Photometry/Astrometry from 2MASS Quicklook Images

1. Overview

The lossy compression algorithm used to reduce the file size of the 2MASS Atlas Images (i.e. the Quicklook images) compromises their photometric integrity. This document characterizes the quality of the Quicklook images by performing point source extraction of 5 such images and then comparing the derived photometry and astrometry with results listed in the final 2MASS Point Source Catalog. The salient results are summarized in Table 2 for the right ascension astrometric accuracy, Table 3 for the declination astrometric accuracy, and Table 4 for the photometric accuracy.

2. Caveats

1. Astrometry and photometry obtained from the Quicklook images are not as accurate as that inherent to the 2MASS Point Source Catalog, which were derived from the uncompressed, stacked raw images.
2. The analysis of the Quicklook images described in this document represents an extremely small subset of the Image Atlas and does not sample the full parameter space that may influence the photometric/astrometric accuracy (e.g. source density). There is no guarantee that the results obtained here are representative of other regions.
3. In the analysis presented here, photometry in the Quicklook images was measured using a fixed aperture, while most objects in the 2MASS Point Source Catalog were measured with point-spread-function (PSF) fitting photometry. More accurate results could potentially be obtained by using PSF fitting photometry on the Quicklook images, especially in source-confused regions.
4. While the 2MASS data reduction pipeline accounted for image distortion when deriving astrometry, time constraints prevented the distortion corrections from being applied to the 2MASS images. Therefore, the astrometry derived from the images and that contained in the 2MASS Point Source Catalog can systematically vary by up to ± 0.2 arcsec offset as a function of pixel position in the image.
5. Aperture corrections are not stored in the 2MASS image headers. If one chooses to obtain photometry from the images, it is recommended that one performs relative photometry on the image and then normalize the relative photometry to calibrated photometry using the 2MASS Point Source Catalog magnitudes for sources in the field.
6. Stars bright than approximately J < 9 mag, H < 8.5 mag and K_s < 8 mag are saturated in the 2MASS images. The precise saturation limit varies depending on the atmospheric seeing conditions and the sky background.
7. The 2MASS coadded images have been smoothed with a gaussian kernel and sampled at half the raw camera pixel size. Therefore, adjacent pixels in the images contain correlated noise, which is often not taken into account in standard photometry packages when estimating photometric uncertainties. The analysis described here does not attempt to validate that the photometric uncertainties obtained from publicly available photometry packages accurately reflect the actual measurement uncertainties.

3. Analysis

a. regions analyzed

To assess the accuracy to which quantitative infromation can be extracted from the Quicklook images, point source astrometry and aperture photometry was performed on 5 such images and compared with values tabulated in the final 2MASS Point Source Catalog. These images were selected to cover a range of source density and time, and to sample images in both hemispheres. These 5 fields will be referred to by the ID number in the first column of the following table:

b. source identification

Point sources were identified in the Quicklook images using DAOFIND as implemented in IRAF using a detection threshold of 5 sigma. The FITS keyword SKYSIG was used as the estimate of the sky noise.

c. aperture photometry

Aperture photometry was performed using the digiphot.apphot package in IRAF. Photometry was measured using an aperture radius of 4" and a sky annulus that extends from 14" to 20". The magnitude zero point was taken from the FITS headers (keyword MAGZP).

d. astrometry

The X/Y pixel centroids computed by the IRAF package apphot were transformed to equatorial coordinates using the FITS CRVAL,CTYPE,CDELT keywords and the SAO WCS C library, version 2.9.1.

4. Astrometric accuracy of the Quicklook images

The results contained in Figure~2 show that the accuracy of the astrometry from the Quicklook images are better than 0.2" over a broad magnitude range. (However, as mentioned previously in Section 2, the images, but not the 2MASS Point Source Catalog, suffer from image distortion and there can be systematic offsets of up to ± 0.2" in the images.)

Figure 1: Right Ascension astrometric residuals vs. Default 2MASS Magnitude

a) Region I		b) Region II		c) Region III		d) Region IV		e) Region V

Figure 2: Declination astrometric residuals vs. Default 2MASS Magnitude

a) Region I		b) Region II		c) Region III		d) Region IV		e) Region V

5. Photometric accuracy of the Quicklook images

Figure 3 shows the photometric residuals as a function of the default 2MASS magnitude in the final 2MASS data release, where again the residuals are in the sense of (IRAF - 2MASS). Table 3 tabulates the average offset and dispersion in the photometric residuals as a function of the default 2MASS magnitude. For the summary tables, Region I-IV have been combined since they represent relatively uncrowded regions, while the noise in the galactic plane field (Region V) is dominated by confusion.

The photometry derived from the Quicklook images is systemtically fainter relative to the Point Source Catalog photometry by ~ 0.03 mag. This photometric offset is a result of the aperture corrections that were not derived/applied to the Quicklook image photometry. The aperture corrections are not stored in the image header, but can be determined empirically from the images. It is recommended however that any photometry be determined on a relative basis with respect to a star on the image that is also listed in the Point Source Catalog.

After accounting for the aperture corrections, Table 4 shows that the photometric accuracy of the Quicklook images is a few percent for stars brighter than approximately 12 mag and about 0.2 mag or more for fainter sources (> 15). The larger photometric offset for bright stars is due to saturation in the coadded images. (Bright star photometry in the Point Source Catalog is obtained from short integration images.) The effects of source confusion in the galactic plane fields are clearly seen, where the aperture magntiudes are systematically brighter than the PSF fitting magnitudes in the Point Source Catalog for magnitudes fainter than ~ 12 mag.

Figure 3: Photometric residuals vs. Default 2MASS Magnitude

a) Region I		b) Region II		c) Region III		d) Region IV		e) Region V