Planetary transits and stellar variability

Abstract

This thesis focuses on the detection of extrasolar planets via the transit method, and more specifically addresses issues relevant to the preparation of upcoming space missions such as CoRoT, Kepler, Eddington, aiming to detect terrestrial planets. The automated detection of transits in tens of thousands of noisy light curves is a challenging task because of the brief, shallow and rare nature of the sought-after signal. I developed a Bayesian transit detection algorithm and tested it through Monte Carlo simulations. This led to a number of improvements to the algorithm, resulting in a maximum likelihood box-fitting approach that is simpler and, for detection purposes, more effective. I also developed an empirical model of the intrinsic micro-variability of the parent stars, which constitutes one of the main performance limitations for space-based transit searches, and used this model to optimise variability filters. These tools were used to identify optimal regions of the stellar parameter space to be targeted by missions such as CoRoT and Eddington, through Monte Carlo simulations and participation in the first CoRoT blind transit detection exercise. Recently, the same tools have been used to analyse ground-based data from the University of New South Wales planet search project, leading to the identification of several potential transiting planet candidates.