Probing the atmospheres of transiting exoplanets using ground-based multi-object spectroscopy

Abstract

The field of exoplanetary science, the study of planets around stars other than the Sun, has undergone rapid expansion over the last few decades, with over 4000 confirmed exoplanets now known. With a wealth of favourable targets identified, it has recently become possible to begin to characterise the atmospheres of these planets in detail. Transmission and emission spectroscopy, performed when a planet passes in front of and behind its host star respectively, are powerful methods for probing atmospheric compositions. However, formidable challenges remain – even for the most favourable targets. Most instruments used for exoplanet spectroscopy were not originally designed for this purpose, leaving these observations vulnerable to instrumental systematics and correlated noise. In this thesis, I present ground-based observations of three hot Jupiters which I modelled using Gaussian processes, a technique which allows me to robustly disentangle the planetary signals from the systematic noise. In addition, I introduce Student’s-T processes for exoplanet time-series, which are a generalisation of Gaussian processes offering enhanced flexibility. My FORS2 observations of WASP-103b reveal a featureless optical spectrum and result in a detection of H2O when combined with GMOS, WFC3 and Spitzer data. The most likely explanation for this featureless spectrum may simply be a combination of low signal-to-noise and the inherently small atmospheric scale height, although patchy/inhomogeneous clouds or hazes may still play a part in damping the absorption features. Next, I present GMOS observations of the well studied hot Jupiter WASP-121b. I find little evidence for either TiO or VO, in agreement with a number of recent studies performed at high-resolution, which suggests that other absorbers, e.g. FeI, may be responsible for the vertical thermal inversion. Since both the GMOS and previously obtained STIS data show excellent repeatability over day-long periods, their difference over longer periods could well be the result of temporal variability in the atmospheric properties (i.e. weather) as predicted by theoretical models of ultra-hot Jupiters. Finally, I present an optical FORS2 reflection spectrum of the hot Jupiter WASP-43b. However, due to the limitations of systematic noise, I was only able to determine a conservative 3-𝜎 upper limit of 0.86 for the geometric albedo. This poorly constrained value highlights the difficulties involved in performing optical emission spectroscopy from the ground and suggests that we will likely need to improve on our techniques if such data are going to contribute to our understanding of optical reflection spectra. These observations demonstrate both the potential and challenges involved for ground-based observations of exoplanet atmospheres.