Detecting molecules in Ariel low resolution transmission spectra

TL;DR

This paper introduces a P-statistic method to detect molecules in low-resolution exoplanet spectra from the Ariel mission, demonstrating its predictive power and limitations based on model complexity.

Contribution

The paper presents a novel P-statistic approach for molecular detection in low-res spectra, assessing its effectiveness and biases in exoplanet atmospheric analysis.

Findings

P-statistic correlates well with true abundances when models are sufficiently complex.

The method's reliability decreases with simpler, less comprehensive models.

Forecasting biases do not significantly impair survey classification.

Abstract

The Ariel Space Mission aims to observe a diverse sample of exoplanet atmospheres across a wide wavelength range of 0.5 to 7.8 microns. The observations are organized into four Tiers, with Tier 1 being a reconnaissance survey. This Tier is designed to achieve a sufficient signal-to-noise ratio (S/N) at low spectral resolution in order to identify featureless spectra or detect key molecular species without necessarily constraining their abundances with high confidence. We introduce a P-statistic that uses the abundance posteriors from a spectral retrieval to infer the probability of a molecule's presence in a given planet's atmosphere in Tier 1. We find that this method predicts probabilities that correlate well with the input abundances, indicating considerable predictive power when retrieval models have comparable or higher complexity compared to the data. However, we also demonstrate that the P-statistic loses representativity when the retrieval model has lower complexity, expressed as the inclusion of fewer than the expected molecules. The reliability and predictive power of the P-statistic are assessed on a simulated population of exoplanets with H2-He dominated atmospheres, and forecasting biases are studied and found not to adversely affect the classification of the survey.