A Cosmic Quest for New Worlds. Characterising Exoplanet Signals via Radial Velocity and Transit Photometry

Sammanfattning: Since the first unambiguous detection of a planet around a Sun-like star, the interest in the new and exciting field of exoplanets has grown immensely. New and exciting developments are seen at a pace unparalleled for most subfields of astronomy. In this thesis, I describe the two most successful techniques for exoplanet detection and characterisation – transits and radial velocities – and the challenges commonly encountered in extracting the planets from the data. Transit photometry allows us to measure the planet radius, while radial velocity measurements give us the planet’s minimum mass. These methods’ true strength, however, manifests in their combination as it allows us to estimate the true mass, which, together with the radius, gives us the planet’s bulk density. This is a powerful quantity, which allows us to construct models and make predictions about the structure and composition of a planet’s interior, as well as its atmosphere. Zeroing in on the latter two is currently one of the biggest challenges for exoplanet characterisation. I describe the process of detecting a planet in a stellar light curve, and how transits and radial velocities are modelled together in order to determine the planet parameters. This is then followed by the ideal theoretical approach, which can be used to study a system in practice. However, the current challenges in exoplanet characterisation surpass the ideal case, leading us to explore more complex models. I then discuss the biggest nemesis to planet discovery, particularly in radial velocity timeseries – stellar activity, and the problem of its often stochastic manifestation. A special focus is given to one method for its mitigation – modelling the radial velocities alongside activity indicators. This is the core concept of multi-dimensional Gaussian process regression, particularly with the quasi-periodic covariance function, which is used in a large part of this work. Finally, the last part if the thesis shows that while the ideal planet case can sometimes be applicable for quiet stars, as is the case of the TOI-2196 system, extending to non-parametric models, such as Gaussian processes, can help us to detect planets in complicated datasets, as demonstrated by the cases of the TOI-1260, TOI-733, TOI-776 and TOI-1416 systems.