Produced by: Tarun Mishra
Modern astronomy heavily relies on AI and machine learning (ML) to manage vast data from telescopes, with ML efficiently identifying patterns in large datasets.
ML plays a crucial role in searching for biosignatures on Earth-like exoplanets, which is a key focus in contemporary astronomy.
Astronomers use transmission spectroscopy to study starlight passing through an exoplanet's atmosphere, identifying molecules by analyzing the split light into different wavelengths.
Identifying chemical biosignatures is challenging due to natural abiogenic processes, stellar activity, weak atmospheric signals, and interference from clouds or haze.
Rayleigh scattering and other "noises" can degrade the signal-to-noise ratio (SNR), complicating the detection of molecular absorption lines in spectroscopic data.
Researchers developed an ML tool to classify transmission spectra with low SNR, focusing on potential biosignatures like methane, ozone, and water for follow-up observations.
The model was trained on synthetic atmospheric spectra, particularly from the TRAPPIST-1e planet, which is ideal for testing due to its Earth-like properties.
The ML system successfully identified synthetic atmospheres containing methane and ozone, relevant markers for biological life, especially similar to Proterozoic Earth's atmosphere.
The model can streamline the search for habitable exoplanets, optimizing JWST's limited observing time by quickly identifying promising candidates for detailed follow-up studies.