CROCODILE: Cross-correlation retrievals of directly-imaged self-luminous exoplanets

Obtaining medium or high-resolution spectroscopy of directly-imaged exoplanets is challenging. Cross-correlation spectroscopy, a technique where well-known molecular spectral signatures originating from the atmosphere of the planet are used to differentiate the signal coming from the planet to that of the PSF of the host star and from noise, has gotten traction in the recent years as a reliable method to detect molecular species in the atmospheres of exoplanets.

My work has been focused on developing and testing the Python package CROCODILE (CROss-COrrelation retrievals of Directly-Imaged self-Luminous Exoplanets), which integrates this format of data into the Bayesian statistical framework of atmospheric retrievals. It can combine medium-resolution cross-correlation spectroscopy with low-resolution spectroscopy and photometry to obtain strong constraints on atmospheric properties of directly-imaged exoplanets such as molecular abundance ratios and thermal structure. I demonstrate the capabilities of CROCODILE using simulations of exoplanetary emission spectra from different configurations of the parameter space of exoplanets. Furthermore, I identify two cases, namely high C/O ratio and low metallicity, where free chemistry fails to reproduce the properties of the input atmosphere although the synthetic spectra are well fit, hinting at a modelling degeneracy.

In summary, CROCODILE provides the statistical framework to interpret the three main observables of directly-imaged exoplanetary atmospheres, namely photometry, low-resolution spectroscopy, and medium (and higher) resolution cross-correlation spectroscopy, all of which will be measured by the next generation of instruments such as ERIS at the Very Large Telescope, MIRI aboard the James Webb Space Telescope, and METIS at the future Extremely Large Telescope.