Probing higher resolution: an asymmetry in the Chandra PSF
Posted: October 2010
As part of efforts to push the spatial resolution of Chandra to the sub-ACIS-pixel regime, we have identified a feature in the Chandra/HRC point spread function within the central arcsecond which may affect high fidelity deconvolutions. There is evidence that the feature is also present in Chandra/ACIS data. The problem does not affect images on scales larger than one arcsecond.
- Chandra's Ultimate Angular Resolution: Studies of the HRC-I Point Spread Function by Michael Juda & Margarita Karovska.
- Analysis of Chandra PSF feature using ACIS data by Vinay Kashyap
What do we know?
The feature's brightness appears to be ~5% (and perhaps somewhat higher) of the total brightness.
The asymmetry (a hook-like structure) seems to be fixed in the spacecraft coordinate and extends roughly between mirror spherical coordinate azimuthal angle, PHI, of 260-310 deg. PHI is measured counterclockwise from the +DETX (Spacecraft +Y) axis; this range corresponds to a position angle of 170-220 deg measured counterclockwise from the +DETY (Spacecraft +Z) axis. (The difference between Fig 1, Fig 2 and the Fig 3 is that the first two are in the HRC instrumental U-V coordinates, and Fig 3 is in spacecraft coordinates.)
Residuals in 2D fits to ACIS data persistently show enhancements relative to the PSF in the +Z direction that are consistent with the PSF hook-like feature.
The asymmetry was discovered using on-axis AR Lac observations (Fig 1, Fig 3), and was confirmed using HRC observations of Capella (Fig 2) and other point sources. Techniques used include HRMA PSF fitting and analysis of residuals; deconvolution using simulated HRMA (ChaRT/SAOTrace PSFS); PSF subtraction (e.g. Fig 1 and Fig 2, Fig 3)
A time series of AR Lac observations obtained almost yearly since October 1999 shows that the asymmetry may have appeared sometime between 2000 Dec 12 and 2002 Jan 27 (Fig 1), and is present in the recent observations obtained in 2009 Sep 24.
What do we suspect?
Analysis of several ACIS images (SER, deconvolution, subtraction) indicate that there is an asymmetry in a similar direction and spatial scale as in the HRC images (Fig 4). The figure of panda counts ratios from Kashyap shows the asymmetry in the counts in the data along +Z compared to counts along -Z.
The analysis of the ACIS data is more complex because of its coarser pixel scale and thus coarser sampling of the mirror+instrument PSF. If the asymmetry is indeed in the ACIS data as well then it would suggest that the origin of this asymmetry is in the HRMA mirrors. The CXC is continuing its investigation.
Advice to users
The make_psf_asymmetry_region tool creates a region file that identifies the area where the asymmetry may be affecting the analysis (especially for a 2-D analysis looking for faint jets, shells, or similar structures). To use make_psf_asymmetry_region, download the CIAO contributed tarfile (release 26 Oct 2010 or later)
Note that although the feature is fixed in detector coordinates, it is not possible to inspect it by binning on DETX, DETY since its offset from the PSF core is completely swamped by the dither. To look for evidence of the feature, make a sky coordinate image using small pixels, e.g. with the binning
The position angle (PA) of a feature on the sky is measured from N through E in the standard astronomical convention; so a feature towards SKY +Y has PA = 0, and towards SKY +X has PA = 270 deg. The PA is related to the detector mirror spherical coordinate PHI by
PA = PHI - ROLL - 90 deg
and so the PSF artifact, which is at 285 +/- 25 deg in PHI, is in the sky direction
PA = 195 - ROLL (+/- 25) deg
where the ROLL can be found by
dmkeypar evt2.fits ROLL_NOM echo+
Users should treat with suspicion any extent in their images at scales less than 1 arcsecond in a direction of 260-310 deg (in spacecraft MSC coordinates) with a flux less than 10 percent of the total brightness of the source. Two observations at different spacecraft rolls will be advisable for searches for shells or jets at scales of less than 1".
Simulations suggest that encircled energy analyses would show detectable effects of this asymmetry for sources with several hundred counts or more.
What is being done?
The CXC is carrying out analysis of point source observations with HRC and ACIS to characterize the asymmetry and its origin and temporal evolution.
Simulations with a distorted HRMA are being carried out in an attempt to replicate and model the asymmetry.
Additional 20 ks observations of AR Lac have been scheduled in Chandra Cycle 13 for calibration purposes.
In the longer term, if the asymmetry is confirmed to be in the HRMA, the HRMA PSF model will be improved and made available to the users via ChaRT/SAOTrace.