ACIS Sub-Pixel Resolution: Improvement in Point Source Detection

Le résumé fourni par la source

We investigate how to achieve the best possible ACIS spatial resolution by binning in ACIS sub-pixel and applying an event repositioning algorithm after removing pixel-randomization from the pipeline data. We quantitatively assess the improvement in spatial resolution by (1) measuring point source sizes and (2) detecting faint point sources. The size of a bright (but no pile-up), on-axis point source can be reduced by 20–30%. With the improved resolution, we detect ∼ 20% more faint sources when embedded in the extended, diffuse emission in a crowded field. We further discuss the false source rate of ∼ 10% among the newly detected sources, using a few ultra-deep observations. 1. How to Obtain the Best Possible ACIS Resolution In order to achieve the best possible ACIS resolution, we apply the following steps to the Chandra X-ray Center (CXC) pipeline products. Then we quantitatively assess the improvement by comparing the point source sizes (§ 2) and detections (§ 3) before and after the procedure. Sub-pixel binning: Chandra coordinates contain positional accuracy finer than one ACIS pixel (0.492 arcsec) through dither and aspect correction. Imaging data binned by ACIS subpixel can already provide an improved resolution. Remove pixel randomization: The current pipeline default is to apply pixel randomization by 1/2 ACIS pixel on the chip coordinate to remove the instrumental “gridded” appearance of the data and to avoid any possible aliasing affects associated with this spatial grid. This pixel randomization has to be removed before applying a sub-pixel algorithm. ACIS sub-pixel algorithm: Positional accuracy can be improved by utilizing the positional information in 3 × 3 event islands. Several sub-pixel event repositioning algorithms have been developed during the first years of the Chandra mission. The first implementation by Tsunemi et al. (2001) applied the knowledge of charge cloud size in 3×3 event islands to corner events with ASCA grade 6 (4–16% of on-axis events). Mori et al. (2001) extended the algorithm to all split pixel events: event grades 2, 3, 4 are shifted 1/2 pixel in one direction, grade 6 events are shifted 1/2 pixel in two directions, while grade 0 events remain centered on the event island. Later Li et al. (2003, 2004) improved the algorithm to SER (subpixel event repositioning) and further to EDSER (energy dependent SER). CXC also implemented EDSER in CIAO 4.3. PSF deconvolution: We do not discuss deconvolution here. We refer the reader to CXC Announcement #64. http://space.mit.edu/CXC/docs/docs.html#subpix http://cxc.harvard.edu/ciao/releasenotes/ciao_4.3_release.html http://cxc.harvard.edu/announcements/announce_64.html

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.