STCTM: a forward modeling and retrieval framework for stellar contamination and stellar spectra

TL;DR

STCTM is a Bayesian framework designed to model and mitigate stellar contamination effects, specifically the transit light source effect, in exoplanet transmission spectra, enabling more accurate characterization of small exoplanet atmospheres.

Contribution

The paper introduces STCTM, a flexible Bayesian retrieval framework that models stellar contamination in transmission spectra and includes a sub-module for data-driven priors from stellar spectra.

Findings

Effective modeling of stellar surface heterogeneities impacts transmission spectra analysis.

The framework allows for inference of stellar surface parameters without planetary signals.

Parallelized implementation enables fast and flexible inferences.

Abstract

Transmission spectroscopy is a key avenue for the near-term study of small-planet atmospheres and the most promising method when it comes to searching for atmospheres on temperate rocky worlds, which are often too cold for planetary emission to be detectable. At the same time, the small planets that are most amenable for such atmospheric probes orbit small and cool M dwarf stars. As the field becomes increasingly ambitious in the search for signs of even thin atmospheres on small exoplanets, the transit light source effect (TLSE), caused by unocculted stellar surface heterogeneities, is becoming a limiting factor: it is imperative to develop robust inference methods to disentangle planetary and stellar contributions to the observed spectra. Here, I present STCTM, the STellar ConTamination Modeling framework, a flexible Bayesian retrieval framework to model the impact of the TLSE on any exoplanet transmission spectrum, and infer the range of stellar surface parameters that are compatible with the observations in the absence of any planetary contribution. With the "exotune" sub-module, users can also perform retrievals directly on out-of-transit stellar spectra in order to place data-driven priors on the extent to which the TLSE can impact any planet's transmission spectrum. The input data formats, stellar models, and fitted parameters are easily tunable using human-readable files and the code is fully parallelized to enable fast inferences. [shortened for arxiv; see full summary in the PDF]