Reading Between the Rainbows: Comparative Exoplanet Characterisation through Molecule Agnostic Spectral Clustering

TL;DR

This paper introduces a molecule-agnostic spectral clustering method to analyze exoplanet atmospheres using JWST data, enabling differentiation of atmospheric types without relying on specific molecular features.

Contribution

The study presents a novel clustering approach that correlates spectral features to planetary characteristics, independent of molecular signatures, improving atmospheric classification accuracy.

Findings

Successfully separated reducing and oxic atmospheres

Constrained CO₂ and O₂ mixing ratios within two orders of magnitude

Demonstrated effectiveness on synthetic JWST spectra

Abstract

Rocky exoplanets are faint and difficult to observe due to their small size and low brightness compared to their host star. Despite this, the James Webb Space Telescope (JWST) has allowed us new methods and opportunities to study them. Accurately characterising exoplanet atmospheres could offer insights not only into the planetary demographics of rocky worlds in the universe, but also how our own Earth compares. Previous work simulating the observational spectra of planets with various models of atmospheric composition has constrained conditions under which specific molecules would be observable. We seek a different approach that does not depend on specific molecular features, but rather on correlating "agnostic" spectral characteristics to each other. This study uses 42 synthetic transit transmission spectra in a range matching that of JWST's MIRI instrument. Spectra were de-noised, normalised and split into bands for which the enclosed areas were calculated. Using the HDBSCAN clustering algorithm, separation patterns and clusters were found across these bands, which correlated to various a priori planet characteristics, such as relative gas concentration. Using 0.7 $\mu$m bands between 7 and 12 $\mu$m, we are able to separate reducing vs. oxic atmospheres and constrain CO$_2$ and O$_2$ mixing ratios accurate to within two orders of magnitude in an unknown sample. Through this, we have demonstrated that it is possible to differentiate and analyse exoplanet atmosphere type from agnostic interrogation of their transit spectra with a method we call "Molecule Agnostic Spectral Clustering".