Gaia DR3 high radial velocity stars: Genuine fast-moving objects or outliers?

Abstract

The third Gaia data release includes 33.8 million radial velocity measurements, extending to a magnitude of G_RVS = 14. To reach this magnitude limit, spectra were processed down to a signal-to-noise ratio (S/N) of 2. In this very low S/N regime, noise-induced peaks in the cross-correlation function can result in spurious radial velocity determinations. Quality filters were applied to the dataset to mitigate such artefacts as much as possible prior to publication. Nevertheless, the high radial velocity (HRV) stars -- defined here as those with radial velocities below -500 or above +500 km/s -- are so sparsely populated that even a few hundred spurious measurements can lead to significant contamination. The objectives of the present study are as follows: (i) to confirm or refute the radial velocity values of the order of one hundred Gaia DR3 HRV stars, (ii) to evaluate the rate of spurious radial velocities in the Gaia DR3 catalogue as a function of S/N and radial velocity, and (iii) to examine the properties of the genuine HRV stars. A total of 134 Gaia DR3 HRV stars were observed using the SOPHIE and UVES spectrographs. (abridged) Ground-based measurements confirm the Gaia DR3 radial velocities of 104 out of our 134 targets, and they refute those of the remaining 30. The combination of these data with the spectroscopic surveys mentioned above enabled an assessment of the rate of spurious measurements as a function of S/N and across three intervals of absolute value of the radial velocity. (abridged) The majority of these stars follow retrograde orbits. Their location in the energy-vertical component of the angular momentum diagram coincides with the region where several structures associated with past merging events have been identified: Sequoia, Arjuna and I'itoi, Antaeus, ED-2, and ED-3. It is likely that most of these HRV stars were accreted.

Figures (9)

• Figure 1: Distribution of the 33.8 million Gaia DR3 radial velocities as a function of Galactic longitude. The 134 SOPHIE and UVES targets and the 47 APOGEE, GALAH, RAVE, and LAMOST stars (referred to as GBS) are represented by filled magenta circles and orange stars respectively. In this and the following figures, the 2 UVES and SOPHIE targets, that are also in APOGEE or LAMOST catalogues, are only indicated with filled circles, and not with stars.
• Figure 2: Gaia DR3 radial_velocity field versus SOPHIE and UVES radial velocities. The 104 confirmed stars are represented by filled cyan circles, the 27 spurious measures with a ground-based radial velocity by filled red circles and the shaded cyan area indicates the validity zone of $\pm 50$$\text{km~s}^{-1}$ around the main diagonal. For the sake of clarity, the central part of the figure, which was empty, has been omitted and has been replaced by a narrow grey band.
• Figure 3: Gaia DR3 radial velocities as a function of the Gaia DR3 S/N, for our 134 targets (filled circles) and for the 47 APOGEE, GALAH, LAMOST and RAVE stars (stars). The confirmed Gaia DR3 values are coloured in cyan and the spurious measures are coloured in red. As in Fig. \ref{['Fig:Confirmed']}, the central part of the figure has been omitted and replaced by a narrow grey band.
• Figure 4: Distribution in Galactic longitude ($l$) and latitude ($b$) of the 181 HRV stars. The figure uses a Mollweide projection. As in Fig. \ref{['Fig:Outliers1']}, our targets are represented by filled circles and the GBS by filled star symbols. Stars with confirmed Gaia DR3 radial velocities are displayed in cyan, while those with spurious measurements are indicated in red.
• Figure 5: Distribution of the Gaia DR3 radial velocities of the sample of 2.7 million stars used to assess the rate of outliers as a function of S/N and for three different intervals of absolute values of the radial velocities. The three intervals are materialised in green ([0, 200) $\text{km~s}^{-1}$), blue ([200, 400) $\text{km~s}^{-1}$) and salmon ([400, 1000) $\text{km~s}^{-1}$) respectively. We note that the Large Magellanic Cloud (LMC) and the Globular Cluster NGC 3201 both produce a peak visible on top of the distribution of Milky Way stars.