Both Northern and Southern Survey Cameras have optical distortions which cause the psf to vary across the focal plane. This psf variation across the focal plane is also a function of temperature and camera focus. The 2MAPPS software assumes the psf is constant across the focal plane (isoplanatic) and varies only as a function of atmospheric seeing. The determination of the psf variation due to seeing is sometimes problematic due to a lack of sufficient high snr point sources, the spectre of two additional dimensions to the problem (focal plane coordinates) motivated a considerable effort to improve the optics such that the psf was isoplanatic. Although the resulting optics are quite good, enough non-isoplanicity exists to cause errors of several percent in the relative psf photometry across the focal plane. The in-scan variations are averaged out by the in-scan stepping providing 6 samples spaced across the focal plane. The effect seems to be a few percent in the north since the fall of 1997, but up to 10% for the southern camera.
The residual cross-scan psf photometric error may be seen in a plot of the psf minus aperture photometry as a function of cross-scan position for the J optics in the Southern Survey camera:
980801s/j044 PSF-aperture photometry vs x-scan
A software fix has been developed which corrects most of the cross-scan photmetric bias most of the time. Problems still exist, however, for calibration scan blocks in low density areas.
The cross-scan offset is determined for each calibration or survey block of scans by computing a robust estimate of the psf-aperture magnitude offsets at ten positions across the focal plane for each band. All the unconfused sources which are not flagged as artifacts or galaxies and have relatively good signal-to-noise ( 9< J < 14.5, 8.5< H < 14.0, 8< K < 13.5), are binned in cross-scan (u-scan), and the mid-average offset is computed and output, as well as the estimated error, and the counts per bin. To avoid a bias in areas of high source density, the threshold is increased by -2.5*log( sources/deg^2/10^4). This output file is read by the calibration subsystem, which applies a correction to each source based on the offsets linearly interpolated to the exact cross- scan coordinate of the source. The results of this process for the above scan are shown in the following plot:
980801s/j044 Corrected PSF-aperture photometry vs x-scan
Problems still exist for calibration scan blocks in low-density areas (high galactic lattitude). For calibration blocks in low-density areas, fewer than 20 sources per cross-scan bin of sufficient quality may be available, resulting in offset determination errors of 3 to 10%. The cross-scan profiles have been seen to change abruptly between blocks, and possibly within blocks, so interpolation from surrounding blocks can be risky. Interpolation between adjacent blocks may still be the best option for blocks with excessive errors in the offset determination. We are in the process of applying the preliminary cross-scan offset determination software to a number of nights of southern survey data to better evaluate the best way to diagnose correction problems and the best course of action when the correction errors are excessive.See Sherry Wheelock's analysis of the cross-scan offset corrections and the correction consistency.
The following plots show the range (max-min) of the computed magnitude correction as a function of the first scan in the block for several full nights of southern data processed so far:
980319s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980321s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980322s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980323s Range of Computed PSF-aperture Photometric Offset vs Scan Block
----> 980323s -note cal scan block 15-20 H & K offset errors ----> this cal block is low density/glat>60
980324s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980325s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980326s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980327s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980328s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980329s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980330s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980331s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980801s Range of Computed PSF-aperture Photometric Offset vs Scan Block
980923s Range of Computed PSF-aperture Photometric Offset vs Scan Block
Of the 45 cal scan blocks in these 5 nights, 18 have expected magnitude correction errors > 0.02 magnitudes.
The psf shape and cross-scan psf magnitude errors are a strong function of the telescope focus. The following plots show the range and the edge difference of the psf-aperture magnitude errors as a function of telescope focus for the 980801s/s080-s085 focus test:
Max-min, cross-scan psf magnitude errors vs focus
Edge difference, cross-scan psf magnitude errors vs focus
In the above plots, J=. , H=+ , K=*
For this focus test (at a tel/air temp ~11.5 C), a focus near 980 would minimize the cross-scan psf magnitude errors.
This page last updated on Oct. 13, 1998.
Gene Kopan - gene at ipac.caltech.edu
