UC Irvine

As one of the ultimate goals of the field of exoplanet science — discovering an Earth-mass planet orbiting a Sun-like star at 1 AU — begins to come into reach from the perspective of instrumentation, there are still many challenges in data analysis to overcome before such a discovery can be realized. The most promising method for discovering this Earth-twin, the radial velocity (RV) method, is one of the most successful ways of finding exoplanets. Through measuring the Doppler shift of light from a star due to the gravitational interaction between star and planet, the RV method looks for periodic signals in the changing velocity of the star. When a periodic signal is identified as planetary in origin, RVs also allow us to measure the mass of the planet.

In addition to a powerful discovery technique, the RV method is a powerful follow up tool. Planet mass, combined with radius as obtained from the transit method, allows us to compute the planet's bulk density. With density, it is possible to theorize what materials make up these planets and begin to ask questions concerning habitability. On this note, without a precise mass one cannot properly interpret transmission spectra and reveal insights into a planet's atmosphere. Planet mass also allows for investigations of planet formation and evolution. Lastly, without planet masses, one cannot investigate the dynamics of a system. Without a well-measured mass from RVs, a planet and its greater system cannot be characterized in detail.

While the new era of Extreme Precision Radial Velocity (EPRV) spectrographs is allowing us to reveal ever smaller RV signals in our data sets, not all of these signals are planetary in origin. We expect a planet's signal to be strictly periodic: the frequency, amplitude, and phase of the signal should be consistent across all time spans. But more and more we are learning that quasi-periodic stellar activity induced signals plague our data sets. These signals are not consistent in time, frequency, amplitude, or phase. They come and go, sometimes stronger or weaker, with or without phase shifts. But nonetheless, our current methods for detecting periodic signals have trouble distinguishing between a strictly periodic planetary signal and a quasi-periodic activity signal. Properly and consistently identifying and characterizing activity will be the prime inhibitor to small planet detection and, by extension, their mass measurements. In this dissertation, I address the issue of distinguishing between strictly and quasi periodic signals in RV data sets.

First, I demonstrate the challenge, importance of, and lessons learned from measuring small RV signals to determine precise masses of planets. I use the exciting 5+ planet system HD 191939 as a demonstration of the importance of detailed intra-system studies. This scaled-down Solar System hosts three inner transiting sub-Neptunes, a shepherding Saturn-mass planet, and a very long period super-Jupiter. Our precise mass measurements allow for a deep dive into the compositions, formation, evolution, and dynamics of this system in a way that many more systems will be treated in the future.

Second, I describe my work on identifying quasi-periodic signals in RV data sets. Using Barnard's star as a test case, I explore how the 145-day stellar rotation period signal was aliased at one year to create a 233-day false positive planet detection. In the process, I develop new techniques for isolating the localized activity signal in time. This example stresses the importance of understanding the lifetime of all signals in a RV data set.

Lastly, with the special case of Barnard's star as motivation, I design and build a new software package, the l1 Apodized Periodogram, or Lia, with the explicit goal of estimating the decay lifetime of periodic signals. Lia will help us better understand our RV data sets, providing a holistic picture of each signal which allows us to more confidently describe their astrophysical origins.

In all, the main effort of my dissertation is to develop new methods and techniques for consistently distinguishing between strictly and quasi periodic signals in RV data sets. This effort will help set us up for discovering small RV signals and more precisely measuring known planet masses. Both of these endeavors will be vital in the broader search for discovering a true Earth-twin planet.