Exploring Kepler's library

Last week's data release from Kepler appears to have temporarily overwhelmed both professional and amateur exoplanet enthusiasts. After the initial flurry of basic overview posts on the Kepler data, I noticed a conspicuous hush fall over many of my favorite astronomy blogs, presumably caused by their authors turning en masse to parse the new treasure trove.

The list of more than 1200 candidate planets will likely yield more than a thousand actual confirmed worlds once culled of false positives. One system, Kepler-11, contains six confirmed transiting planets, and another system, KOI 157, has five candidates. Eight systems have been found with four candidate planets, along with 45 triple-planet and 115 double-planet systems. It's a lot to digest, and only represents the first four months of data from a 3.5-year mission.

Fortunately, the incapacitating shock of the breadth and depth of the new dataset seems to be wearing off, and researchers are beginning to reveal some of their initial explorations. In particular, Daniel Fabrycky, a University of California-Santa Cruz astronomy post-doc and member of the Kepler team, has created an impressive visualization of projected orbital motions for all the multi-planet systems Kepler has discovered to date. Within each system, planets are color-coded according to size, with the redder planets being larger and bluer planets being smaller.

It's hard to overstate the magnitude of the insights that can potentially be extracted from novel presentations of Kepler's raw data and its present and future planetary ensemble. Though each planetary system constitutes an essentially static snapshot of only one outcome from eons-long stochastic processes, lurking in the aggregate are lessons about how exactly planets form, how orbital configurations change over time, the relative distributions of planetary size, and frequency and how a star's age, size, and mass determine the sorts of planets it produces.

Sometimes these rules and the relationships between them may clearly manifest through a simple chart or graph of two key variables, but in other cases they may only reveal themselves through more dynamic presentations and multivariate analyses that better leverage the pattern-recognition capabilities of people. In this way, Kepler's large, diverse data sets may stimulate not only a more robust understanding of stellar and planetary science, but also significant progress in the effective design and usage of scientific data visualization.

For example, novel visualizations of stellar light curves from Kepler's first batch of data, released in June 2010, allowed members of the citizen-science project Planethunters.org to preliminarily identify 83 candidate planets that were only confirmed in last week's data release.

The same visualizations, which plot dips in the brightness of the more than 150,000 Kepler monitors, also yielded what may prove to be 47 additional candidate planets that slipped through the Kepler team's automated pipeline. Many fainter, subtler signals of smaller planets in habitable orbits around larger stars are certainly present unrecognized in the most recent dataset—borderline events that won't trigger a flag in a software routine but will catch the human eye. More people should be looking—there is a not-insignificant chance that with a bit of luck and careful observation, you could discover a potentially Earth-like world in Kepler's data even before the mission's scientists do.


  1. The one thing that I would like to know about all of these worlds is the “reverse look.”

    That is, if we were on a planet in, say, the KOI 157 system, looking back at the sun with Kepler-style eyes, what would we see?

    Would we be able to detect Earth, Mars, Venus? Or would we be limited to the larger gas giants? Or are they even too far from the sun to cause a detectable wobble.

    If we were there, could we see us?

    1. Re: the reverse look from KOI 157… It depends chiefly on the mutual orientations of planetary orbits in each system. So, we can see the KOI 157 planets because from our perspective in the sky they are all more or less edge-on to their star. Someone on one of those worlds could see ours in the exact same way if some of the planets in our solar system transit in that particular line-of-sight.

      I’m not certain what the mutual orientations actually are in the case of KOI 157, but assuming they were both exactly edge-on, the observers on KOI 157 probably would still have trouble detecting all of the planets in our solar system using a twin of the Kepler mission. The chances of any planet transiting diminishes as its orbital separation increases and its angular size decreases…

      1. I would have thought that all of the new found systems would ALL be pretty much in the same line as ours…that is in line with the overall general rotation of the galaxy. Centrifugal effect and all that…..Is this not the case?

        1. Nope. Star formation is much more chaotic than the small contribution from galactic rotation, and systems form at nearly every angle.

        2. I bet an actual astronomer would have a better answer, but here’s mine: I don’t think so, no. The large-scale rotation of the galaxy is pretty disconnected from the rotation of individual stars. I am *guessing* that the orbital inclinations of exoplanetary systems would be pretty randomly distributed with regard to the plane of the galaxy. But it would be really interesting if there were a correlation like that…

    2. As Lee Billings suggested, observers of our solar system would only be able to detect its planets with a Kepler-like method if they were in line with the plane of our solar system. So, given that Kepler’s own field of view is far from the ecliptic and thus does not intersect with the ecliptic plane (the plane of Earth’s orbit) and that the other planets have orbital planes very similar to Earth’s, I would say that our solar system would be undetectable to astronomers in the KOI 157 system looking in the direction of our sun with a Kepler-like instrument.

      There are other techniques that they could use, but planetary transits would not work.

  2. “…and only represents the first four months of data from a 3.5-year mission.”

    How big is the chance that there’ll be more data releases of this size? I would imagine that the first planets discovered are also the easiest to discover, and that it will be increasingly difficult to detect more after that.

    1. The chance of more releases this size are fairly high, though you’re right that it’ll slow down exponentially (given a consistent press release cycle, not an assumption I’d necessarily make).

      Right now we’re seeing the huge, fast-moving planets. Their ‘years’ are measured in our days or months. Given the requirement to see at least two transits to be confident that there’s something there, it would take two years to monitor Earth’s own transit from a distant star. As the time it takes to measure the next set of orbits doubles, the frequency of discoveries will likely halve.

      Keep in mind, though: We’re able to make out the difference between day & night on some of these planets, and even measure their atmosphere! Kepler is very sensitive.

  3. This whole project is truly fascinating and awe-inspiring, but I’m forced to wonder how ‘old’ the data we’re looking at is. Correct me if I’m wrong, but aren’t most of these stars so vastly far away that what we’re seeing is data from centuries or millennia ago? Even if we could get live data from a suitable planet *right now* what are the odds that, even if there is or was human-like life there, that it’d already be long gone… Truly the mind boggles.

    1. Almost all the planets discovered are within 5000 years, and most are within 2500 light years. But a substantial number are much closer. 16 Cygni B for example is only about 75 light years away, and there are a substantial number of candidates within 200 light years. I think 16 Cygni B is the closest confirmed planet. I don’t know what the closest unconfirmed candidates are, they might be closer but I don’t think so. Also, very few of the close to Earth-like planets are very close. Kepler-10b which is pretty rocky is around 560 light years away (also it is way too close to be in the habitable zone). Right now though the data isn’t very well organized (or if it is, I don’t know where that is in an easily searchable form) , so I might be missing important closer stuff.

  4. I wonder what the Sol system would look like if Kepler was at Alpha Centauri, Barnard’s star, Tau Ceti, etc?

  5. Any idea what fraction of stars Kepler looked at have observable planets? Any idea how to extrapolate that to what fraction have planets we could observe if the star’s planets’ ecliptic plane were oriented properly?

    I would imagine we are only seeing planets around only a very small fraction of those stars that have planets, due to constraints of the geometry for detection.

    1. It’s true, Kepler is only skimming a vanishing fraction of the planets that exist around the stars in its field-of-view. I wouldn’t have been so certain when I began this guest-blog stint, but the latest data release makes me confident: Transits are rare, but planets are common as dirt.

  6. I love the discussion. I love the video. I love these posts.

    I do think that visualization would be seriously improved by including our solar system as a reference for size and speed.

Comments are closed.