a.  Identification of suspicious objects:

• Color-color and color-magnitude diagrams (Figure 1)

The examination of photometry in a 4 square degrees field in the Taurus molecular cloud leads to the identification of about 50 objects with H-K_s<-0.5 and J-K_s<0. The color of these objects cannot be explained by the presence of the cloud itself: extinction reddens stars and young stellar objects associated with a cloud never have any blue excess. The color-color diagram indicates that the problem is with the K_s photometry.
The blue stars have uniform spatial distribution in the 4 square degrees field and their K_s magnitude range from 11 to 18. Because of the uniform distribution of the blue stars, and because they are mixed in with stars having normal colors, the erroneous K_s photometry is not caused by the photometric normalization error described in the Gizis's page.

Figure 1

b.  Understand what occurs:

• Comparison with the aperture photometry

Figure 2 compares the aperture photometry with the photometry obtained by PSF fitting for J, H and K_s. The K_s photometry can be incriminated: PSF fitted photometry in K_s appears to be from 1 to 3 magnitudes fainter than aperture photometry for the blue stars. Blue diamonds in Figure 3 represent the PSF fitted photometry of the blue stars and green diamonds represent same stars with the aperture photometry. The locus of selected stars seems correct with aperture photometry (stars are spread along the main sequence line).

Figure 2	Figure 3

• Information about the PSF fitting in the catalogue

The Point Source Catalogue contains the K_s band reduced chi-squared value of fit. Figure 4 shows this value for all stars in the field (black) and for blue stars (red diamonds). Virtually all of the sources with bad K_s magnitudes have large chi-squared value. However, other sources with large chi-squared value show usual color.
Figure 5 presents J, H and K_s chi-squared value on the same graph: most of stars have good value in J and H.

Figure 4	Figure 5

• Example of blue objects

Objects incriminated in this analysis have a close companion or an extension which contaminates the stellar flux (Figure 6).
For this field, a companion is detected in 80% of the case at a distance lower than 6 arcsecs and it has always the same chi-squared value of fit in K_s than the brightest star. Companions are also detected in J and H, but the photometry in those bands is not affected.

Figure 6

• Solution

The problem has been identified: the anomalous K_s magnitudes are due to the inadequate design of the passive deblending algorithm currently used in 2MASS photometric processing (PROPHOT). Sources are photometered in decreasing order of brightness, and sources that fall within a certain critical radius of a given source are reduced
together as a group. This would have been okay, except that the neighboring sources were never revisited during subsequent reduction, which meant that no account was taken of the possibility that the outlying neighbors
might have been contaminated by other sources outside the original radius.

The passive-deblending deficiency has been rectified in a new version of PROPHOT that will be delivered with the new production PSFs. As an illustration of the new process, scan 063 from 971129 was reprocessed using the revised PROPHOT, and the results are presented in Figure 7. The chi squared values for the formerly-blue stars are now different. The values are still high because of the contamination by the close companion but the systematic shift in photometry toward blue color is no longer present.

Figure 7