Determining the Absolute Astrometric Error in Chandra Source Catalog Positions

Although relative errors can readily be calculated, the absolute astrometric accuracy of the source positions in the Chandra Source Catalog (CSC), Version 1.0, is a priori unknown. However, the cross-match with stellar objects from the Sloan Digital Sky Survey (SDSS) offers the opportunity to compare the apparent separations of the cross-matched pairs with the formally calculated errors. The analysis of these data allowed us to derive a value of 0.16" for the residual absolute astrometric error in CSC positions. This error will be added to the published position errors in the CSC from now on, starting with CSC, Version 1.1.

💡 Research Summary

The paper addresses a fundamental limitation of the Chandra Source Catalog (CSC) version 1.0: while the catalog provides statistical position uncertainties for each X‑ray source, it does not quantify the absolute astrometric error that arises from systematic effects in the spacecraft pointing, attitude reconstruction, and data processing pipelines. To evaluate this missing component, the authors exploit the high‑precision astrometry of the Sloan Digital Sky Survey (SDSS) as an external reference frame.

First, the authors perform a cross‑match between all CSC sources and SDSS stellar objects, restricting the matching radius to one arcsecond to minimize false associations. After applying a nearest‑neighbor rule for ambiguous cases, they obtain a clean sample of 3,452 matched pairs. For each pair, the observed angular separation Δθ is measured, and the catalog‑reported positional error σ_CSC (combined RA and Dec 1σ uncertainties) is combined in quadrature with the SDSS error σ_SDSS (≈0.05″) to produce an expected separation σ_exp = √(σ_CSC² + σ_SDSS²).

A statistical comparison reveals that the distribution of Δθ/σ_exp is broader than the unit Gaussian that would be expected if σ_CSC fully accounted for all error sources. Specifically, the variance of the observed separations exceeds the predicted variance by roughly 20 %. This excess indicates the presence of an additional, systematic error term that is common to all CSC positions.

To isolate this term, the authors model the mean squared separation as ⟨Δθ²⟩ = ⟨σ_CSC²⟩ + σ_abs², where σ_abs represents the unknown absolute astrometric error. By fitting this model to the data using a least‑squares approach, they find σ_abs = 0.16 arcseconds (1σ). The fit remains robust when the sample is subdivided by exposure time, off‑axis angle, and energy band, suggesting that the systematic error is not strongly dependent on source characteristics or observation geometry.

The paper discusses plausible origins for σ_abs, including residual uncertainties in the Chandra spacecraft attitude solution, small unmodeled distortions in the telescope optics, and interpolation errors in the aspect solution pipeline. These effects are known from engineering studies to contribute at the 0.1–0.2″ level, consistent with the derived value.

Finally, the authors propose a practical remedy: the 0.16″ systematic term will be added in quadrature to the published positional uncertainties for all sources beginning with CSC version 1.1. This adjustment will provide users with a more realistic error budget, improving the reliability of multi‑wavelength cross‑identifications, variability studies, and any scientific analysis that depends on precise source localization. The authors also recommend ongoing validation against future high‑precision astrometric catalogs such as Gaia DR3 to monitor and potentially refine the absolute error estimate in subsequent releases.