Skip to the navigation links
Last modified: 22 November 2017


213th Meeting of the American Astronomical Society

4 January 2009 - 8 January 2009
Long Beach, CA

Chandra X-ray Observatory Source Catalog

Poster Session: 7 January 2009

[poster thumbnail] The Chandra Source Catalog (PDF) - Ian N. Evans

[poster thumbnail] The Chandra Source Catalog: Source Properties and Data Products (PDF) - Arnold H. Rots

[poster thumbnail] The Chandra Source Catalog: Algorithms (PDF) - Jonathan C. McDowell

[poster           thumbnail] The Chandra Source Catalog: Background Determination and Source Detection (PDF) - Michael L. McCollough

[poster thumbnail] The Chandra Source Catalog: Source Variability (PDF) - Michael A. Nowak

[poster thumbnail] The Chandra Source Catalog: Spectral Properties (PDF) - Stephen Doe

[poster thumbnail] The Chandra Source Catalog: Statistical Characterization (PDF) - Francis A. Primini

[poster thumbnail] The Chandra Source Catalog: X-ray Aperture Photometry (PDF) - Vinay L. Kashyap

[poster thumbnail] The Chandra Source Catalog: User Interface (PDF) - Nina R. Bonaventura

[poster thumbnail] The Chandra Source Catalog: Processing and Infrastructure (PDF) - Janet D. Evans

[poster thumbnail] The Chandra Source Catalog: Automated Source Correlation (PDF) - Roger Hain

[CSC handout           thumbnail] CSC handout (front) [CSC handout           thumbnail] (back)


The Chandra Source Catalog (CSC) is ultimately intended to be the definitive catalog of X-ray sources detected by the Chandra X-ray Observatory (CXO). When compared to all previous and current X-ray missions, Chandra breaks the resolution barrier with an arcsecond scale on-axis point spread function. The combination of excellent spatial resolution, a reasonable field of view, and low instrumental background translate into a high detectable-source density, with low confusion and good astrometry. The wealth of information that can be extracted from identified serendipitous sources is a powerful and valuable resource for astronomy.

The aim of the CSC is to disseminate this wealth of information by characterizing the X-ray sky as seen by Chandra. The CSC provides simple access to Chandra data for individual sources or sets of sources matching user-specified search criteria. The catalog is intended to satisfy the needs of a broad- based group of scientists, including those who may be less familiar with astronomical data analysis in the X-ray regime. For each detected X-ray source, the catalog lists the source position and a detailed set of source properties, including multi-band aperture fluxes, X-ray colors and hardness ratios, spectra, temporal variability information, and source extent estimates. In addition to these traditional elements, the catalog includes file-based data products that can be manipulated interactively, including images, photon event lists, light curves, and spectra for each source individually from each observation in which a source is detected.

The first release of the CSC includes information for ~100,000 X-ray sources detected in a subset of public imaging observations from the first eight years of the Chandra mission. Only point sources, and compact sources with extents <~30 arcsec, are included. Highly extended sources, and sources located in selected fields containing bright, highly extended sources, are excluded from this release.

The Chandra Source Catalog (CSC) is breaking new ground in several areas. In particular, there are two aspects of interest to the users: its evolution and its contents.

  • The CSC will be a living catalog that becomes richer, bigger, and better in time while still remembering its state at each point in time. This means that users will be able to take full advantage of new additions to the catalog, but still able to back-track and go back to what was extracted in the past.
  • The CSC also sheds the limitations of flat-table catalogs. Its sources will be characterized by a large number of properties, as usual, but each source will also be associated with its own specific data products, allowing users to perform mini custom analysis on the sources.

Source properties fall in the spatial (position, extent), photometric (fluxes, count rates), spectral (hardness ratios, standard spectral fits), and temporal (variability probabilities) domains, and are all accompanied by error estimates. Data products cover the same coordinate space and include events lists, images, spectra, and light curves. In addition, the catalog contains data products covering complete observations: event lists, background images, exposure maps, etc.

Creating the Chandra Source Catalog (Evans et al.) required adjustment of existing pipeline processing, adaptation of existing interactive analysis software for automated use, and development of entirely new algorithms. Data calibration was based on the existing pipeline, but more rigorous data cleaning was applied and the latest calibration data products were used. For source detection, a local background map was created including the effects of ACIS source readout streaks. The existing wavelet source detection algorithm was modified and a set of post-processing scripts used to correct the results. To analyse the source properties we ran the SAOTrace ray trace code for each source to generate a model point spread function, allowing us to find encircled energy correction factors and estimate source extent. Further algorithms were developed to characterize the spectral, spatial and temporal properties of the sources and to estimate the confidence intervals on count rates and fluxes. Finally, sources detected in multiple observations were matched, and best estimates of their merged properties derived. In this paper we present an overview of the algorithms used. More detailed treatment of some of the newly developed algorithms are presented in companion posters. For details of the software and processing, see J. Evans et al.; for a description of the user interface which allows access to the catalog, see Bonaventura et al.

The Chandra Source Catalog (CSC) is a major project in which all of the pointed imaging observations taken by the Chandra X-Ray Observatory will be used to generate one of the most extensive X-ray source catalog produced to date. Early in the development of the CSC it was recognized that the ability to estimate local background levels in an automated fashion would be critical for essential CSC tasks such as source detection, photometry, sensitivity estimates, and source characterization. We present a discussion of how such background maps are created directly from the Chandra data and how they are used in source detection.

The general background for Chandra observations is rather smoothly varying, containing only low spatial frequency components. However, in the case of ACIS data, a high spatial frequency component is added that is due to the readout streaks of the CCD chips. We discuss how these components can be estimated reliably using the Chandra data and what limitations and caveats should be considered in their use.

We will discuss the source detection algorithm used for the CSC and the effects of the background images on the detection results. We will also touch on some the Catalog Inclusion and Quality Assurance criteria applied to the source detection results.

The Chandra Source Catalog (CSC) contains fields of view that have been studied with individual, uninterrupted observations that span integration times ranging from 1 ksec to 160 ksec, and a large number of which have received (multiple) repeat observations days to years later. The CSC thus offers an unprecedented look at the variability of the X-ray sky over a broad range of time scales, and across a wide diversity of variable X-ray sources: stars in the local galactic neighborhood, galactic and extragalactic X-ray binaries, Active Galactic Nuclei, etc. Here we describe the methods used to identify and quantify source variability within a single observation, and the methods used to assess the variability of a source when detected in multiple, individual observations. Three tests are used to detect source variability within a single observation: the Kolmogorov-Smirnov test and its variant, the Kuiper test, and a Bayesian approach originally suggested by Gregory and Loredo. The latter test not only provides an indicator of variability, but is also used to create a best estimate of the variable lightcurve shape. We assess the performance of these tests via simulation of statistically stationary, variable processes with arbitrary input power spectral densities (here we concentrate on results of red noise simulations) at variety of mean count rates and fractional root mean square variabilities relevant to CSC sources. We also assess the false positive rate via simulations of constant sources whose sole source of fluctuation is Poisson noise. We compare these simulations to a preliminary assessment of the variability found in real CSC sources, and estimate the variability sensitivities of the CSC.

The first release of the Chandra Source Catalog (CSC) will contain sources identified from eight years' worth of publicly accessible observations. The vast majority of these sources have been observed with the ACIS detector and have spectral information in 0.5-7 keV energy range. Here we describe the methods used to automatically derive spectral properties for each source detected by the standard processing pipeline and included in the final CSC. Hardness ratios were calculated for each source between pairs of energy bands (soft, medium and hard) using a Bayesian algorithm (BEHR, Park et al. 2006). The sources with high signal to noise ratio (exceeding 150 net counts) were fitted in Sherpa (the modeling and fitting application from the Chandra Interactive Analysis of Observations package, developed by the Chandra X-ray Center; see Freeman et al. 2001). Two models were fitted to each source: an absorbed power-law and a blackbody emission. The fitted parameter values for the power-law and blackbody models were included in the catalog with the calculated flux for each model. The CSC also provides the source energy flux computed from the normalizations of predefined power-law and blackbody models needed to match the observed net X-ray counts. In addition, we provide access to data products for each source: a file with source spectrum, the background spectrum, and the spectral response of the detector.

The Chandra Source Catalog (CSC) will ultimately contain more than ~250000 x-ray sources in a total area of ~1% of the entire sky, using data from ~10000 separate ACIS and HRC observations of a multitude of different types of x-ray sources (see Evans et al.). In order to maximize the scientific benefit of such a large, heterogeneous dataset, careful characterization of the statistical properties of the catalog, i.e., completeness, sensitivity, false source rate, and accuracy of source properties, is required. Characterization efforts of other, large Chandra catalogs, such as the ChaMP Point Source Catalog (Kim et al. 2007, 2007ApJS..169..401K) or the 2 Mega-second Deep Field Surveys (Alexander et al. 2003, 2003AJ....126..539A), while informative, cannot serve this purpose, since the CSC analysis procedures are significantly different and the range of allowable data much less restrictive. We describe here the characterization process for the CSC. Our plans include both a comparison of real CSC results with those of other, deeper Chandra catalogs of the same targets as well as extensive simulations of blank-sky and point source datasets. We present preliminary results from our work to date.

The Chandra Source Catalog represents a reanalysis of the entire ACIS and HRC imaging observations over the 9-year Chandra mission. Source detection is carried out on a uniform basis, using the CIAO tool wavdetect, and source fluxes are estimated post-facto using a Bayesian method that accounts for background, spatial resolution effects, and contamination from nearby sources. We use gamma-function prior distributions, which could be either non-informative, or in case there exist previous observations of the same source, strongly informative. The resulting posterior probability density functions allow us to report the flux and a robust credible range on it. We also determine limiting sensitivities at arbitrary locations in the field using the same formulation.

The Chandra Source Catalog (CSC) is ultimately intended to be the definitive catalog of all X-ray sources detected by Chandra. The CSC is presented to the user in two tables: the Master Chandra Source Table and the Table of Individual Source Observations. Each distinct X-ray source identified in the CSC is represented by a single master source entry and one or more individual source entries. If a source is unaffected by confusion and pile-up in multiple observations, the individual source observations are merged to produce a master source. In each table, a row represents a source, and each column a quantity that is officially part of the catalog.

The CSC contains positions and multi-band fluxes for the sources, as well as derived spatial, spectral, and temporal source properties. The CSC also includes associated source region and full-field data products for each source, including images, photon event lists, light curves, and spectra. The master source properties represent the best estimates of the properties of a source, and are presented in the following categories: Position and Position Errors, Source Flags, Source Extent and Errors, Source Fluxes, Source Significance, Spectral Properties, and Source Variability.

CSCview is a GUI which provides direct access to the source properties and data products contained in the catalog. The user may query the catalog database via a web-style search or an SQL command-line query. Each query returns a table of source properties, along with the option to browse and download associated data products. CSCview is designed to run in a web browser with Java version 1.5 or higher, and may be accessed via a link on the CSC website homepage ( As an alternative to the GUI, the contents of the CSC may be non-interactively accessed through a URL, using a command-line tool such as cURL.

Chandra Source Catalog processing recalibrates each observation using the latest available calibration data, and employs a wavelet-based source detection algorithm to identify all the X-ray sources in the field of view. Source properties are then extracted from each detected source that is a candidate for inclusion in the catalog. Catalog processing is completed by matching sources across multiple observations, merging common detections, and applying quality assurance checks.

The Chandra Source Catalog processing system shares a common processing infrastructure and utilizes much of the functionality that is built into the Standard Data Processing (SDP) pipeline system that provides calibrated Chandra data to end-users. Other key components of the catalog processing system have been assembled from the portable CIAO data analysis package. Minimal new software tool development has been required to support the science algorithms needed for catalog production. Since processing pipelines must be instantiated for each detected source, the number of pipelines that are run during catalog construction is a factor of order 100 times larger than for SDP. The increased computational load, and inherent parallel nature of the processing, is handled by distributing the workload across a multi-node Beowulf cluster. Modifications to the SDP automated processing application to support catalog processing, and extensions to Chandra Data Archive software to ingest and retrieve catalog products, complete the upgrades to the infrastructure to support catalog processing. The end product is a catalog of Chandra sources, associated catalog user interfaces, and forthcoming data analysis tools, that will allow users to query the catalog, retrieve relevant data, and perform interactive scientific analysis on those results.

Chandra Source Catalog (CSC) master source pipeline processing seeks to automatically detect sources and compute their properties. Since Chandra is a pointed mission and not a sky survey, different sky regions are observed for a different number of times at varying orientations, resolutions, and other heterogeneous conditions. While this provides an opportunity to collect data from a potentially large number of observing passes, it also creates challenges in determining the best way to combine different detection results for the most accurate characterization of the detected sources.

The CSC master source pipeline correlates data from multiple observations by updating existing cataloged source information with new data from the same sky region as they become available. This process sometimes leads to relatively straightforward conclusions, such as when single sources from two observations are similar in size and position. Other observation results require more logic to combine, such as one observation finding a single, large source and another identifying multiple, smaller sources at the same position.

We present examples of different overlapping source detections processed in the current version of the CSC master source pipeline. We explain how they are resolved into entries in the master source database, and examine the challenges of computing source properties for the same source detected multiple times. Future enhancements are also discussed.