Correcting HIRES radial velocities for small systematic errors

TL;DR

This paper identifies and corrects small systematic errors in HIRES radial velocity data, improving long-term precision and aiding exoplanet searches.

Contribution

The study reveals previously unrecognized nightly zero-point effects in HIRES RVs and provides corrected data, enhancing the accuracy of stellar RV measurements.

Findings

Detected a 1.5 m/s jump in 2004 due to instrument changes

Identified a long-term drift of up to 1 m/s

Found a 0.051 m/s/hr intra-night RV drift

Abstract

The HIRES spectrograph, mounted on the $10$-m Keck-I telescope, belongs to a small group of radial-velocity (RV) instruments that produce stellar RVs with long-term precision down to $\sim1$ ms$^{-1}$. In $2017$, the HIRES team published $64,480$ RVs of $1,699$ stars, collected between $1996$ and $2014$. In this bank of RVs, we identify a sample of RV-quiet stars, whose RV scatter is $<10$ ms$^{-1}$, and use them to reveal two small but significant nightly zero-point effects: a discontinuous jump, caused by major modifications of the instrument in August $2004$, and a long-term drift. The size of the $2004$ jump is $1.5\pm0.1$ ms$^{-1}$, and the slow zero-point variations have a typical magnitude of $\lesssim1$ ms$^{-1}$. In addition, we find a small but significant correlation between stellar RVs and the time relative to local midnight, indicative of an average intra-night drift of $0.051\pm0.004$ ms$^{-1}$hr$^{-1}$. We correct the $64,480$ HIRES RVs for the systematic effects we find, and make the corrected RVs publicly available. Our findings demonstrate the importance of observing RV-quiet stars, even in the era of simultaneously-calibrated RV spectrographs. We hope that the corrected HIRES RVs will facilitate the search for new planet candidates around the observed stars.