My main research concerns the understanding and mitigation of stellar activity. Under the supervision of Prof. Xavier Dumusque, I am part of the project ''Signal COrrection to Reveal other Earths'' (SCORE) funded by the European Research Council (ERC).
Sun-as-a-star observations and their quality control
The HARPS and HARPS-N instruments have obtained continuous, high S/N, high precision radial velocities (RVs) of the Sun. However, without simultaneous meteorological monitoring, these observations will occasionally be influenced by clouds and, in the case of HARPS-N located in La Palma, dust winds from the Sahara called calima. These weather phenomena introduce one-sided outliers in the linear relation expected between the apparent magnitude of the Sun and its airmass, known as the extinction, as it traverses the sky. The extinction slope is also of importance to correct the RVs from differential extinction due to the tilt of the Sun's rotation axis relative to the horizon normal, causing a flux inbalance between the red- and blueshifted hemispheres.
As a consequence of the non-gaussian nature of the noise, the method of least-square fitting becomes biased and insufficient. Instead, I perform Markov Chain Monte Carlo (MCMC) analysis to simultaneously model a fore- and background population by exploring the parameter space. Marginalizing over the likelihoods allows for the definition of quality flags on each individual measurement, which are readily available on the ''Data & Analysis Center for Exoplanets'' (DACE) platform.
Radial velocities based on the formation temperature of spectral lines
Convection in the envelopes of low-mass stars create surface granulation patterns consisting of ascending hot granules and descending cool intergranular lanes. Light emitted from these different regions will contain spectral lines with varying strengths (due to the difference in intensity) and Doppler shifts (due to the convective motion). For unresolved sources, i.e. most stars, the disk-averaged spectral lines will be asymmetrical and have a net blueshift with respect to the stellar restframe.
Spectral lines formed at different depths will also be differently affected by the convection, known as the third signature of granulation. In previous efforts to mitigate the impact of stellar activity on RV measurements, several studies have investigated the dependency on a line-by-line (LBL) basis. It has been found that some lines are more strongly correlated with activity signals, and that the signal amplitude decreases as a function of line depth when averaging over certain depth intervals.
In my work, I explore a new approach in calculating LBL RVs based on the average formation temperature, obtained from spectral synthesis, of sampled points across individual line profiles. RVs are then computed and averaged on segments of lines divided by binning the range of formation temperatures. By doing so, the aim is to disentangle activity signals originating at different layers, corresponding to different temperatures, of the stellar atmosphere.
