Last modified: 21 February 2024

URL: https://cxc.cfa.harvard.edu/csc/proc/master.html

After the Stacker pipeline completes for all detected sources in a stack, the Master Match pipeline compares that stack to others which overlap the same region of sky to match sources identified in multiple stacks. The resulting overall list of 'master sources' is considered to be a list of individual astronomical X-ray sources.

Stacks contain observations whose pointing direction is the same to within 1 arcminute. Since the field of view of a single ACIS chip is 8 arcminutes, there are many observations which significantly overlap but are not in the same stack. As a consequence, a given source may appear in these observations with very different PSF sizes, which complicates the cross-stack identification process.

Match sources from different stacks

The Master Match pipeline cross-matches each stack detection with all existing stack detections in the catalog database, and either inserts the derived source properties into the database in the case of a newly identified source, or merges the newly measured source properties with the existing catalog data in the case of an already cataloged source.

Source matches may be classified in several different ways:

  • New: the source is newly identified and does not overlap any other stacks, and so does not need to be merged.

  • Simple: a detection in one stack overlap at most one detection in another stack.

    [Thumbnail image: A simple merge case, where sources (shown by an enhancement in the number of white points on a black background) are surrounded by two (blue and green) ellipses representing the detections, where each green/blue pair only overlap each other.]

    [Version: full-size]

    [Print media version: A simple merge case, where sources (shown by an enhancement in the number of white points on a black background) are surrounded by two (blue and green) ellipses representing the detections, where each green/blue pair only overlap each other.]
    A simple merge case.
  • Confused: a single source in one stack overlaps more than one source in another stack.

    [Thumbnail image: A large green ellipse overlaps four small blue ellipses.]

    [Version: full-size]

    [Print media version: A large green ellipse overlaps four small blue ellipses.]
    A confused source merge case.
  • Too hard: everything that does not fall into one of the other categories (e.g. "chained" sources). These matches were sent to a manual quality assurance process to find a resolution.

    [Thumbnail image: There appear to be two sources (from the pixel data) but there are multiple ellipses (blue, greem, and yellow) that overlap.]

    [Version: full-size]

    [Print media version: There appear to be two sources (from the pixel data) but there are multiple ellipses (blue, greem, and yellow) that overlap.]
    A "too hard" source merge case.

Link sources

At this stage, the master sources are assigned their final 2CXO catalog designations.

All sources within a given observation are included in the Catalog as a unit, i.e. the full-field observations including those sources are linked together under each source name in the Master Catalog. This allows for straightforward updating and the ability to roll-back a single source if a problem is found.

Bi-directional links within the database record the connections between the Master Catalog entries and associated individual stack detection entries, so that users can view all observations of a single source and how they contribute to the master entry.

Because the Chandra point spread function (PSF) is a strong function of off-axis angle, detections far off-axis in one observation stack may overlap multiple resolved sources detected on-axis in another observation stack. Such detections are flagged as "confused." The individual source entries are linked to the Master Catalog entries for each overlapped source, but the data are not used to compute the master source properties.

Post-pipeline Tasks

After this pipeline runs, sources are flagged for removal if the significance is lower than the threshold. When a source is removed, it is not included in the database.

Sources may be detected in one stack but not in another stack in the same ensemble. The undetected source in this case is called a 'no-detect'. We want to know, for example, what the upper limit is to the flux of this source in the stack it wasn't seen in. For these no-detect sources the pipeline creates per-observation point spread function (PSF) images and per-stack background regions, which allow upper limit and variability calculations to be performed for these source/observation combinations later on, in the Source Properties pipeline.