I. Profile-Fit Photometry Normalization

We assume that aperture photometry of isolated, relatively bright (SNR>20) sources provides a "truth" reference for 2MASS point source brightness estimation. Profile-fit photometry is the default point source brightness estimator, though, because it provides better measurements for the more numerous fainter sources, does a better job of measuring sources in crowded environments, and avoids inducing biases due to lost pixels in undersampled 2MASS data.

Profile-fit photometry from PROPHOT is tied to the true photometric scale in 2MAPPS by normalization to curve-of-growth corrected aperture photometry. This normalization occurs in two places during scan processing:

1. MAPCOR
Profile-fit photometry is normalized to aperture photometry in MAPCOR band-by-band by measuring the mean offset between the two for sources in the brightness range 8 < mag < 12 in J, H and K_s. The mean offset is then added to the profile-fit magnitudes for all sources so that the mean offset for the sources in the 8-12 mag brightness range should be zero. This is done separately for all sources in each scan having the same seeing-shape value, so there can be up to several normalizations per band per scan.

2. DMAGCOR

Shortly after the start of southern operations, it was found that there was an apparent bias between profile-fit and aperture photometry that was a function of source cross-scan position. This bias arises because of the variation of PSF shape across the focal plane. PROPHOT employs PSF's that are constructed from ensembles of sources from all focal plane positions, so towards the edges of the arrays, where the distortion is largest, there can be a mismatch between the PROPHOT PSF and actual image profile.
The DMAGCOR subsystem was designed to compensate for this by deriving a second-order PSF-fit/aperture photometric normalization that is a function of cross-scan position. DMAGCOR computes the mean offset between PSF-fit and aperture magnitudes in ten position bins across the focal plane using sources in brightness ranges 9 < J < 14.5, 8.5 < H < 14.0, 8 < K_s < 13.5. The mean offsets are then applied to the profile-fit magnitudes of sources according to their cross-scan position. The statistical accuracy of the measurement is improved by using a wider magnitude range and by computing the correction for blocks of scans.
The DMAGCOR correction was run on all southern data during preliminary data processing, but only on northern data taken after 980605n.

The objective of these two corrections is for there to be as little possible bias between profile-fit and aperture photometry for relatively high signal-to-noise sources in the survey. Neither of these adjustments can compensate for any magnitude-dependent bias between the photometric measurements, though.

II. PSF-fit vs. Aperture Photometry in the RTB Night Data

The following plots show the measured differences between profile-fit and aperture photometry in recent 2MAPPS v3.0 RTB night runs. All sources with |b| > 20^o in each night were used in the evaluations. The rows labelled "v2.x" in the Proc_date column contain the results from preliminary processing (2MAPPS v2.x).

Figures 1-7a,f,k show the photometric offsets as a function of the profile-fit magnitude. The small black points are individual stars and the larger red points are the trimmed average offsets in bins of 0.1 magnitude. The horizontal green lines are drawn at -0.01, 0.0 and +0.01 mag for reference.

Figures 1-7b,g,l show the mean photometric offsets in the 0.1 magnitude bins for each night with the black lines showing the means of all sources, the green points are the mean offsets of sources within 50" of the east edge of the focal plane, and the red points are the mean offsets of sources within 50" of the west edge of the arrays.

Figures 1-7c,h,m show the mean photometric offsets of sources as a function of cross-scan position, in 5.1" cross-scan bins. The blue points show the mean offsets for sources with J < 13 mag, H < 12.5 mag and K_s < 12.0 mag, and the red points show the mean offsets for sources fainter than those limits.

Figures 1-7d,i,n show the reduced chi-squared as a function of the profile-fit magnitude. The small black points are individual stars and the larger red points are the trimmed average chi-squared in bins of 0.1 magnitude. The horizontal green line is drawn at chi-sqared = 1.0 for reference.

Figures 1-7e,j,o show the mean chi-squared values of sources as a function of cross-scan position, in 5.1" cross-scan bins. The blue points show the mean offsets for sources with J < 13 mag, H < 12.5 mag and K_s < 12.0 mag, and the red points show the mean offsets for sources fainter than those limits.

Table 1 - Mean PSF-fit versus Aperture Photometry Differences for RTB nights

Date/hemis	Proc_date	M_psf-M_ap versus Mag	M_psf-M_ap versus Mag Edge Effects	M_psf-M_ap versus Cross-Scan Position	Chi-squared versus Mag	Chi-squared versus Cross-Scan Position
971116n	v2.x	Fig 1a	Fig 1b	Fig 1c	Fig 1d	Fig 1e
	010504	Fig 1f	Fig 1g	Fig 1h	Fig 1i	Fig 1j
	010512	Fig 1k	Fig 1l	Fig 1m	Fig 1n	Fig 1o
980319s	v2.x	Fig 2a	Fig 2b	Fig 2c	Fig 2d	Fig 2e
	010505	Fig 2f	Fig 2g	Fig 2h	Fig 2i	Fig 2j
	010512	Fig 2k	Fig 2l	Fig 2m	Fig 2n	Fig 2o
980403n	v2.x	Fig 3a	Fig 3b	Fig 3c	Fig 3d	Fig 3e
	010505	Fig 3f	Fig 3g	Fig 3h	Fig 3i	Fig 3j
	010514	Fig 3k	Fig 3l	Fig 3m	Fig 3n	Fig 3o
990523n	v2.x	Fig 4a	Fig 4b	Fig 4c	Fig 4d	Fig 4e
	010505	Fig 4f	Fig 4g	Fig 4h	Fig 4i	Fig 4j
	010512	Fig 4k	Fig 4l	Fig 4m	Fig 4n	Fig 4o
990723s	v2.x	Fig 5a	Fig 5b	Fig 5c	Fig 5d	Fig 5e
	010506	Fig 5f	Fig 5g	Fig 5h	Fig 5i	Fig 5j
	010513	Fig 5k	Fig 5l	Fig 5m	Fig 5n	Fig 5o
990923s	v2.x	Fig 6a	Fig 6b	Fig 6c	Fig 6d	Fig 6e
	010504	Fig 6f	Fig 6g	Fig 6h	Fig 6i	Fig 6j
	010511	Fig 6k	Fig 6l	Fig 6m	Fig 6n	Fig 6o
000317n	v2.x	Fig 7a	Fig 7b	Fig 7c	Fig 7d	Fig 7e
	010506	Fig 7f	Fig 7g	Fig 7h	Fig 7i	Fig 7j
	010513	Fig 7k	Fig 7l	Fig 7m	Fig 7n	Fig 7o

III. Observations

The v2.x processed data show relatively little magnitude-dependent offsets between psf-fit and aperture photometry, with the glaring exception of 990523n. For 990523s, the offsets are near zero for bright stars, but increase systemmatically towards fainter magnitudes. The effect is most pronounced in the K_s band.

The cross-scan distortion bias is apparent in the J-band for the two early northern nights for which DMAGCOR was not run (971116n and 980403n). For the other nights, the cross-scan dependence on the bias is small and roughly constant, again with the exception of 990523n. For this night, there is a relatively large (0.01-0.02 mag), approximately constant offset between the profile-fit/aperture residuals between the bright and faint stars.

With the exception of the two early northern nights, rhere is no apparent profile-fit/aperture photometry offset as a function of cross-scan position for bright stars in both the original v2.x and v3 processing. This is expected because the MAPCOR and DMAGCOR correction are derived from the high signal-to-noise stars.

The scatter in the profile-fit/aperture photometry residuals for brighter stars is smaller in the v3 test runs.
The v3 data show different and slightly worsemagnitude dependent profile-fit/aperture biases than v2.x data on several nights. See in particular the J-band for 971116n, K_s band on 980319s, all bands on 980403n and 990923s.

Because there are some relatively strong magnitude-dependent biases between the profile-fit and aperture photometry in the v3 data (and 990523n v2.x), there are resulting systemmatic cross-scan position-dependent differences of up to 2% between bright and faint sources.

IV. Link to Discussion on the Origin and Mitigation of the Offsets by Ken Marsh

Last Update - 14 May 2001
R. Cutri - IPAC