wants to know whether there is life on other planets,” Mr. Borucki said, noting that it could take decades. The effort to get an answer, he said, reminds him of the building of the great cathedrals in Europe, in which each generation of workers had to tell themselves that “someday it will be built.”

“In a sense, we, too, are doing these things,” he said.

Well spoken. The Kepler work segues nicely into a new piece by astrobiologist and author Caleb Scharf (Columbia University), who discusses the search for habitable worlds in a guest post for Scientific American. Scharf makes the point that most stars can be described with only a few parameters. Analyze mass, age, and abundance of elements and it’s fairly easy to create the stellar taxonomy that categorizes them. Planets, however, are a different matter. With them, we consider orbit, type of primary, atmospheric composition, axial tilt, gravitational tides…

Well, you get the idea. The complete list takes in magnetic field, geology, chemistry and more. The problem this presents is that while we want to choose the right candidate planets to study for extraterrestrial life, we’re dealing with space-based observatories with finite resources. We’d like to come up with nearby stars to examine that are not so different from our Sun, but 75 percent of all stars are less than half as massive as Sol, and most nearby stars are red dwarfs like these, where planetary and stellar conditions may make life problematic.

What to do? Scharf suggests looking at the big picture by putting statistics to work:

Suppose there are biospheres scattered across many systems, perhaps driven by the same kind of microbial machinery that dominates our own. Conditions can vary tremendously among these worlds, but biospheres still persist. Environments on such planets may be held in subtly different equilibria than their sterile equivalents – seen through atmospheric composition, reflectivity, and temperature. No single world may actually present enough of a smoking gun to let us say, “there be life,” but put the data from all these planets together, and that signature might be detectable.

Could a statistical approach to astrobiology succeed? Scharf continues:

Statistics are wonderful, if finicky, things. They let us cut through the haze and see things we would otherwise never find. Suppose we accumulate crude, rudimentary data on not just a few planets, but on hundreds or thousands. No single observation of a planet may tell us if it is teeming with life, but the cumulative weight of different parts of the planetary zoo might at least tell us if there is life on some of them. By letting go of our desire to locate a single instance of life, we’d stand to answer the global question.

Interesting to keep this in mind as we learn more about Kepler’s findings. For Kepler is itself statistically based, looking at a huge number of stars with the expectation, now being borne out, that a number of these will have systems tilted just the right way so that we can observe planetary transits from our perspective here on Earth. Scharf explores the kind of signal we might dig out of the statistical noise in his Life, Unbounded blog, and it’s heartening to think that in an era when budget realities keep terrestrial planet finder spacecraft indefinitely shelved, we might still locate habitable worlds by virtue of big datasets and brute force computing.

tzf_img_post

{ 36 comments }

Reyn June 15, 2010 at 10:43

Correct me if I’m wrong, but these hundreds of possible planet candidates have almost no chance of being earth like, right? Most likely hot jupiters? Wouldn’t the observations in a few years be more likely to be “earths” in habitable zones because their orbits are farther out?

Mithridates June 15, 2010 at 11:00

Yeeeeeeaaaaaaaaaaaaahhhh!!! Looking at that video reminds me of just how tense I was the day Kepler launched because of just how impressive this telescope is. I’m firmly convinced that the discovery of one or more extrasolar Earths is going to be a game-changer that defines the 2010s just as the internet did the 1990s. Just as it’s hard to imagine waiting days and days for a letter to arrive, so too will it soon be difficult to imagine what it was like when Earth was the only planet of our type we knew.

Adding to this the Falcon 9 launch and this has been a pretty good two weeks. Rosetta’s flyby is coming up soon too.

Elon June 15, 2010 at 11:02

So exciting, so…uplifting! Cathedral indeed – my feelings at having lived to see even this day give me an inkling of how emotional my reaction is going to be if and when we get a clear indication of having found another habitable world: like attending the first mass in a freshly roofed cathedral must have felt like to my ancestors. We’re not there yet but some things sure are ‘looking up’.

Some impatient questions: Do we know if the Kepler team held back their favorites, or was it random, or what? Is there a read on how this result so far compares with expectations? Is there any other broad characterization of this preliminary data statistically, for instance in terms of the distribution of the length of time of occlusion?

Daniel Fischer June 15, 2010 at 11:17

You shouldn’t compare planet candidates (detected only via transits so far) with the RV-confirmed planets in the catalog – and remember what you yourself reported in March about over 100 planet candidates found by the vastly smaller and cheaper CoRoT satellite. The jury is still out on the planet hunting efficiency per invested dollar … The main difference so far is that the multi-national European project only releases the coordinates of confirmed planets (15 so far) and strictly withholds information on the candidates’ whereabouts while NASA is running a mixed – but at least as controversial – strategy in preliminary data releases.

Erik Anderson June 15, 2010 at 12:40

Where is the data posted?

Administrator June 15, 2010 at 12:54

Daniel Fischer writes:

You shouldn’t compare planet candidates (detected only via transits so far) with the RV-confirmed planets in the catalog – and remember what you yourself reported in March about over 100 planet candidates found by the vastly smaller and cheaper CoRoT satellite. The jury is still out on the planet hunting efficiency per invested dollar … The main difference so far is that the multi-national European project only releases the coordinates of confirmed planets (15 so far) and strictly withholds information on the candidates’ whereabouts while NASA is running a mixed – but at least as controversial – strategy in preliminary data releases.

No comparison of planet candidates as opposed to RV-confirmed planets made in the post. The difference seems clearly stated. I’ve made no comment on planet-hunting efficiency because the two teams have different policies re information release. The article makes no comparisons with CoRoT.

I seem to have blundered into a CoRoT vs. Kepler argument, and it’s not one that I want to pursue. Both missions have done superb work thus far, and I celebrate the successes of both teams.

Administrator June 15, 2010 at 12:58

Elon writes:

Do we know if the Kepler team held back their favorites, or was it random, or what? Is there a read on how this result so far compares with expectations? Is there any other broad characterization of this preliminary data statistically, for instance in terms of the distribution of the length of time of occlusion?

Yes, Kepler held back the top 400. See the Overbye article I refer to in the post for more on this policy, which we’ve also discussed in these pages. As to how this compares with expectations, I’m afraid I don’t have a good read on that, but maybe others will. Ditto with your second question.

Administrator June 15, 2010 at 13:00

Erik Anderson writes:

Where is the data posted?

Don’t know. Right now it’s hard to even find the announcement video, which has been declared ‘private’ and no longer accessible. Decidedly odd.

Administrator June 15, 2010 at 13:04

Reyn writes:

Correct me if I’m wrong, but these hundreds of possible planet candidates have almost no chance of being earth like, right? Most likely hot jupiters? Wouldn’t the observations in a few years be more likely to be “earths” in habitable zones because their orbits are farther out?

I think it depends on what we mean by ‘Earthlike.’ I suppose we may have some interesting worlds in this mix around M-dwarfs, but we don’t know whether a planet in the habitable zone of such a star could sustain life, and it wouldn’t be truly Earthlike in most respects even if it could. So yes, I agree with you, Earthlike planets in habitable zones around Sun-like stars will take longer to confirm, and that could be a matter of a couple of years.

andy June 15, 2010 at 13:44

Correct me if I’m wrong here, but RV confirmation of Kepler terrestrial planet candidates is expected to be essentially impossible due to the faintness of the host stars. I doubt we’ll get many confirmed terrestrial exoplanets (if any) from this mission, but at least some idea of the statistics of the planet distribution.

Of those 750 candidates, I wonder how many are in the open clusters in the Kepler field: we only know of a handful of planets in star clusters, I’m hoping Kepler will be able to shed some light on how the population of cluster planets differs from the population around field stars.

Administrator June 15, 2010 at 14:21

andy, this is interesting:

Five Kepler target stars that show multiple transiting exoplanet candidates

http://arxiv.org/abs/1006.2763

Abstract: We present and discuss five candidate exoplanetary systems identified with the Kepler spacecraft. These five systems show transits from multiple exoplanet candidates. Should these objects prove to be planetary in nature, then these five systems open new opportunities for the field of exoplanets and provide new insights into the formation and dynamical evolution of planetary systems. We discuss the methods used to identify multiple transiting objects from the Kepler photometry as well as the false-positive rejection methods that have been applied to these data. One system shows transits from three distinct objects while the remaining four systems show transits from two objects. Three systems have planet candidates that are near mean motion commensurabilities – two near 2:1 and one just outside 5:2. We discuss the implications that multitransiting systems have on the distribution of orbital inclinations in planetary systems, and hence their dynamical histories; as well as their likely masses and chemical compositions. A Monte Carlo study indicates that, with additional data, most of these systems should exhibit detectable transit timing variations (TTV) due to gravitational interactions – though none are apparent in these data. We also discuss new challenges that arise in TTV analyses due to the presence of more than two planets in a system.

Mike June 15, 2010 at 16:08

Wow!,if this number of candidates is detected in the first 43 days of observing one wonders how many will be spotted in 3 years of observing with ever improving data processing algorithms.The RV teams will be busy for many years on the follow-up observations.
Also considering the advantages of having a large data base to work with and the extremely detailed and precise light curves that Kepler has shown it can produce,I think this will help in verifying planet candidates as experience is gained in matching transiting light curves to RV confirmed planets.This increasing experience will give astronomers more confidence in interpreting Kepler’s light curves alone with out the RV follow-up as the follow-up will take many years and we still lack the capablity to identify Earth mass planets through the stellar radial velocity method in a timely way.
That being said I still believe RV follow-up still must be attempted,especially on the more Earth mass and orbit like detections. Better instrumentation is on the way.

Anders Feder June 15, 2010 at 18:37

Actually, Mike, there must be some form of diminishing returns involved here. Kepler has now spotted the most easily detected exoplanet candidates. Since we obviously won’t be counting the easy ones twice, the second batch of data may not yield nearly as many candidates.

Dave June 15, 2010 at 18:58

Am I correct in assuming the 750 planetary candidates do not include any transits that have only occurred once? Obviously any planet in the habitable zone of a Sun-like star will take around a year to transit, so we would not have seen two transits from those planets yet, if any.

NS June 15, 2010 at 19:23

A small number of Kepler datasets are available here, with links to ways of getting more.

http://archive.stsci.edu/prepds/kepler_hlsp/

I downloaded one and am in the middle of fiddling around with a program that graphs the light curve. Even with my very simple approach the periodic blips in the data are obvious.

Querying the actual Kepler data archive looks a bit intimidating. Hopefully more datasets like those at the link above will be created by people who know what they’re doing!

Mike June 15, 2010 at 20:48

Anders,what Kepler has seen are the candidates with the shortest periods.
These would likely be hot gas giants,hot rocky worlds and maybe an Earth mass planet or two orbiting a low mass M-dwarf. All the short-period transiting planets that have produced 3 detectable transits in 43 days or less.
The next block of Kepler observations has run from June 2009 t0 June 2010. Now I wonder how many candidates will have produced 3 transits within that year.That would be periods of 4 months or less.Then we are getting into the habitable zone for K-dwarfs and getting into really unknown territory for exoplanetary populations. This is what Kepler was built for.
An exoplanet census taker. What I mean to say is Kepler has detected the easier planets yes,but probably not the most numerous.That will take more time. I think the next list of candidates will number into the thousands.

Phil June 15, 2010 at 21:18

Most of the easy exoplant candidates may have been found, but perhaps many of the remainder will be more interesting – at least from the terrestrial/astrobiological POV – because presumably they will have longer periods?

Extrapolating these ‘porthole in space’ discoveries onto the whole area of the sky is truly mind boggling.

p

Thomas June 16, 2010 at 2:13

@andy

RV Confirmation of Kepler candidates benifits from knowing the there is something there, and has a well-defined period. Thus, finding an RV fit is much more trivial than it would be if it were a blind search.

Without transits, fitting a faint RV signal of a planet would suffer from significance tests. The signal is so weak, how can we be sure it’s real? Detection of transits argues strongly in favour of a planetary interpretation and therefore allows much lower mass-planets to be confidently confirmed.

@ Dave
Correct. All of these candidates have transited at least three times.

tesh June 16, 2010 at 7:03

Stunning!

Cannot wait for their 300 odd “held back” planets!

kzb June 16, 2010 at 7:49

750 (divided by 2) out of 156,000 is not actually a very big proportion (it’s 0.2%).

I’m not sure if this data is telling us anything much in terms of frequency of planets and what type of planets can be predicted from a given stellar type.

I believe that current thinking remains that planetary systems should be more or less universal. However if it turned out the frequency is only 0.2%, we would need a big re-think. If they are universal it means that at least 99.8% (and probably more because more than one planet per star is expected) of the planets around these stars have NOT been detected yet. What does that mean for observational bias?

tesh June 16, 2010 at 9:00

However if it turned out the frequency is only 0.2%, we would need a big re-think. If they are universal it means that at least 99.8% (and probably more because more than one planet per star is expected) of the planets around these stars have NOT been detected yet. What does that mean for observational bias?

these are only transiting planets, i.e. planets that pass across their stars within our line of site. Kepler will only get those planets.

Mike June 16, 2010 at 10:28

To kzb,
Finding exoplanets by detecting their transits is complicated by the fact that the orbital plane of planets around their stars appears randomly orientated.
To observe a transist the planet must cross the face of it’s star as seen from our direction. This is more likely to occur with large close-in planets then with smaller worlds orbiting farther out. Either way the chances of viewing the transists of any paticular exo-system are in the very low percentage,say .5 to 2% depending on the orbits and diameters of the exoplanets.
What that means is we would never get 50 to 100 % detections with Kepler.
The .2% detection from the first 43 days is pretty healthy and a damned good start.
About 1 percent is the expectation for transit detections in Kepler’s target stars. The idea being when the survey is completed and the data analysed and the follow-up finished up you total up the number of exoplanets detected and then multiply by 100 and that extrapolation will give you a rough census of the number and type of planets orbiting main sequence dwarf stars in a typical patch of spiral arm in our galaxy. Ofcourse there is always the chance of some surprises as well,this is a very unknown area once we start looking at the lower mass planets.
The observational bias you mention is very pronounced with the radial velocity method of detecting exoplanets because the more massive and close-in worlds are much easier to detect then smaller worlds.It is not possible at this time to detect the smallest mass planets using the RV method except maybe by the Debra Fischer technique she and others are attempting on Alpha Centauri.That is a special case.
Kepler was designed to have the ability to detect Earth mass planets orbiting Sun like stars using the transist method.If they are there Kepler will spot them.

kzb June 16, 2010 at 13:32

Mike and Tesh, OK I suppose I am a bit happier about it now.

If we know what the statistical probability of detection of each planet type within orbit bins is, we can multiply up by the inverse of that factor to obtain a census estimate.

What I was getting at was, the “Oh wow we’ve discovered another hot Jupiter” angle does not get us anywhere much on this forum.

andy June 16, 2010 at 13:50

@Thomas: actually a lot of the objects in this initial sample are based on single transits. In the case of a single transit, orbital period is estimated based on assumption of zero eccentricity, etc. Note the longest period object has an estimated period of about 28 years (!), no way have we got three transits of this object yet…

Bill Parkyn June 16, 2010 at 15:15

1. The transit duration is a key parameter distinguishing slow-moving terrestrial planets from fast-moving close planets, and from other false alarms. Even for the minority of terrestrial transits on short chords, the slowness of the transition will still be diagnostic, in spite of the smaller planet size. Limb darkening changes will also be slower during a short-chord transit.

2. The 156,000 target stars are all solar-type, no M-dwarfs (just F5-K5, I believe).
At terrestrial distances such stars subtend ~0.01 radians, the source of the 1% number. This would mean 1560 transits per year in the 12-hr class, or 183 in 43 days. BUT recent discoveries have shown that only 10% of solar stars have spectra with the characteristic iron deficit caused by terrestrial-planet formation. That means only 18 of the transits in the 43 day period would be in the 12-hr range. This could be as high as fifty, given that the sun has 3 Kepler-detectable terrestrial planets, although those are not coplanar enough to show up together from most directions that can see any one. So maybe 30-40 non-repeating long transits.

3. I hope somebody digs out such slow transits and makes a table of them.

Mike June 16, 2010 at 16:37

About 2 percent of Kepler’s target stars are M-Dwarfs.About 60 percent are G-Dwarfs,25 percent K-Dwarfs and the rest are F-Dwarfs. Despite the greater number of M-Dwarf stars they are too faint to provide a large sampling for Kepler’s survey.But some are bright enough to be included in the target list.
Could Andy provide more information about the single transit candidates?

philw1776 June 16, 2010 at 17:13

“The 156,000 target stars are all solar-type, no M-dwarfs (just F5-K5, I believe).”

Not really. Kepler is looking at stars from late Fs to ~M5 with more astrophysical restrictions on Ms (lots of younger Ms are variable flare stars and yuh the M5s are geting too faint).

And yes somewhere from 2% (stars with close in planets) to 0.5%
(planets an AU or more out) will fortuitously have their orbital planes aligned so Kepler can detect occulting planets. Assuming an avg of 1% aligned ‘right’ out of each 100,000 stars examined, then ~1,000 of these stars should have detectable planets, assuming they have planets.

andy June 16, 2010 at 17:50

I wonder if the disappearance of the video is anything to do with the frankly bizarre situation where NASA HQ is apparently ordering no discussion about the exoplanet candidates until the papers have been peer reviewed, despite the fact that the papers are now publically accessible.

Mike June 16, 2010 at 23:30

Some of those single transit candidates in Kepler’s public data release look quite interesting. But these are not the best verified candidates and their transits should be re-observed with time. One may have a long period that a second transit will never be seen by Kepler. That is #771.01 in the list of first public release candidate exoplanets.
Lots of new data for astronomers to work with.
We live in great times people! With tools that astronomers from the recent to distant past could only have dreamed about.I feel so fortunate to have lived to see answers being found to some of the really ancient and profound questions that any thinking person has pondered and speculated about.
And I wonder if we will be lucky enough to witness bio-signatures being detected by the James Webb Space Telescope a few years down the road.

kzb June 17, 2010 at 7:38

Sounds like a data handling nightmare -glad I don’t have to do it !

Seriously, do we know for a fact that the data are going to be treated in such a way that realistic estimates can be made about the frequency of terrestrial planets in habitable zones?

Mike June 17, 2010 at 11:20

To kzb.
That is the main reason for the Kepler mission,to attempt a census of terrestial and other planets in the habitable zones of sun like stars.
There are other astronomical information being gathered as well but that main reason is what the Kepler spacecraft and ground teams were designed to do.
I would recommend the Kepler web site as a rich source of information for any questions you may have.

spaceman June 17, 2010 at 15:50

Is there any indication as to what is the rough numbers of planets of different radii in this initial sample?

Ronald June 18, 2010 at 7:57

Ballpark of course, but the mentioned number of candidates out of a sample of 156,000 stars, in only 43 days of observation time, seems like a nice and promising result.

I mean, assuming an average detection (transit) chance of 1% implies that, even if all those stars had planets, the total end result would still only be some 1560 (stars, not planets, since one star can show multiple transits).

Hence, 350 – 400 serious candidates (assuming half will be false positives) in such a short time actually seems quite good.
On the downside, it can also mean that hot Jupiters and Neptunes (and inward migration!) are the common rule.

Greg Laughlin’s systemic site has some very interesting additional information on this: http://oklo.org/2010/06/16/the-312-candidates/

His own Monte-Carlo estimate (last part of his article) shows that, taking the 312 released candidates out of a sample of 88,196 target stars: if half of the stars have a planet from earth mass up to Neptune mass (17 Me) and with orbits of less than 50 days, *1100* planet candidates should have been present in this sample. The present harvest is almost 30% of that. Suggesting that about 15% of those stars have a close-in (< 50 d orbit) earthsize to Neptune size planet (?).

spaceman June 20, 2010 at 1:02

So many candidates in such a short period (no pun intended) is indeed a promising sign. If the notion was not dead already (as it should be based on the exoplanet findings thus far), then the idea that planets are rare will certainly be dead as a door nail with the advent of the Kepler mission.

Interestingly, Greg Laughlin’s preliminary calculations seem somewhat at odds with the Swiss team’s radial velocity results. The Swiss team, whose data still has yet to be published for whatever reason, indicate that ~30% of FGK stars have super-Earths in 1 to 50 day orbital periods. On the other hand, the American NASA-UC Eta Earth team’s results indicate that this number may be closer to the ~15% implied by Greg’s calculation. So, if a more refined analysis of the Kepler results AND the full NASA-UC Eta Earth team’s results agree with each other about close-in exoplanet frequencies and DISAGREE with the Swiss team’s exoplanet frequency data, then it will seem likely that the Swiss team is not correct. To be perfectly clear: the Swiss team’s results which, though unpublished, imply that ~30% of FGK stars have short period super-Earths may NOT be correct. So, speaking of likelihoods, a looming exoplanet frequency controversy seems very likely indeed at this point!

spaceman June 20, 2010 at 7:56

Swiss radial velocity derived frequency of exoplanets amiss in light of NASA-Eta Earth survey results and preliminary Kepler data?

spaceman June 23, 2010 at 0:27

This many candidates in 43 days. It will be interesting to see if a similar number of planet candidates emerges from an additional 40-50 days of data.

Comments on this entry are closed.

{ 3 trackbacks }