A new view of the galaxy: Exclusive Kepler data visualization by Jer Thorp

Note from Lee: Video is best viewed in HD, full-screen mode.

Following up on yesterday's post about Dan Fabrycky's festive rendition of Kepler's candidate multi-planet systems, I'm proud to unveil an exclusive new visualization of Kepler's candidate planets, courtesy of the multitalented Jer Thorp, who has an excellent blog at blprnt.blg.

Jer is a native of Vancouver, Canada, but he currently makes his home in New York City, where he is the New York Times' Data Artist in Residence and a visiting professor at New York University. He's also a contributing editor at Wired UK. Shortly after the Kepler data release, Jer and I started talking about how he could dynamically visualize both the magnitude and the nuance of the discoveries. This is what we came up with.

What you're looking at is every Kepler candidate, arranged as if orbiting a single star. In the real world, this would result in a catastrophic gravitational destruction derby, but in our virtual world it simply normalizes the candidates and allows a proper sense of scale to be perceived. Each candidate's estimated size, orbital speed, and orbital separation is accurately depicted, and each world is color-coded according to its estimated effective temperature, with red being relatively hot and deep blue/violet being relatively cold. Mercury, Mars, Earth, and Jupiter are added for context; the high-value Kepler candidates KOI 326.01 and 314.02 are also highlighted.

The color scale is calibrated so that Earth is a pale blue dot. This is the color it would display across the gulfs of interstellar space, a hue that suggests blue oxygen-rich skies and deep liquid-water oceans. Two concentric rings plot distances of 0.5 and 1 astronomical units from the central star, and a pale blue line delineates Earth's location on two self-organizing charts. In the video posted above, the first chart in the sequence plots semi-major axis (i.e., average orbital separation) versus effective temperature, while the second plots semi-major axis versus planetary size.


Important data trends prominently emerge from this visualization. The abundance of smaller candidates and relative sparsity of larger ones clearly indicates that there are many tiny, meek worlds for every giant planet. But curiously there is a relatively stark drop-off in the frequency of Kepler candidates at or below approximately Earth-size. My gut says this is probably an observational bias that later data releases will smooth out, but perhaps it's not, and Mother Nature actually prefers slightly chubbier planetary progeny, with Earth being a slender outlier.


Similarly, Kepler's nascent sensitivity to smaller planets with larger orbital separations is also suggested from the pile up of small, hot, fast-moving planets in very short-period orbits. The tight clustering of worlds large and small, cold and hot, all well within the orbit of Mercury testifies to the fact that most of Kepler's haul thus far comes from small, cool stars. Only later data releases will probe the habitable zones of larger, hotter stars like our Sun. Hopefully, the best is yet to come.

The visualization also highlights some things that are just plain weird. For instance, several other apparently promising candidates accompany both KOI 326.01 and 314.02 in terms of estimated habitability. Why weren't they assigned higher values based on Greg Laughlin's equation? I'm guessing part of the answer is that they orbit very distant and dim stars that offer extremely marginal chances of follow-up observations, something that Laughlin's formula substantially penalizes. The rest can be explained by the fact that any candidate large enough to be easily noticeable in this visualization is several times the size of the Earth, and hence of dubious habitability. The prevalence of apparently promising pale blue dots is an artifact of just how great the disparity is between the largest and smallest candidates.

Speaking of which, the largest candidate, an enormous pale blue orb, strikes me as extremely strange. It's several times larger than our solar system's heavyweight champion, Jupiter, yet it's also relatively distant from its star, and estimated to be rather cool. If I recall correctly, planets puffed-up to such large sizes get that way through being extremely hot and close to their stars. Perhaps this candidate will turn out to be something distinctly unplanetary, like a brown dwarf.

What do you see in this visualization? Do any clarifying insights or unsolved puzzles leap out at you?

How do you think it could be improved?

Let me know in the comments, and stay tuned—there may be an updated, interactive version of this coming soon.


  1. Is there a minimum distance in orbital separation from the star, or a sharp fall-off in frequency at some distance? I would think that the current Kepler data would be most sensitive to these nearest orbits, for any planet size. If many of the near-star planets migrated there, why don’t they migrate even closer?

    1. I remember reading in one of the discovery papers that there does seem to be a substantial fall-off in planetary candidates with less than 3-day orbital periods. It doesn’t appear to be purely an observational bias, either, because the fall-off also appears in other non-transit surveys. So it may represent an inner boundary of sorts past which planets tend to fall into their star or perhaps get ejected from the system.

  2. at another site i saw a map of “space” and what Kepler was looking into,it’s nothing not even a drop in the bucket.if that much can be seen by looking at a pin head worth of “space” think about what the rest could hold.

    1. I actually think it’s rather premature to give this object its own Wikipedia page. There’s still a lot of uncertainty associated with this object and others like it, objects that may be terrestrial planets orbiting in the habitable zones of their stars. It’s awesome, sure, but it’s entirely reasonable that no one is shouting from the rooftops yet. I’m actually a bit confused why KOI 701.03 has been singled out over all the other potentially terrestrial candidates in the habitable zone.

    2. In other words, yes this is totally amazing! But it will be a while yet before KOI 701.03 and other similar objects will be on sufficiently firm footing to really raise a ruckus. Looking forward to seeing how things unfold when that happens. :)

  3. Maybe the data of all other discovered exoplanets could be incorporated into these charts?

    A chart showing the spectral class of a star against their exoplanets could be interesting, too.

  4. Nifty visualization! A couple of comments here–

    “But curiously there is a relatively stark drop-off in the frequency of Kepler candidates at or below approximately Earth-size. My gut says this is probably an observational bias…”

    Your gut is correct :)
    One of the main takeaway points of the Kepler paper is that even though the survey is incomplete so far, and still contains plenty of observational biases, it so far looks like small planets are inherently *more* plentiful than large ones. That’s a big, big deal– something we really didn’t know before.

    As for KOI 701.03, it certainly is an interesting candidate– remember though that we are still trying to figure out whether super Earths can be habitable planets. Some of them may be more like little versions of Neptune (not such a clement place to live), but some of them may have big oceans, which might be great places to live. Usually when we find planets, we try to understand them in the context of planets we have in our own solar system– and understandable bias, since we live here. In the case of super Earths, though, we don’t really have an up-close-and-personal planet to compare them to. So while it’s really exciting to find these worlds– not just because they’re new and different but also because it means we’re honing in on finding even tinier planets– we still don’t know whether super Earths are habitable or not.

  5. this is lovely – but there’s a key to temporal normalization that’s missing (at least i couldn’t find it). you said
    “Each candidate’s estimated size, orbital speed, and orbital separation is accurately depicted, and each world is color-coded according to its estimated effective temperature, with red being relatively hot and deep blue/violet being relatively cold. ”

    but i’m watching many of these worlds whiz around their star in seconds. how many seconds of video time = 1 day of orbital time?

    1. Hi Jill,

      Good catch. I’m guessing the timescale of the video could be derived from tracking the orbital motion of a solar system planet like Mercury or Earth and comparing it to the actual known value, but yeah, that’s a bit complicated. Maybe Jer can clarify. I’ll ping him and we’ll see.

  6. My gut says this is probably an observational bias that later data releases will smooth out << planets closer to their star make the round trip more often, increasing the probability of us seeing it.

  7. Improvements
    Assuming each type of star has a habitable zone that can be estimated….

    1) Make the center star change type and show their respective habitable zone
    2) Whenever the center star changes type, highlight the orbiting planets that are actually orbiting this type of star.
    3) Whenever the center star changes type, highlight the planets that fall within that star’s habitable zone temperature, excluding those stars highlighted in 2
    4) Join the planets from 2&3 for some interesting results (if there’s even enough data)

  8. Hi Jill,

    The timescale is adjustable – I picked a value that seemed to work out well.

    Right now, 1 year = 50,000 frames. The animation runs at 60 frames per second, so a year is about 833 seconds.

    A day, then, is about 2.2 seconds.

    I will be releasing an interactive version of this, in which the timescale will be one of the adjustable parameters.

    Thanks for the question.


  9. Hi Jill,

    The timescale of the visualization is easily adjusted. In the interactive version, this will be a parameter than can be changed on the fly.

    In this video, 1 year == 50,000 frames, or about 833 seconds.

    So, one day is about 2.2 seconds.

    Hope that helps.


  10. Can someone clear something up for me? My assumption from reading a lot about exoplanets is that the large close orbit planets are wrong. Flat wrong.

    My theory is, what if the very large inside orbit is actually the star being pulled by multiple smaller planets that are at normal orbits? Instead of one ginormous planet orbiting closer than Mercury, what if there were, say 5 Mars-to-Jupiter sized planets at roughly normal distances?

    1. Hi Anon,

      You know, a lot of people were skeptical of the “hot Jupiters” after the first batch was discovered in the mid-1990s. They thought that something so strange couldn’t exist, that the signals were probably starspots or some other scarcely understood variety of stellar noise.

      But then astronomers found one that transited (http://en.wikipedia.org/wiki/HD_209458_b), and it all became clear. You can measure the radius of these things, and they’re huge. It’s a pretty direct confirmation that these sorts of bizarre planets are real.

      Kepler’s planetary candidates, remember, are almost exclusively transiting. So we’re measuring very directly their radii. There are of course still substantial uncertainties, but not as much as are associated with detections that occur only through radial-velocity surveys.

      For what it’s worth, the very effect you describe, where the RV signals of multiple planets blend to produce illusory planets, is something that planet-hunters are extremely careful about and sensitive to. A lot of the real innovation that occurs in RV searches these days involves designing software routines and data analysis pipelines that address exactly this issue.

      The biggest problem I see with your theory that some large, close-in planets are just the blended signals of smaller planets in more distant orbits is that this would suggest the periodicity of the smaller planets’ orbits would have to somehow be synchronized so that they all induced their maximum and minimum pulls at the same time. And I’m not sure that’s even physically possible. (anyone who actually knows better want to comment?)

  11. Just what the dominant cultural paradigm needs. a cosmologically-scaled statement in favor of size- the Universe appears to be a size queen? That’s just *Great*. Hey! Universe! It’s not the size, it’s what you DO with it! See: Earth (and thank you Very Much, Earth, for being both an excellent counterpoint and decent planet.)

  12. It would be even more cool to see a similar graphic of the Earth’s human tribes, especially with the challenge of screening out the 98% of human asteroid choss living in metropolitan ghetto blight. Just Earth’s few self-sufficient tribes, ex-insanofantasiaurbania, like Shishmaref, Alaska, for example, far out in the EMS Region 5A of the Norton Sound Region, in the Chukchi Galaxy N51.

  13. Great but this doesnt prove anything.There is life out there,the u.s. and other goverments have created a reality machine in which they decide what you beleive.Apparently the only habitable planets they will show probably will be uninhabited planets.The ones that are inhabited are kept secret with a sort of area 51 glimmer.It would be absurd to assume that w are alone for a variety of reasons,one being that we are not alone.Based upoun placement in a steller group.Read more my book if released in the future.

Comments are closed.