2015ApJ...808..171D

The HARPS spectrograph is showing an extreme stability, close to the m/s level, over more than 10 years of data. However, the radial velocities of some stars are contaminated by a spurious one-year signal with an amplitude that can be as high as a few m/s. This signal is in opposition of phase with the revolution of Earth around the Sun and can be explained by the deformation of spectral lines crossing block stitchings of the CCD when the spectrum of an observed star is alternatively blueshifted and redshifted due to the motion of Earth around the Sun. This annual perturbation can be suppressed by either removing those affected spectral lines from the correlation mask used by the cross-correlation technique to derive precise radial velocities, or by simply fitting a yearly sinusoid to the radial velocity data. This is mandatory if we want to detect long-period low-amplitude signals in the HARPS radial velocities of quiet solar-type stars.