Abstract
Context. Extreme-precision radial velocity observations used to search for low-mass extrasolar planets are hampered by astrophysical noise originating from stellar photospheres. Starspots are a particular nuisance when observing young and active stars. New algorithms are needed to overcome the stellar noise barrier in radial velocity measurements.
Aims. Using simulations of stellar spectra, we aim to test a technique, which we call GUSTS, that directly measures the contribution from starspots by using spectral features that are distinct from the rest of the stellar photosphere. Their contributions are expected to be anti-correlated with the starspot-induced radial velocity jitter of the star. This is reminiscent of high-dispersion observations of a transiting planet, which causes a Rossiter-McLaughlin effect but also leaves an atmospheric transmission signature that, in the case of spin-orbital alignment, is anti-correlated in radial velocity with the Rossiter-McLaughlin effect.
Methods. We simulated rotating stars with a single starspot to test the method. Synthetic spectral time series were averaged to obtain a virtual spot-free star spectrum. The individual spectra were subsequently convolved with a kernel, using single value decomposition, to match the average spectrum as closely as possible, after which this average spectrum was removed from the individual spectra. The residual spectra were subsequently searched for spot signatures using the template spot spectrum used to make the synthetic stars. We tested this method on a variety of spectra with different signal-to-noise ratios and investigated what data quality is needed to use this technique in practice.
Results. We demonstrate that the new GUSTS technique can work to reduce radial velocity jitter from starspots given a highly sampled, high S/N dataset. Though this alone cannot take radial velocity jitter down to the level to see Earth-like planets, it can be combined with other methods and can be used on starspot-dominated stars to detect smaller and farther planets. This technique could be useful for the future Terra Hunting Experiment, which will provide high S/N data with large samples.