Wednesday, February 26, 2014

Holy S***! NASA Announces 715 New Confirmed Exoplanets!

Yes, the title pretty much says it all. NASA scientists have confirmed the existence of 715 former planet candidates, all of which are in multi-planet systems. Sounds cool and all, but what does that actually mean?

The original goal of the Kepler mission was to continuously observe 156,000 stars and monitor how their brightnesses changed over time. One of the possible ways that a star's brightness can change is if a planet orbiting that star passes in front of the star, blocking some of the star's light from reaching us. This is what we call a transit. Observing transiting exoplanets was the primary mission of Kepler, and it has found nearly 3,500 candidates and confirmed planets to date. The pattern of light we would expect to see from a planet transiting its star is shown below for a single planet.
Image courtesy of

Why am I making the distinction between "planet candidates" and "confirmed planets"? It turns out that there are a number of astrophysical phenomena that can pretty easily mimic the signal we expect to see from a planet passing in front of its star. Among these are eclipsing binary stars, where two stars orbiting one another pass in front of each other regularly, which temporarily blocks the light from either star. In the right situations, this can look exactly like the above. Because Kepler collects so much data from so many stars, the data are analyzed by computer algorithms written to look for transit-like signals. As such, it will occasionally pick out something that looks right, but is actually the result of an eclipsing binary or perhaps a pattern of starspots moving across the star's surface. But because these signals are not the result of a planet, we group them together under a single term, and call such events "false positives".

There are some ways to exclude false positives from the detection algorithms if they're REALLY obviously not planets, but probably roughly 5 to 10 percent of the time (the algorithms have been improved over time), some non-planets sneak in to our list of planet candidates. We only call these algorithm-vetted planets "candidates" because they could still be particularly devious false positives, and there is really no way to tell actual planets apart from the really tricky false positives without further observations.

So how do we pick the real planets apart from the false positives? The traditional way to do this would be to observe the stars using the Doppler wobble method, which would allow us to measure the mass of the other objects in the system so we can tell the planets apart from the stars. Of course, this involves getting observation time on some of the biggest and best telescopes in the world (the twin Keck telescopes are usually the workhorses for these observations) for multiple nights to get sufficient data, so this gives us a bit of a bottleneck.

Another more recent way to confirm exoplanets that only works in multi-planet systems relies on the mutual gravitational forces the planets exert on one another. Traditionally, planetary orbits are very regular; if a planet's orbit is 10 days long, we should see a transit once every 10 days like clockwork. However, if there is another planet in the system (or possibly more!), the two planets can pull on one another as well, which can change the exact time at which a planet transits by a little bit. A fantastic animation of this phenomenon is shown below in a video created by NASA's Kepler team. These slight changes in the planet's orbital period are called "transit timing variations". On the down side, this method only works if the planets are large enough and/or close enough to one another that the changes are detectable to us.

In 2012, Jack Lissauer of NASA Ames Research Center published a paper in which he described a new statistical method that could allow for the confirmation of various planets in multi-planet systems. The basic idea behind this method is that, while false positives can fairly easily resemble the types of signals we expect from single transiting planets, the likelihood of a false positive in a multi-planet system decreases with every successive planet candidate in that system. In the best possible systems, the probability of a given planet being a false positive is low enough that the planets pass the probability threshold necessary for the Kepler team to consider an object "confirmed".

Performing this analysis on the existing Kepler multi-planet systems yielded a total of 305 planetary systems with a total of 715 planets among them, leading to the announcement made just a few hours ago today.

Even more interestingly, the newly confirmed exoplanets (roughly equal to the number of previously confirmed exoplanets) dramatically affect the demographics of confirmed exoplanets. The figure below, taken right from NASA's presentation materials (which can be found here), shows exactly how the numbers of different sized planets have increased as a result of the new findings.
While most of the planets confirmed were Neptune-sized planets, the part of this distribution to experience the largest increase was the super-Earths, of which over 300 are now confirmed to exist. Last, but definitely not least, the number of Earth-sized planets increased by an additional 400%, such that over 100 Earth-sized exoplanets are now confirmed. None of these numbers are particularly surprising when you consider what Kepler has told us about the sizes of exoplanets out there. The only reason Jupiter-size planets are in the majority is because they're the easiest to find. Relative to the other sizes, Kepler observed VERY few gas giants, which tells us that they're actually very uncommon compared to the smaller planets, as shown below.

The percentages shown on the above graph represent the percent increase in each category since the previous data release. In short, the longer we observe, the more small planets we see. So we have every reason in the universe to believe that planets like Earth are common.

Of course, a lot of news has talked about how Kepler is now out of order due to its reaction wheels failing. However, the Kepler team has come up with a very innovative way to keep Kepler going and gather more high-quality data over the next three years or so. But that's a post for another time (by which I mean, I was actually planning to write about that before this came out).

Also, I have to use this space to give a shout-out to my friend and colleague Kim Star for having her name on one of the papers released as a part of this announcement!


  1. This was a good read -- thanks for sharing. So does earth experience significant transit time variations? Or are we small enough / far enough from other planets (which are also small) that it's entirely negligible? I have never heard of any, but now I am curious.

    I wonder whether we would be able to observe Earth if we were elsewhere looking at Sol.

    1. First, I don't think we'd see TTVs for Earth. The planets nearest Earth are all pretty small, so the gravitational effect would likely be undetectable without ludicrously high precision observations. Interestingly enough, it *could* be possible to detect timing variations that occur because of the Moon's orbit, but it would, again, require obscene precision that we probably just don't have at the moment.

      We would probably be able to detect Earth transiting though, if we use a telescope like Kepler. Kepler was specifically designed to be able to detect Earth-like planets around Sun-like stars, so with the right alignment, it's totally observable.