The Science Behind the Validation of 1 284 Transiting Extrasolar Planets

An artistic representation of exoplanets. (Credit: NASA/W. Stenzel)
An artistic representation of exoplanets. (Credit: NASA/W. Stenzel)

On May 10, 2016 the Kepler team announced the validation of 1 284 planets which represents the largest haul of extrasolar planets in a single study.

The validation of a planet is the confirmation that a transit signal detected from Kepler observations is indeed due to a planet. The Kepler instrument detects planets by continuously monitoring over 150 000 stars to observe when the light from the star decreases due to an orbiting planet passing in front of the star. This event is known as a planetary transit and will occur periodically as the planet orbits its host star.   To date, Kepler has detected more than 5000 transit events that may be due to a transiting extrasolar planet. The key to the excitement over the validation announcement is that we no longer need to say we may have found planets, but can now, with high statistical significance, state that we have found planets, and not just a few, but 1 284. From this sample, more that 200 of the newly validated planets have radii that are similar to the Earth which suggests they could have a rocky composition and one these newly validated planets is at the right distance from its host star to be located in the habitable zone which permits the existence of liquid water.

The term validated indicated that there is a greater than 99% probability that the transit signal observed by Kepler is due to a transiting extrasolar planet and not due to other astrophysical phenomena or instrumental effects. The most common astrophysical phenomena are eclipsing binaries which are gravitationally bound stellar systems that are observed to eclipse one another. There are a few observational differences between eclipsing binaries and transiting planets, specifically: transit-depth, transit-shape, secondary events and photo-centroids.   The transit depth is proportional to the radius of the transiting body relative to the host star. Planets range in size from smaller than Mercury to moderately larger than Jupiter depending on the mass, composition and temperature of the planet. The smallest stars are about the size as Jupiter, whereas stars like the Sun are 10 times larger. This, the depth of an observed transit combined with knowledge of the physical properties of the host star is a good indicator as to whether the transit is due to a planet or a star.

A complication occurs when multiple the light from multiple stars is blended together. Kepler determines the brightness of a star through the use of CCD detectors that are similar to the detectors found in digital cameras. A CCD detector is composed of an array of pixels that, combined with the optics of the telescope, defines the resolution of the imager. Thus, if two stars are both imaged on the same pixel the light from both stars becomes blended. If one of the stars has a transiting companion the depth of the transit will be observed to be too shallow. The consequence is that a blended eclipsing stellar binary can appear to have the same depth as a transiting extrasolar planet.


An example of a light curve: the brightness of the star as a function of time, where we can see the shape of the transit in the form of a “U” (in the case of this figure) or a “V”. (Credit: NASA)

The transit shape is also an indicator of the size of a transiting body.   When the transiting object is small relative to the host star a transit lightcurve has a characteristic ‘U’ shape. When the size of the transiting body is comparable in size with the host star then the transit becomes ‘V’ shaped as the circumference of the transiting object never passes fully in front of the host star. The transit shape is another diagnostic to distinguish between a transiting planet and eclipsing binary.

The Kepler CCD imager and optics were designed to have the light from a point source, such as a distant star, spread across multiple pixels. This design enables the use of photo-centroids to determineif a stellar blend is present and which star is the likely source of the transit signal. Imagine you are viewing a car at a distance with both headlights on. If one of the highlights is obscured or turned off you would observe the center of light to move towards the brighter headlight. The same principle applies to the detection of transiting exoplanets: if the center of light is observed to move when a transit occurs then an additional source of light is present.

The observables from the Kepler lightcurve can be combined with known or calculable occurrence rates of stellar binaries and stellar blends to estimate the probability that any observed transit is due to a planet. The probability calculation was carried out for the entire Kepler planet candidate sample and then a simple question was asked: which planet candidates have a greater that 99% probability of being a planet. The answer is more than 2 000 with 1 284 being newly validated. By combining a statistical model with observables provided by Kepler we now have a powerful tool to sift through the large number of exoplanet discoveries which will only increase with the launch of future missions such as the NASA TESS mission which will follow up on the legacy of Kepler.