Characterizing Earth Analogs in Reflected Light: Atmospheric Retrieval Studies for Future Space Telescopes

Space-based high contrast imaging mission concepts for studying rocky exoplanets in reflected light are currently under community study. We develop an inverse modeling framework to estimate the science return of such missions given different instrument design considerations. By combining an exoplanet albedo model, an instrument noise model, and an ensemble Markov chain Monte Carlo sampler, we explore retrievals of atmospheric and planetary properties for Earth twins as a function of signal-to-noise ratio (SNR) and resolution ($R$). Our forward model includes Rayleigh scattering, single-layer water clouds with patchy coverage, and pressure-dependent absorption due to water vapor, oxygen, and ozone. We simulate data at $R = 70$ and $R = 140$ from 0.4-1.0 $\mu$m with SNR $= 5, 10, 15, 20$ at 550 nm (i.e., for HabEx/LUVOIR-type instruments). At these same SNR, we simulate data for WFIRST paired with a starshade, which includes two photometric points between 0.48-0.6 $\mu$m and $R = 50$ spectroscopy from 0.6-0.97 $\mu$m. Given our noise model for WFIRST-type detectors, we find that weak detections of water vapor, ozone, and oxygen can be achieved with observations with at least $R = 70$ / SNR$\ = 15$, or $R = 140$ / SNR$\ = 10$ for improved detections. Meaningful constraints are only achieved with $R = 140$ / SNR$\ = 20$ data. The WFIRST data offer limited diagnostic information, needing at least SNR = 20 to weakly detect gases. Most scenarios place limits on planetary radius, but cannot constrain surface gravity and, thus, planetary mass.