Origin and Extent of the Opacity Challenge for Atmospheric Retrievals of WASP-39 b

Abstract

As the James Webb Space Telescope (JWST) came online last summer, we entered a new era of astronomy. This new era is supported by data products of unprecedented information content that require novel reduction and analysis techniques. Recently, Niraula et al. (N22) highlighted the need for upgraded opacity models to prevent facing a model-driven accuracy wall when interpreting exoplanet transmission spectra. Here, we follow the same approach as N22 to explore the sensitivity of inferences on the atmospheric properties of WASP-39 b to the opacity models used. We find that the retrieval of the main atmospheric properties from this first JWST exoplanet spectrum is mostly unaffected by the current state of the community’s opacity models. Abundances of strong opacity sources like water and carbon dioxide are reliably constrained within ∼0.30 dex, beyond the 0.50 dex accuracy wall reported in N22. Assuming the completeness and accuracy of line lists, N22's accuracy wall is primarily driven by model uncertainties on broadening coefficients and far-wing behaviors, which we find to have marginal consequences for interpreting the transmission spectra of large, hot, high-metallicity atmospheres such as WASP-39 b, in opposition to emission spectra and climate modeling, which depend on deeper parts of a planetary atmosphere. The origin of the opacity challenge in the retrieval of metal-rich hot Jupiters via transmission spectroscopy will thus mostly stem from the incompleteness and inaccuracy of line lists.

Updated Spectral Characteristics for the Ultracool Dwarf TRAPPIST-1

“…This could be due to both limitations in stellar models and the marginalization of unknown systematic effects with retrieval tools, leading to large data-model residuals (Rackham & de Wit 2023). In addition-as found previously for opacity models (Niraula et al 2022(Niraula et al , 2023, we find a model-driven accuracy wall more than one order of magnitude above the precision accessible with JWST data, due to the lack of model fidelity (e.g., intermodel differences on the derived effective temperature at the level of ∼100 K while the instrument-driven uncertainty is at the level of ∼4 K). As highlighted by recent studies (Iyer et al 2023, and references therein), current M dwarf atmospheric models encounter several limitations.…”

“…Indeed, while the effective temperature of the dominant surface feature (T 1 ), for example, is consistent within the inferred uncertainties for a set of retrievals with a given stellar model, it significantly differs between models. For example, a fit of the NIRISS data with 1-comp using PHOENIX (fit 1) yields T 2662 1 4 4 = -+ K, while using SPHINX (fit 7) yields T 2567 1 5 5 = -+ K. In other words, while the instrument can yield constraints on the effective temperature T 1 to within ∼4 K, differences between stellar models (or their lack of fidelity) give rise to discrepancies between fits at the level of ∼100 K. Such a model-driven accuracy wall in the JWST era has previously been reported for opacity models (Niraula et al 2022(Niraula et al , 2023.…”

Empirically Constraining the Spectra of Stellar Surface Features Using Time-resolved Spectroscopy

“…Currently, the dominant bottlenecks for transmission spectroscopy are associated with imperfections in our opacity models (Niraula et al 2022) and stellar contamination (Rackham & de Wit 2023). The current limitations in opacity models result in an accuracy wall preventing constraints on most atmospheric properties beyond ∼0.5 dex for all planets but large, hot, and highly metallic ones (Niraula et al 2023). Future efforts supporting the standardization of databases, and the improvement of treatments of broadening and far-wings of line profiles, should mitigate this bottleneck.…”