How the transit method finds planets

Most planets beyond the Solar System have never been seen directly. They are found from their effect on their star’s light. The most productive way to do that is the transit method: watch a star’s brightness for long enough, and if a planet’s orbit is lined up with our view, the star will dim slightly each time the planet crosses its face. Around three quarters of the more than 6,000 exoplanets confirmed so far were found this way.

What a transit looks like

A transit is the small dip in brightness as a planet crosses its star. Astronomers measure it with a light curve, a record of a star’s brightness over time. Outside a transit the light curve is flat, apart from noise. When the planet moves onto the disc, the brightness falls; while the planet is fully in front of the star it stays low; and it recovers as the planet moves off.

The bottom of the dip is curved rather than flat because a star is darker near its edge than at its centre, an effect called limb darkening. A planet blocks more light when it covers the bright centre of the disc than when it covers the edge.

How deep is the dip?

The fraction of light blocked is roughly the area of the planet’s disc divided by the area of the star’s disc:

depth ≈ (R_p / R_★)²

The dips are small. Jupiter’s radius is about a tenth of the Sun’s, so a distant observer would see the Sun dim by about 1% as Jupiter crossed it. Earth’s radius is 0.00916 of the Sun’s, which gives a dip of about 84 parts per million, or 0.0084%. That is like a 12,000-watt floodlight dimming by one watt. Measuring Earth-sized transits requires photometry from space, above the turbulence of the atmosphere.

Because the depth depends on the ratio of radii, the same planet is easier to detect around a smaller star. An Earth-sized planet crossing a red dwarf a fifth the size of the Sun produces a dip 25 times deeper.

How often, and how long?

A transiting planet repeats its dip once per orbit, so the spacing between dips gives the orbital period directly. Kepler’s third law then gives the size of the orbit from the period and the star’s mass. The length of each transit depends on the size of the star and how fast the planet moves across it: Earth would take about 13 hours to cross the Sun, while a hot Jupiter on a three-day orbit takes two to three hours.

Only a small fraction of planets transit at all. A transit is visible only if the orbit is nearly edge-on to us, and the chance of that is about the star’s radius divided by the orbit’s radius:

probability ≈ R_★ / a

For Earth around the Sun that is about 0.47%, or 1 in 215. For a hot Jupiter at 0.05 AU it is about 9%. This is why transit surveys watch tens of thousands to millions of stars at once, and why they find close-in planets far more easily than distant ones.

Planets on long orbits are hard to catch for a second reason. A survey that watches a patch of sky for a month will see a planet with a one-year orbit transit at most once, if at all. A single transit gives a depth and a duration but no period, and pinning that period down takes further observations.

What a transit tells you, and what it does not

• Size. The depth gives the planet’s radius relative to its star. The planet’s radius in kilometres is only as good as the estimate of the star’s radius, which is why precise stellar measurements, such as distances from the Gaia mission, matter so much.
• Orbit. The period, and with the star’s mass the orbital distance. The shape of the dip also constrains how centrally the planet crosses the disc.
• Not the mass. A transit alone says nothing about how heavy the planet is. Mass comes from the radial velocity method, which measures the star’s wobble, or from small variations in transit times caused by other planets. Radius and mass together give the planet’s density, and so a first idea of whether it is rocky or gaseous.

What can imitate a transit

Not every periodic dip is a planet. Most transit-like signals in survey data turn out to be something else, and telling them apart is most of the work.

• Eclipsing binaries. Two stars orbiting each other can eclipse one another. When one star only grazes the other, the eclipse can be as shallow as a planetary transit.
• Blends. A faint eclipsing binary close to the target on the sky, often unresolved by the telescope, adds its deep eclipses to the target’s light. Diluted by the brighter star, they look like a shallow transit.
• Stellar variability and instrument noise. Starspots, pulsations, spacecraft motion and detector artefacts can all produce dips, sometimes repeating ones.

Several clues give an impostor away. An eclipsing binary often produces a V-shaped dip rather than a U-shaped one, alternating dips of different depths, or a shallower secondary eclipse halfway between the main ones. A planet’s transits are the same depth each time and its secondary eclipse is usually too faint to detect. If the dimming comes from a neighbouring star, the centre of light on the detector shifts during the dip, which can be measured by comparing images taken in and out of transit. Spectra of the star, high-resolution imaging to look for close companions, and radial-velocity measurements rule out the remaining cases.

From signal to planet

Because false positives are common, the words matter. A signal is a detected dip that has not yet been checked. A candidate is a signal that has passed vetting against the false-positive scenarios above. A planet is a candidate confirmed by further observation, usually a mass measurement from radial velocities, or validated statistically by showing that every false-positive explanation is far less likely than a planet. Many candidates wait years for that last step.

The missions behind the method

NASA’s Kepler mission, launched in 2009, stared at a single field of about 150,000 stars for four years and then observed a series of fields along the ecliptic as K2. It showed that small planets are common and found more than 2,600 confirmed planets. NASA’s TESS, the Transiting Exoplanet Survey Satellite, launched in April 2018 and surveys nearly the whole sky in strips, watching each patch for about 27 days and returning to many of them in later years. It concentrates on bright, nearby stars, whose planets are the easiest to follow up.

Both missions’ data are public. Light curves and images are archived at MAST, the Mikulski Archive for Space Telescopes, and anyone can download and analyse them. Candidates and follow-up observations are coordinated through ExoFOP, and confirmed planets are catalogued in the NASA Exoplanet Archive.