Convective blueshifts in the solar atmosphere II. High-accuracy observations of the Fe I 6173.3 Å line and deviations of full-disk Dopplergrams

Context. Granular convective motions reach into the lower solar atmosphere, typically causing photospheric spectral lines to exhibit a differential line shift. This Doppler shift to shorter wavelengths is commonly known as convective blueshift. Aims. Spectroscopic high-accuracy measurements provide us with a refined determination of the absolute convective blueshift and its atmospheric distribution from disk center to the solar limb. Methods. We performed systematic observations of the quiet Sun with the Laser Absolute Reference Spectrograph (LARS) at the German Vacuum Tower Telescope. The solar disk was scanned along the meridian and the equator, from the disk center toward the limb. The solar spectrum around 6173 angstrom was calibrated with a laser frequency comb on an absolute wavelength scale with an accuracy of a few meters per second. We applied a bisector analysis on the spectral lines to reveal the changes of convective blueshift and line asymmetry at different heliocentric positions. Results. Being a signature for convective motions, the bisector curve of Fe I 6173.3 angstrom describes a C-shape at disk center. When approaching the solar limb, the bisector transforms into a \-shape. The analysis of the time-and bisector-averaged line shifts yields three distinct results. Firstly, the center-to-limb variation of Doppler velocities measured with LARS reveals a significant discrepancy (up to 200 m s(-1)) to the full-disk Dopplergrams of the Helioseismic and Magnetic Imager (HMI). Secondly, we obtained a significant decrease of convective blueshift toward the solar limb. Thirdly, the line-of-sight effect of solar activity, including p-mode oscillations and supergranular flows, leads to a scatter of up to +/- 100 m s(-1) at intermediate heliocentric positions. Conclusions. The accurate observation of the absolute convective blueshift with LARS allows the identification of systematic discrepancy with Doppler velocities measured by HMI. The center-to-limb variation of HMI suffers from an additional blueshift for mu < 0.9 that is incompatible with our results. LARS measurements can be taken as a reference for the correction of systematic errors in the synoptic HMI Dopplergrams.