Bayesian atmospheric retrieval for exoplanets : uniqueness of exoplanet spectra, characterizations of super-earths, and evaluations of dedicated space telescope designs

Abstract

After centuries of studying the eight planets in our solar system, recent improvements in technology have given us the unprecedented opportunity to detect planets orbiting stars other than the sun, so-called exoplanets. Recent statistical studies based on 800 confirmed planets and more than 3000 planet candidates suggest that our galaxy is teeming with billions of planets. Many of them are likely to orbit their host stars at a distance where liquid water and potentially life can exist. Spectroscopic observations of exoplanets can provide us with information about the atmospheres and conditions on these distant worlds. This thesis presents a Bayesian retrieval framework to analyze spectroscopic observations of exoplanets to infer the planet's atmospheric compositions, the surface pressures, and the presences of clouds or hazes. I identify what can unambiguously be determined about the atmospheres of exoplanets by applying the retrieval method to sets of synthetic observations. The main finding is that a unique constraint of the atmospheric mixing ratios of all infrared absorbing gases and up to two spectrally inactive gases is possible if the spectral coverage of the observations is sufficient to (1) determine the broadband transit depths in at least one absorption feature for each absorbing gas and (2) measure the slope and strength of the molecular Rayleigh scattering signature. For the newly discovered class of low-density super-Earths, with radii and masses intermediate between Earth and Neptune, I present an observational approach to distinguish whether these planets more closely resemble the giant planets in our solar system or whether they represent a completely new, potentially water vapor-rich type of planet. The approach discussed in this work represents the science case for the largest Hubble Space Telescope program ever awarded for a single exoplanet. The numerical methods and the conceptual understanding of atmospheric spectra presented in this thesis are key for the design of future space telescopes dedicated to the characterization of transiting exoplanets. I present an integrated design evaluation framework for the proposed Exoplanet Characterization Observatory (EChO) that simultaneously models the astrophysical signal and the telescope's payload module. I demonstrate that costly cryogenic cooling to observe the mid-infrared spectrum beyond ~ 11 [mu]m is not required while visible light observations down to - 400 nm are essential for the mission success. The observational study of exoplanet atmospheres is in its infancy and its pace is poised to accelerate as observational techniques are improved and dedicated space missions are designed. The methods developed in this thesis will contribute to constraining the atmospheric properties of a wide variety of planets ranging from blazingly-hot gas giants to temperate Earth-like planets.