Combined analysis of stellar and planetary absorption lines via global forward-transit simulations

Abstract

Context. Transit spectroscopy of exoplanets has led to the detection of many species whose absorption signatures trace their atmospheric structure and dynamics. Improvements in resolution and sensitivity have, however, revealed biases induced by stellar lines occulted by the transiting planet. Aims. We characterise the planet-occulted line distortions (POLDs) in absorption spectra that arise from proxies used for the occulted stellar lines and investigate the impact of stellar rotation, centre-to-limb variations, and broadband limb-darkening. Methods. We used the EVaporating Exoplanets (EVE) code to generate realistic stellar spectra during the transit of exoplanets, accounting for the 3D geometry of the system’s architecture and atmospheric transit, as well as for spectral variations over the stellar disc. The absorption spectra were calculated using approaches drawn from the literature and compared to the expected signal. Results. The POLDs from stellar rotation are dominant for moderate to fast rotating stars, reaching amplitudes comparable to atmospheric signals, but they can be mitigated by shifting the stellar line proxies to the radial velocity of the planet-occulted region. Centre-to-limb variations become dominant for slow rotators and are more easily mitigated at the stellar limb. We re-interpret the ESPRESSO data of two iconic systems and confirm that the sodium signature from HD 209458 b mainly arises from POLDs. However, we unveil a possible contribution from the planetary atmosphere that warrants further observations. For MASCARA-1 b, we did not find evidence for atmospheric sodium absorption and we can fully explain the observed signature by a POLD for super-solar stellar sodium abundance. Conclusions. We studied POLDs dependency on star and planet properties, and on the proxy used for planet-occulted lines. Distinguishing planetary absorption signatures from POLDs is challenging without access to accurate estimates of the local stellar spectrum and system orbital parameters. We propose a way to mitigate POLDs and improve atmospheric characterisation, by using simultaneous forward modelling of both the star and the planet to simulate the global observed signatures.