2MASS Northern Observatory Survey Camera

Causes of Offsets Between Read2 and Read1 Sources

Howard McCallon and Chas Beichman have found differences between the Read1 and Read2 source position residuals ~0.1 arcsec when measured with respect to the TYCHO catalog positions at the output of the 2MAPPS processing. The residuals with respect to the TYCHO catalog should be quite small, as the TYCHO catalog is the reference for pegging the frames on the sky. They have in fact found residuals of 0.07 arcsec and more, with Read1 and Read2 differing by ~0.1 arcsec as detailed by Howard.

Two contributing factors have been identified which combine to result in this offset.

Read1 and Read2 sources are extracted from the individual frames by FREXAS, and these extractions are used by the POSFRM subsystem to determine offsets between the frames, and the offsets, scale factors, and twist angles between the bands, and peg the frames to the sky using the TYCHO catalog. The Read1 sources are confirmed and combined in POSFRM, and the geometric information is passed in the fpos files to PIXPHOT where the Read2 sources are detected and measured by the PROPHOT sub-task using a table of psfs as a function of seeing and band. POSFRM removes any offsets between the reads and bands using the FREXAS extractions, but position offsets introduced by differences between the FREXAS and PROPHOT extraction processes are not specifically removed before the Read1 and Read2 are combined by MAPCOR. Photometric offsets between Read1 and Read2 are removed in MAPCOR, but not postion offsets.

In early 1998, new psfs were generated for FREXAS and PROPHOT in preparation for the start of production processing. An effort was made to reference the FREXAS psfs with respect to the source centroids. At that time, differences were noted between the FREXAS and PROPHOT positions in band scan coordinates, however, the PROPHOT psfs were not yet complete, so the difference was not resolved.

In order to trace the cause of the current R2 to R1 position offsets, the offsets between the centroid and psf positions output by FREXAS were again investigated, and the offsets between the confirmed FREXAS Read2 extractions mapped to band-scan coordinates (using the POSFRM fpos files) and the Read2 positions output by PROPHOT were also investigated. Small residual offsets (~0.01-0.02 pixel) were noted in all three bands and both axes between the FREXAS centroid and psf positions for both hemispheres except for the Y coordinate in band Ks in the northern camera, where a residual offset was found ~0.06 pixel = 0.12 arcsec. This Ks offset could result in an offset ~0.04 arcsec for a 3 band source or the full 0.12 arcsec for a Ks only source, but not ~0.1 arcsec for all sources.

The following plots show the results for the 2MASS Sampler night (971116n), a night when offsets >0.1 arcsec were found between the Read1 and Read2 sources:

Average offsets between FREXAS psf and centroid source positions for 971116n scans

Average offsets between FREXAS and PROPHOT psf source positions for 971116n scans

In the first plot, a consistent offset ~0.06 pixel may be seen between the FREXAS psf and centroid positions on the Y axis in Ks band. In the second plot, variable offsets of up to 0.16 pixel = 0.32 arcsec may be seen between the FREXAS and PROPHOT postions, which could account for 0.1 arcsec offsets for 3 band sources. Variable offsets as a function of PSFID have been previously noted in band-to-band comparisions of the PROPHOT output sources. This plot shows the offset in the y axis of K band toggling with the PROPHOT psf as a function of scan length for 971116n scan 38.. These offsets may also be seen in the statistics file output by BANDMERGE, with a superimposed cross-scan slope of amplitude ~0.15 arcsec between the Ks band and the J and H band merged extractions see 971116n.s011.bmgstats for example.

Summarizing, the offsets seen between R2 and R1 in the TYCHO residuals are found to be partially due to an error ~0.12 arcsec in the FREXAS psf centering on the Y axis of Ks band, and probably a similar error in the centering of one or more PROPHOT psfs of ~0.2 arcsec.

2MAPPS V3 testing on 980403n shows offsets between FREXAS and PROPHOT source postions in band scan coordinates:

Average offsets between FREXAS psf and centroid source positions for 980403n scans

Average offsets between FREXAS and PROPHOT psf source positions for 980403n scans

Since there is very little offset between the FREXAS psf and centroid positions,the offsets in cross-scan between FREXAS and PROPHOT are probably due to PROPHOT psfs. Since POSPTS shifts the source positions to remove any residual offset in the PROPHOT positions, any descrepancy between FREXAS and PROPHOT will show up as an offset in the R1 source positions.

2MAPPS V3 testing on 971116n shows similar results, although the offset is smaller in H band:

Average offsets between FREXAS psf and centroid source positions for 971116n scans

Average offsets between FREXAS and PROPHOT psf source positions for 971116n scans