Measuring precise radial velocities on individual spectral lines. III. Dependence of stellar activity signal on line formation temperature

Context. To enable radial velocity (RV) precision on the order of ~0.1 m/s required for the detection of Earth-like exoplanets orbiting solar-type stars, the main obstacle lies in mitigating the impact of stellar activity. Aims. This study investigates the dependence of derived RVs with respect to the formation temperature of spectral line segments. Methods. Using spectral synthesis, we compute for each observed wavelength point of unblended spectral lines the stellar temperature below which 50% of the emergent flux originates. We can then construct RV time series for different temperature ranges, using template matching. Results. With HARPS-N solar data and HARPS $\alpha$ Cen B measurements, we demonstrate on time intervals of prominent stellar activity that the activity-induced RV signal has different amplitude and periodicity depending on the temperature range considered. We compare the solar measurements with simulated contributions from active surface regions seen in simultaneous images, and find that the suppression of convective motion is the dominant effect. Conclusions. From a carefully selected set of spectral lines, we are able to measure the RV impact of stellar activity at various stellar temperatures ranges. We are able to strongly correlate the effect of convective suppression with spectral line segments formed in hotter temperature ranges. At cooler temperatures, the derived RVs exhibit oppositely directed variations compared to the average RV time series and stronger anti-correlations with chromospheric emission.
Comment: Accepted for publication in A&A