The Detection, Characterization, and Retrieval of Terrestrial Exoplanet Atmospheres

Abstract

With the upcoming launch of the James Webb Space Telescope (JWST), the first light of thirty-meter class ground based telescopes in the 2020s, and plans in the works for a next-generation space-based telescope designed from the outset to probe exoplanet habitability and biosignatures, we are entering the era of terrestrial exoplanet science. Soon we will have the opportunity to study the presence, composition, habitability, and biosignatures of terrestrial exoplanet atmospheres, enabling new perspectives on comparative planetology, astrobiology, and our place as humans within the cosmos. In this dissertation, we explore optimal astronomical observations and theoretical modeling approaches for constraining terrestrial exoplanet environments. In the first part of this dissertation, we investigate the extent of terrestrial exoplanet science that may be achievable in the JWST era. We begin by studying the feasibility and observational cost of detecting and characterizing the composition of the seven Earth-sized TRAPPIST-1 exoplanets with JWST. By positing several terrestrial atmospheric compositions that are plausible given the tumultuous history of atmospheric escape expected for this late M dwarf system, we use JWST noise models to determine the optimal observing modes and number of transits/eclipses necessary to detect the planet's atmospheres. We show that the likely common presence of CO2 in terrestrial atmospheres should drive our ability to detect all seven TRAPPIST-1 planet atmospheres with JWST's NIRSpec Prism, although as many as 30 transits may be required if the planets are cloudy Venus-like worlds. We find that water may be prohibitively difficult to detect in both Venus-like and habitable atmospheres due to its presence lower in the atmosphere where transit transmission spectra are less sensitive. Although the presence of biogenic O2 and O3 will be extremely challenging to detect, abiotically produced oxygen from past ocean loss may be detectable for all of the TRAPPIST-1 planets via O2-O2 collisionally induced absorption features at 1.06 and 1.27 µm, or via NIR O3 features for the outer three planets. These results constitute a suite of hypotheses on the nature and detectability of highly evolved terrestrial exoplanet atmospheres that may be tested with JWST. Although the presence of clouds in the TRAPPIST-1 planet atmospheres are not expected to prevent the atmospheres from being detected, we use models of cloudy Venus-like exoplanets to demonstrate that their atmospheres may remain significantly misunderstood. We find that JWST/Mid-IR Instrument (MIRI) Low Resolution Spectrometer (LRS) secondary eclipse emission spectroscopy in the 6 µm opacity window could probe at least an order of magnitude deeper pressures than transmission spectroscopy, potentially allowing access to the sub-cloud atmosphere for the two hot innermost TRAPPIST-1 planets. In addition, we identify two confounding effects of sulfuric acid aerosols that may carry strong implications for the characterization of terrestrial exoplanets with transmission spectroscopy: (1) there exists an ambiguity between cloud-top and solid surface in producing the observed spectral continuum; and (2) the cloud-forming region drops in altitude with semimajor axis, causing an increase in the observable cloud-top pressure with decreasing stellar insolation. Taken together, these effects could produce a trend of thicker atmospheres observed at lower stellar insolation––a convincing false positive for atmospheric escape and a "mirage" of the cosmic shoreline. These studies emphasize that JWST should be able to detect terrestrial exoplanet atmospheres around the smallest of stars, but characterizing the unique nature of secondary atmospheres, including habitability and biosignature detection, may be beyond JWST's capabilities. In the second part of this dissertation, we investigate terrestrial exoplanet science cases that a future space telescope could undertake. A future coronagraph-equipped, direct-imaging telescope may offer the best chance to directly search for oceans on the surface of exoplanets. To explore the depths of ocean detection, we introduce a novel approach for exoplanet ocean detection that combines two previously distinct techniques––rotational mapping and ocean glint––into the most rigorous approach to date for determining exoplanet habitability. We demonstrate that time-series observations of Earth-like exoplanets at crescent phase can be used to search for the "blinking" of ocean glint as continents, which interrupt the ocean glint spot, rotate into and out of the illuminated planet crescent. This effect causes large-amplitude, time-varying signals in the reflected light, which can be used to longitudinally map the ocean glint. In our simulations, we find that the retrieved apparent albedo of ocean-bearing longitudes is increased by a factor of 5 at crescent phase due to the contribution from glint, compared to the albedo of the same longitudes at quadrature phase. Meanwhile, the predominately land-bearing longitudes exhibit no significant change in apparent albedo with phase. Thus, by using time-dependent information we can decompose the planet into unique spatial regions that provide greater leverage to identify the scattering properties of liquid water. Although these observations would be difficult to acquire in practice, we calculate that NASA's LUVOIR mission concept could use this method to detect oceans for the closest Earth-like exoplanet targets (e.g. within 10 pc). Constraining the prevalence of habitable and inhabited exoplanets is one of the main science objectives of the LUVOIR telescope concept. Accomplishing this ambitious objective will require directly-imaged reflectance spectra of many Earth-like exoplanets that can be analyzed for signs of habitability and biosignatures. We use the LUVOIR-A (15 m) and LUVOIR-B (8 m) coronagraph design specifications provided in the LUVOIR Final Report to perform exposure time calculations and determine the number of potentially habitable Earth-like exoplanets for which LUVOIR can undertake a robust search for habitability and biosignatures. We find that a dedicated one-year characterization program with LUVOIR-A (LUVOIR-B) would be capable of observing the spectra of 24 (11) exo-Earth candidates between 0.3 - 1.5 µm. We highlight the challenges faced when observing exo-Earths in the farthest UV and farthest NIR bandpasses, and we explore the effect to our spectral yields of imposing a target prioritization strategy based on preliminary evidence about each target. These spectral yields demonstrate that a LUVOIR-like telescope could be used to characterize the nature of a large enough sample of HZ terrestrial exoplanets to place statistical constraints of the prevalence of planetary habitability and life in the Solar neighborhood. In the third part of this dissertation, we develop, validate, and apply a novel terrestrial exoplanet atmospheric retrieval model. Exoplanet atmospheric retrieval models have become invaluable tools for the interpretation of observed exoplanet spectra, however surprisingly little work has been done to develop and test retrieval methods for terrestrial exoplanets despite the excitement for upcoming spectral observations of terrestrial exoplanets with JWST. We describe a novel exoplanet atmospheric retrieval algorithm built specifically for terrestrial exoplanets that may span a broad range of atmospheric temperatures, pressures, and compositions. Our retrieval is built around the physically rigorous line-by-line radiative transfer code, SMART, originally built for Solar System terrestrial remote sensing. Our new SMART Exoplanet Retrieval code (SMARTER) has been developed with a modular design featuring a suite of different forward and inverse models that enable atmospheric inference for a variety of different observing techniques––including transmission, emission, and reflectance spectroscopy––coupled to an assortment of different Bayesian retrieval methods––including optimal estimation, Markov chain Monte Carlo, nested sampling, and a novel machine learning approach. We validate SMARTER against high-resolution transmission spectrum observations of Earth observed in solar occultation, demonstrating key capabilities and challenges for interpreting the spectra of terrestrial exoplanets with complex atmospheres. Finally, to assess the sensitivity to biosignatures and test common retrieval assumptions, we employ our new SMARTER model to interpret the atmosphere of TRAPPIST-1e as if it were observed by an intensive JWST transmission spectroscopy campaign, assuming several different plausible habitable and uninhabitable archetypal planetary states. We find that the CH4-CO2 disequilibrium biosignature pair, previously suggested by Krissansen-Totton et al. for anoxic biospheres like the Archean Earth, may also be retrieved if the planet possesses a modern Earth-like atmosphere due to the increased photochemical lifetime of CH4 in the atmospheres of late M dwarf planets. However, enhanced CH4 absorption features only worsen the already dim prospects for detecting H2O, which could complicate the interpretation of planetary habitability. We also find that stacking 25 transits could lead to a weak detection of CO2 in a cloudy Venus-like atmosphere, or alternatively reveal O2 as the bulk atmospheric gas via O2-O2 CIA for an oxygen-dominated post-runaway atmosphere. We uncover that adding refraction to our transmission retrievals modifies the known degeneracy between reference radius and pressure allowing for an additional mode of interpretation wherein the spectral continuum is controlled by refraction rather than an opaque medium. We also identify a strong implicit prior on the mean molecular weight caused by the common assumption of a background gas (such as H2/He or N2) that can significantly affect the retrieved temperature via scale height degeneracies. Finally, we verify that the TRAPPIST-1e planet mass constraints from Transit Timing Variation (TTV) measurements are sufficient to enable dozens of observed transmission spectra to be stacked without incurring additional posterior uncertainties on atmospheric parameters. The results of this dissertation both underscore the compelling future and highlight some of the key challenges ahead for terrestrial exoplanet atmospheric characterization. In the near term, observations with JWST are well-positioned to begin detecting and characterizing terrestrial exoplanet atmospheres, but are ultimately unlikely to constrain the presence of habitable conditions or biosignatures even for the most optimistic targets. Investing resources in a future space telescope that is capable of directly imaging Earth-like exoplanets is crucial to optimize the ambitious pursuit of habitable and inhabited planets. In advance of these upcoming missions, our new retrieval model can continue to be used, tested, and improved upon to maximize our chances of correctly interpreting the first precise spectra of rocky exoplanets.