Nominal thresholds for good astrometric fits, and prospects for binary detectability, for the full extended \textit{Gaia} mission

TL;DR

This study tackles how Gaia's extended $10$-year baseline improves identification of unresolved astrometric binaries through RUWE (Unit Weight Error) deviations. Using simulated binaries from the Gaia Universe Model Snapshot and the astromet pipeline, the authors derive nominal RUWE thresholds for DR4 and DR5, showing limits of $\mathrm{RUWE}_{\mathrm{DR4,lim}}=1.15$ and $\mathrm{RUWE}_{\mathrm{DR5,lim}}=1.11$, and quantify detectability as a function of orbital period and other binary properties. They find that longer baselines widen the detectable period range (roughly days to up to $\sim$100 years, with highly eccentric systems potentially detectable at longer times) and increase the number of detectable binaries by about 10–20% per data-release, while short-period detections rise by ~5–10%. An analytic comparison using a renormalised $\chi$-distribution provides a lower-bound, per-$N_{\mathrm{obs}}$ perspective on RUWE thresholds and highlights the role of sampling, suggesting refinements where thresholds could depend on $N_{\mathrm{obs}}$. These results offer practical guidance for binary discovery in Gaia data releases and illuminate the relationship between RUWE spread and observation statistics.

Abstract

The full extended Gaia mission spans slightly over 10 years of data, whilst the current data releases represent only a fraction of that timescale (DR3, 34 months). The longer baseline improves the quality of astrometric fits, lowering the noise floor and making consistently bad fits (for example, due to binarity) more apparent. In this paper, we use simulated binaries from the Gaia Universe Model to examine the long-term astrometric behaviour of single stars and stellar binaries. We calculate nominal upper limits on the spread of goodness of astrometric fits for well-behaved single stars. Specifically, for the RUWE parameter, for upcoming DR4 ($\rm RUWE_{lim}=1.15$) and DR5 ($\rm RUWE_{lim}=1.11$), using the full mission nominal scanning law. These can be used to identify poor astrometric fits, and in particular can flag potential binary systems. We show the increase in the number and type of binaries detectable through RUWE. With our updated RUWE thresholds, the number of detectable low-period binaries increases by 5-10% with each subsequent data release, suggesting detections may be possible for orbital periods down to days. The number of detectable long-period systems increases by 10-20%, with periods up to 100 years, causing significant deviations in low moderate-eccentricity binaries. Very eccentric systems with much longer periods (thousands of years) can still be detected if they pass through periapse during the observing window. Finally, we compare our results to the analytic estimate for the spread in UWE, which we predict from a $χ$-distribution moderated by the number of observations. These agree with our inferred population limits but suggest that we may be biased by a small number of poorly sampled systems. In regions of the sky that are more frequently observed, lower limits could be employed, potentially bringing even more binaries above the threshold for detectability.

Figures (8)

• Figure 1: Left: The distribution of the total number of observations per system, $N_\mathrm{obs}$, in each data release for GUMS sources within 200 pc. Right: the corresponding distribution of the number of visibility periods, $N_\mathrm{vis}$, (number of transits separated by more than X hours). Both histograms show a bimodality, mainly due to the ecliptic pole regions ($l<-45$° and $l>45$°) being scanned with more visibility periods Lind2018.
• Figure 2: Example astrometric tracks and fits for seven different binaries from GUMS, of increasing period. Simulated data and fits are shown for DR3 (green), DR4 (orange), and DR5 (red). The coloured curves trace the true photocentre motion, points mark the times of observations (9 CCD observations distributed along the scan direction by the scanning law), and black lines represent ensembles of single-body tracks drawn from the distribution of fitted parameters. Triangular markers indicate the first Gaia observation of each system. The rightmost column reports the true period and parallax of the fit, along with the fitted UWE and parallax.
• Figure 3: The distribution of simulated $\rm UWE$ for single (dashed) and binary (solid) stars, as shown for each data release.
• Figure 4: The fraction of systems below a given $\rm UWE$ is shown for single stars (dashed lines). We fit each with a Student's-$t$ distribution (dotted line) to determine the approximate upper limit on $\rm UWE$ (vertical solid lines), corresponding to a threshold that would exclude fewer than one in a million well-behaved single stars.
• Figure 5: The fraction of detectable binaries as a function of period. A system is considered detectable when $\mathrm{UWE_{DRN}} > \mathrm{UWE_{DRN,limit}}$ (solid lines). The same set of curves has also been plotted using a fixed threshold of 1.4 (dashed lines). Vertical lines show the observational baseline corresponding to each data release.