Date: Thu, 21 Dec 2000 17:07:31 -0500 (EST) From: "Eric D. Feigelson" Subject: Optimal point source extraction in ACIS To: karovska@head-cfa.harvard.edu, lvs@cfa.harvard.edu, elvis@cfa.hravard.edu, afruscione@cfa.harvard.edu, kashyap@head-cfa.harvard.edu, acis_psu, mwb@space.mit.edu Mime-Version: 1.0 To: Margarita Karovska, Martin Elvis, Antonella Fruscione (CXC SDS group) Leon Van Speybroeck (CXC Telescope Sci) Vinay Kashyap (wavdetect group) CXC Helpdesk acis_psu, Mark Bautz From: Eric Feigelson, Pat Broos & Jim Gaffney, ACIS Team, Penn State Re: Optimal extraction radius for point sources in ACIS We would like to report our work on this issue, which has raised over the past few months by us (see Helpdesk tickets 2403 & 3682). The question is: what is the optimal extraction radius as a function of off-axis distance and source S/N for point sources? This matter should be of interest to a great number of Chandra observers, and it is appropriate that CXC provide easily accessible information or guidance for the Chandra scientist. Here is a brief history of our effort. First, we thought that the quantity PSF_SIZE given in wavdetect srclist.fits files, generated from a program called psf_calc_size, would do a good job. But examination showed that it produced extraction circles too large far off-axis. Second, we computed 90%/95%/99% enclosed energy radii vs. off-axis distance using mkpsf and the CIAO 1 point spread function database. A monochromatic 1.49 keV source is assumed. The result seemed to match source sizes better than PSF_SIZE, but there was a disturbingly sudden change in PSF characteristics at 6' off-axis. The attachment extraction_ciao1.ps summarizes these efforts. Our third, and final step was to compute 90%/95%/99% curves using the denser CIAO 2 PSF database (extraction_ciao2.ps). The results are improved -- the jump found earlier is attributable to inaccurate interpolation in CIAO 1's crude PSF grid. These curves complement the 85% enclosed energy radii vs. off-axis distance vs. energy curves recently produced by Margarita in her draft document "Chandra PSF Library". So we can now turn to the observer's question: What extraction radius should I use? It depends on the relative strength of the source and background. Qualitatively, one should choose a radius such that the incremental increase in background events is just outweighed by the increase in source events as one increases the radius. For very strong sources, it is probably always best to choose a large extraction radius such as 99%. This will give the most counts for spectral and variability study and, in particular, will capture the high energy photons that preferentially populate the PSF wings. For very weak sources off-axis sources, then a conservative radius like 90% might be chosen. This will maintain a good S/N ratio in the extracted events. Of course, one must avoid overlapping extraction radii in crowded fields, and must record which curve is used for which source so that the flux can be correctly scaled to 100%. For several forthcoming studies, we are choosing for simplicity just two options: the 95% curve for sources <1000 counts, and the 99% curve for sources >1000 counts. We would appreciate hearing from anyone who establishes more quantitative rules. Many thanks to Margarita Karovska and Vinay Kashyap for discussion and resources. We would appreciate hearing from others on this matter.