By Phil Plait
Astronomers working with data from the Kepler space telescope have announced the verification of a stunning 1,284 new exoplanets, worlds orbiting other stars. Even better, nine of them are at the right distance from their host stars (the “Goldilocks zone”) to have liquid water on their surface; that is, they’re potentially (so kinda maybe) habitable.
I want to be very careful here. These planets have been verified using a statistical analysis, and each has been found to have a greater than 99 percent chance of being real. That’s very high confidence, but I’ll note that given how many were found, even a 99 percent chance means that some of them will likely turn out to be false positives. But that number will be low, and this new harvest is still huge, the largest single group of validated planets to date.
So how was this done?
Kepler orbits the Sun, on a path very similar to Earth’s orbit but just outside ours. During its main mission, it stared at a single spot in space, observing 150,000 stars all the time. It looked for slight dips in the light from each star, which could be due to the planet passing directly between the star and us (if this sounds familiar, it’s precisely what Mercury did a few days ago, transiting the Sun).
These transits only occur for planets whose orbits are nearly edge-on as seen from Earth, which means that for every one we see, there are probably quite a few we miss. So Kepler puts a lower limit on how many planets are out there.
Even so, quite a few have been found! Up until this latest release, more than 1,000 such transiting verified exoplanets had been dug up from the Kepler data.
The problem is that word: “verified.” How do you know what you’re seeing is an actual planet? A lot of astronomical sources can muck with the data, creating false positives: things that look like planets, but are spurious. For example, some stars are binary, two stars orbiting each other. Some of these are eclipsing binaries; we see their orbit edge-on, so each star passes in front of the other every orbit. If they’re far enough away they look like a single star that goes up and down in brightness, just like an exoplanet transit.
Some of those can be eliminated using various methods. But that still leaves a big passel of candidate planets—ones that might be real—that needs to be verified. Most of these stars are too faint to get reliable results from the ground (that’s why Kepler was launched into space!), so observatories on Earth can’t always follow up. And looking at thousands of stars for the length of time needed to confirm the planets would take years. Decades.
This is where the astronomers in the new study were clever. They took all the candidate planets in the catalog—7,470 “Kepler Objects of Interest”—and ran a statistical analysis on them. This included fitting a simple model to the transit plot, fitting various characteristics of the host star, and assigning predetermined probabilities that the transit is from some spurious source like an eclipsing binary.
Out of the 7,470 input stars, their software ran successfully on 7,056 (the remaining 414 had other problems that made the software results untrustworthy). Of these stars, they found that 1,935 had a 99 percent chance or greater of being real, live transiting exoplanets. Of those, 651 had previously been determined to be real in the Kepler database.
That left 1,284 possible exoplanets with a very high chance indeed of being real. As I said, a handful of these may yet turn out to be false positives, but the overwhelming majority are very likely to be real.
Continue reading by clicking the name of the source below.