Quasar emission lines are often shifted from the systemic velocity due to various dynamical and radiative processes in the line-emitting region. The level of these velocity shifts depends both on the line species and on quasar properties. We study velocity shifts for the line peaks of various narrow and broad quasar emission lines relative to systemic using a sample of 849 quasars from the Sloan Digital Sky Survey Reverberation Mapping (SDSS-RM) project. The coadded (from 32 epochs) spectra of individual quasars have sufficient signal-to-noise ratio (SNR) to measure stellar absorption lines to provide reliable systemic velocity estimates, as well as weak narrow emission lines. The sample also covers a large dynamic range in quasar luminosity (~2 dex), allowing us to explore potential luminosity dependence of the velocity shifts. We derive average line peak velocity shifts as a function of quasar luminosity for different lines, and quantify their intrinsic scatter. We further quantify how well the peak velocity can be measured for various lines as a function of continuum SNR, and demonstrate there is no systematic bias in the line peak measurements when the spectral quality is degraded to as low as SNR~3 per SDSS pixel. Based on the observed line shifts, we provide empirical guidelines on redshift estimation from [OII]3728, [OIII]5008, [NeV]3426, MgII, CIII], HeII1640, broad Hbeta, CIV, and SiIV, which are calibrated to provide unbiased systemic redshifts in the mean, but with increasing intrinsic uncertainties of 46, 56, 119, 205, 233, 242, 400, 415, and 477 km/s, in addition to the measurement uncertainties. These more realistic redshift uncertainties are generally much larger than the formal uncertainties reported by the redshift pipelines for spectroscopic quasar surveys, and demonstrate the infeasibility of measuring quasar redshifts to better than ~200 km/s with only broad lines.
Download