Answers are grounded in this paper and cite their sources.
Reflected-light spectra of Earth-sized exoplanets can probe habitability and biosignatures, but interpreting atmospheric oxygen is challenging because O 2 can arise from both biological and abiotic processes. In addition, reflected-light retrievals exhibit degeneracies among atmospheric composition, clouds, surface reflectance, and gravity. We use the Bayesian retrieval framework ExoReL ℜ , leveraging recent model developments that incorporate (i) O 2 –O 2 collision-induced absorption (CIA), which produces the O 4 bands, and (ii) a wavelength-dependent (step-function) surface albedo parameterization. Focusing on a modern-Earth analog, we compare retrievals with and without O 2 –O 2 CIA using spectra from 0.25–1.8 μ m and perform signal-to-noise ratio (SNR) sweeps for two observing configurations motivated by the Habitable Worlds Observatory: a resolved-visible case ( R VIS = 140) and a broadband-visible case (visible summarized into three broad bins). Omitting O 2 –O 2 CIA allows the retrieval to fit the spectrum with an incorrect O 2 -dominated atmosphere. Including O 2 –O 2 CIA leverages the O 4 bands to break this degeneracy and recover the correct N 2 -dominated, O 2 -rich solution without requiring a tight mass prior. Adopting a robustness criterion of ≥95% posterior mass in the N 2 -dominated region, both observing configurations require SNR ≳ 20 to identify the correct dominant gas, with a higher cloud fraction increasing this threshold. Reducing near-IR coverage to 1.4 μ m prevents recovery of the correct N 2 -dominated solution across the explored SNR range and instead favors a spurious O 2 -dominated, CO 2 -rich atmosphere. These results support near-IR coverage to ∼1.7 μ m and an SNR of at least ∼20 for robust oxygen interpretation on Earth-like planets.
No takes yet. Share an insight, caveat, or question.
Burri et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: