Using Statistical Methods to Identify Hidden Visual Binaries in Gaia DR3

In the Gaia era, resolved binaries have never been easier to detect via. common parallaxes/proper motions. However, Gaia astrometry is not always available for both components at < 2.5", meaning a significant number of nearby binaries are hiding in the Gaia dataset. To this end, we developed a statistical method to identify likely visual binaries that doesn’t rely on astrometry. This method utilizes differing PSF sizes, where at < 2.5" two stars may be unresolved in 2MASS but resolved by Gaia. Here, the unresolved 2MASS source associated with the resolved Gaia sources has a predictable excess in the J-band that depends on the ∆G from Gaia. This relationship between J- band excess and ∆G differs for chance alignments, as compared to true binaries, when various other parameters are considered, allowing the chance likelihood of any candidate pair to be quantified. We evaluate this for pairs within 200 pc, resulting in a catalog of 68,725 likely binaries. We then obtained Gemini speckle observations of 16 of these systems. With the Gaia and speckle positions, we assess the likelihood of the systems being true binaries vs. chance alignments based on their apparent motion. We find all 16 systems are true binaries if the total average measurement error is ∼4.3%. This estimate of the error for close separation binaries will be crucial when examining time series data in Gaia DR4 and will facilitate more robust error estimates of mass determinations for these systems.