Executive summary: Approximately one million stars
in the Point Source Catalog have been found to have erroneous J-band
photometry. The affected stars fell on the central column of the
J-band array of the Southern hemisphere 2MASS system. Stars that fell
on this column have reported J-band fluxes that are approximately 10%
lower than their actual fluxes and thus 2MASS J-H and J-K colors that
are too red. Queries of the 2MASS All-Sky Release PSC, Survey
Point Source Reject Table, and the 6x and Calibration Point Source Working
Databases can avoid/identify these stars by including the
hemis='s' and dist_edge_ew=251 and (dist_edge_flg='se' or dist_edge_flg='ne').
The 2MASS Calibration Field scans are distinguished among the 2MASS data sets in that they consist of highly redundant observations of selected small regions of the sky. Since most stars did not vary significantly during the duration of the Survey, this database can be used to evaluate the consistency of photometric extractions. When observing light curves for individual stars observed more than 3000 times in calibration field 90067, approximately 1 in 10 stars exhibit anomalously low J-band fluxes in about 1 in 50 observations (Figure 1). This effect is correlated with the star being located at dist_edge_ew of between 250.8 and 252.2E on the southern hemisphere J-band array. This dist_edge corresponds to the center of the array. A similar anomalous photometric behavior, although more sparse, occurs in southern J-band for sources with dist_edge_ew=130-132W. The photometric bias is typically about 0.1 mag faint compared with nominal photometry for the source.
Because the calibration data were generated by the same pipeline that produced the main survey data, this contamination should appear in the PSC as well. To examine this possibility a "badpsc" was generated by selecting southern hemisphere sources with dist_edge_ew between 250.8 and 252.2. Figure 2 shows a blink between high latitude PSC stars with ?_cmsig<0.05 and "badpsc" stars with a similar constraint on ?_cmsig. Indeed the photometry for the "badpsc" is both biased and noisy at about the 0.1 mag level. This problem primarily affects the bright end of the catalog since measurement uncertainty will dominate at fainter magnitudes. There are 1.2 million southern hemisphere sources in the PSC that match the bad dist_edge_ew constraint and have ?_cmsig<0.10.
|Figure 1 - Light curve for star with gcntr=11884290. Beginning around JD=2451200 occasional J-band observations appear about 0.1 mag too faint.||Figure 2 - Blink comparison of a sampling of the high-latitude PSC with a selection from the PSC that required hemis='s' and ?_cmsig<0.05. Photometry from dist_edge_ew=251 is offset and noisy. Click on the thumbnail to view the blink animation.|
The suppression of estimated J-band flux for stars traversing the central column of the array appears to arise because the J-band flat fields (obtained under bright twilight illumination) were not appropriate in that column for the low background J-band survey images. Evidence for this can be found in southern hemisphere coadded images. Figure 3 compares J, H, and Ks Atlas Images from scan 015 on 990205. A bright star crosses the array around frame 160. In the H- and Ks-bands the scattered light from this star shows a smooth monotonic variation away from the star. In the J-band image this scattered light is suppressed along the central column of the array indicating that the flux was not scaled properly during image calibration.
|Figure 3 - Comparison of 2MASS Atlas Images containing a saturated star near the center of the array. For the H-band and Ks-band the scattered light from the star varies smoothly across the central column. In the J-band the star's scattered light flux is suppressed along the central column indicating that the applied flat field was not correct for that column.|
The suppression of sky background in this central column will NOT appear in J-band coadd images of blank sky. During 2MASS image calibration (IV.2d) a median sky background illumination correction was subtracted from each image after the flat field correction was applied. This "sky offset" correction tends to remove any fixed illumination pattern that is present on the focal plane during the interval of frames from which the sky offset correction was constructed. Hence the suppressed central column in the southern J-band images is effectively hidden for constant dark background levels.
Figures 4 and 5 contain raw image frames that contributed to the J-band Atlas Image shown in Figure 3. These frames have been calibrated by applying the standard
calibrated = (frame - dark) / (flat - dark)where the flat and dark frames are single frames from the flat and dark sequences taken on the same night. In these images the central column is suppressed for the entire extent of the array because no median sky background subtraction is performed following the flat fielding operation and the uniform airglow background is erroneously scaled in the central column. Figure 6 contains a plot of the average of 34 rows of one of the calibrated raw images away from the bright star (the positive spikes are stars somewhere along those particular rows). The sky flux in the suppressed column is 460 compared with a sky flux of 530 over the rest of the array - an error of 13%. Since a star crosses one full column in this band as it scans from the extent of the array and since star images extend over more than one column even when centered on a column, this error is consistent with the 5-10% flux suppression seen in pipeline source extractions from stars traversing this column.
|Figure 4 - A raw J-band Survey frame calibrated using the flat and dark frames acquired earlier in the night, but with no sky illumination correction. The flat field is inappropriate for the central column as revealed by the suppression of the atmospheric airglow flux in this column.||Figure 5 - Same as Figure 4, but later in the scan.||Figure 6 - Average flux from a cut across the image in Figure 4. The downward spike at the central column (column 127) amounts to a flux suppression of 13%. The upward spikes are stars included in the region.|
[Last Updated: 2008 August 1; M. Skrutskie]