How can machine learning help astronomers find Earth-like exoplanets? This is what a study recently accepted by Astronomy & Astrophysics seeks to clarify. A team of international researchers investigated how a novel neural network-based algorithm could be used to detect Earth-like exoplanets using data from the radial velocity (RV) detection method. This study could help astronomers develop more efficient methods to detect Earth-like exoplanets, which are traditionally difficult to identify in RV data due to the intense stellar activity of the parent star.
The study states: “Machine learning is one of the most efficient and successful tools for processing large amounts of data in the scientific field. Many machine learning-based algorithms have been proposed to reduce stellar activity and better detect planets with low mass and/or long orbital periods. These algorithms can be divided into two categories: supervised learning and unsupervised learning. The advantage of supervised learning is that the proposed model contains a large number of variables and is able to make relatively accurate predictions based on the training data.”
For the study, the researchers applied their algorithm to three stars to determine its ability to identify exoplanets in the stellar activity data: our Sun, Alpha Centauri B (HD 128621), and Tau ceti (HD 10700), with Alpha Centauri B being about 4.3 light-years from Earth and Tau ceti being about 12 light-years from Earth. After plugging simulated planetary signals into the algorithm, the researchers found that their algorithm was successful in identifying simulated exoplanets with potential orbital periods ranging from 10 to 550 days for our Sun, 10 to 300 days for Alpha Centauri B, and 10 to 350 days for Tau ceti. It's important to note that Alpha Centauri B currently has several potential but unconfirmed exoplanet detections, while Tau ceti's system currently has eight exoplanets listed as “unconfirmed.”
Furthermore, the algorithm identified that these results are consistent with Alpha Centauri B and Tau ceti potentially possessing exoplanets that are about four times the size of Earth and are also located in the habitable zones of these stars. After the researchers fed more stellar activity data into the algorithm, they found that the algorithm successfully identified a simulated exoplanet that is about 2.2 times the size of Earth and orbits at the same distance from our Sun as Earth.
The study's conclusions state: “In this paper, we developed a neural network to efficiently mitigate stellar activity at the spectral level and improve the detection of low-mass planets on timescales ranging from a few days to a few hundred days, corresponding to the habitable zone of Sun-like stars.”
While the study focused on searching for Earth-like exoplanets in the RV data, the researchers point out that additional data, including transit time, phase, and space-based photometry, could be used to identify Earth-like exoplanets. They stress that this is possible with the European Space Agency's PLATO space telescope mission, which is currently under development and scheduled for launch in 2026. After launch, it will be stationed at the Sun-Earth L2 Lagrange point on the opposite side of Earth from the Sun, where it will use the transit method to search for exoplanets in up to a million stars, with a focus on terrestrial (rocky) exoplanets.
PLATO mission discussed around 9:00 am
This study comes as the number of exoplanets confirmed by NASA has reached 5,632 at the time of this writing, including 201 terrestrial exoplanets, and it provides ample opportunity for the upcoming PLATO mission to discover many more terrestrial exoplanets in our Milky Way galaxy.
How will machine learning help astronomers discover Earth-like exoplanets in the coming years and decades? Only time will tell, and that's why we do science!
And as always, keep doing science and keep looking up!
Like this:
Is loading…
