| dc.description.abstracteng | One of the big driving forces behind exoplanet research is the search for an Earth-twin to answer the old question of whether or not humanity is alone in the universe. According to simple logic as well as our current understanding of planet formation, planet Earth should have a twin somewhere within a few hundred lightyears. A planet of similar mass, composed primarily of rock and orbiting within the habitable zone around a star similar in age and temperature to our Sun. Yet, after nearly three decades of active searches for planets around other stars, we have not managed to find one. For the first twenty years this was expected as Earth is a rather small planet compared to even the giants in our Solar System, while our Sun is well above average in terms of stellar mass, leading to only a very weak signal and requiring an instrumental precision far in excess of what was available at the time. Over the last decade, instrumentation has reached the point where detecting an Earth twin would become possible, if it was not for variable signals intrinsic to the stars themselves that can hide and even mimic the signals of small exoplanets. Overcoming the challenge of these variable stellar signals by finding ways to mitigate their effects on, and disentangle them from, observations is a major part of contemporary exoplanet research. To this end, three main contributions to stellar variability and their interplay were investigated: (i) Acoustic oscillations of the entire star, excited by turbulent motion. (ii) Starspots and rotationally modulated phenomena related to magnetically active regions. (iii) Convective blueshift, its suppression through magnetic activity, and how it can be robustly determined. The scaling of acoustic oscillations with stellar mass along the main sequence and with age along the red giant phase were explored as well as how mitigating oscillatory radial velocity variations through well chosen integration times could be possible. Starspot covering maps for a range of stellar effective temperatures and activity levels were created and its effect through rotation on observed line profiles and radial velocites simulated. An empirical, data-driven, model-independent way to determine convective blueshift strengths was developed and its efficacy analyzed. The basis is an ultra-high quality solar template that was created and limits for the applicability of the technique in signal-to-noise, instrumental resolution and stellar rotation were determined. The technique was applied to determine convective blueshift strengths along the main sequence for 810 stars between 3500\,K and 6200\,K and for 241 post-main sequence stars spanning from subgiant to asymptotic giant stars. A strict scaling relation of convective blueshift along the main sequence to the third power of the effective temperature was found together with a plateau for K-type dwarfs. M dwarfs showed no discernible convective shift. Post-main sequence stars show stronger convective shifts compared to main sequence stars of similar temperature but vary a lot less overall with increasing age. A small minimum was found for the earliest red giants, just after the subgiant transition, while convective blueshift strengths determined in this way were found to increasingly decouple past that point from expectations from analytical models, macroturbulent dispersion and 3D MHD derived velocities, likely due to changes in the large-scale convective structure. | de |