Using NASA data and machine learning, scientists have found over 10,000 possible new planets in a single survey.
In a new study, researchers used machine learning to perform a sweeping survey of data from NASA’s exoplanet-hunting Transiting Exoplanet Survey Satellite (TESS). As a result, they and uncovered exactly 10,091 candidate planets that had never been seen before. To clarify, when planets beyond our solar system (otherwise known as exoplanets) are first spotted, they are considered “candidates” until they can be confirmed as such with the right amount of evidence. Some of these candidates might not end up being planets after all — some could end up being other objects or even just “noise” in the data.
This major haul of exoplanet candidates, however, may beg the question: Why haven’t these thousands of planets been seen before? In fact, TESS has been operational since 2018, and has been continuing on its extended mission since 2020.
Well, TESS operates by observing planets “transiting,” or passing in front of, their stars. In other words, when an exoplanet is orbiting its star, at some point it will cross the face of the star from TESS’ view. When this happens, the star appears to dim. TESS can measure that dimming, thereby revealing information about the planet passing in front of its star. Planets orbiting brighter stars are easier to spot because the transits are clearer.
However, this new study pulled data from fainter stars.
The survey actually looked at stars 16 times fainter than those typically targeted by TESS. Using machine learning, a type of artificial intelligence, the team surveyed over 83 million stars that were observed during TESS’ first year of observations. Of these many millions of faint stars TESS looked at, 10,091 appeared to have transiting, planet-like objects never seen before.
Jackpot
Of course, some of the 10,091 planet candidates might turn out to not be planets, so this research team wanted to start further testing to confirm these results. And in doing so, they were able to confirm one of the candidates: a planet called TIC 183374187 b. This world appears to be a hot Jupiter, a gas giant that orbits very close to its host star — which is why it’s so hot — and has a mass similar to Jupiter‘s.
But besides confirming all the exoplanets spotted with this research, the team also intends to take things a step further. While this study used the first year of TESS data, the research team aims to continue with a follow-up study using TESS’ second year of data, lead author Joshua Roth, a graduate researcher at Princeton University, told IFLScience.
This boom in exoplanet discovery comes just about 30 years since the first exoplanet to be confirmed, 51 Pegasi b, was found in 1995. With that first confirmation, scientists were able to definitively say what they had thought likely true for years — that planets exist around other stars outside of our solar system. Since then, NASA missions like TESS and Kepler have provided data that has grown the field tremendously.
Looking to the future, in addition to incorporating techniques like machine learning as this team has done, NASA’s upcoming Nancy Grace Roman Space Telescope will continue to expand exoplanet science with its Coronagraph Instrument which will make direct observations of worlds beyond our cosmic neighborhood. Rather than increase the number of exoplanets discovered, this instrument will allow for more in-depth study of these worlds and their atmospheres. As of writing this article, that telescope is slated to launch no earlier than early September 2026.
And even farther into the future, NASA aims to expand the field yet again with its Habitable Worlds Observatory, which is currently being built.
This work was published April 28 in The Astrophysical Journal.
