2018MNRAS.481.4698F

A comprehensive analysis of 38 previously published Wide Field Camera 3 (WFC3) transmission spectra is performed using a hierarchy of nested-sampling retrievals: with versus without clouds, grey versus non-grey clouds, isothermal versus non-isothermal transit chords, and with water, hydrogen cyanide, and/or ammonia. We revisit the 'normalization degeneracy': the relative abundances of molecules are degenerate at the order-of-magnitude level with the absolute normalization of the transmission spectrum. Using a suite of mock retrievals, we demonstrate that the normalization degeneracy may be partially broken using WFC3 data alone, even in the absence of optical/visible data and without appealing to the presence of patchy clouds, although lower limits to the mixing ratios may be prior-dominated depending on the measurement uncertainties. With James Webb Space Telescope-like spectral resolutions, the normalization degeneracy may be completely broken from infrared spectra alone. We find no trend in the retrieved water abundances across nearly two orders of magnitude in exoplanet mass and a factor of 5 in retrieved temperature (about 500-2500 K). We further show that there is a general lack of strong Bayesian evidence to support interpretations of non-grey over grey clouds (only for WASP-69b and WASP-76b) and non-isothermal over isothermal atmospheres (no objects). 35 out of 38 WFC3 transmission spectra are well fitted by an isothermal transit chord with grey clouds and water only, while 8 are adequately explained by flat lines. Generally, the cloud composition is unconstrained.