2MASS All Sky Images

Table 1

Image	Linear stretch	Log stretch	Number of stars	Selection Criteria
Star count maps
Level 1 specs	J   H   K	J   H   K	N(J) = 261,201,372 N(H) = 246,302,241 N(K) = 170,397,798	J <= 15.8 for J maps H <= 15.1 for H maps K <= 14.3 for K maps PH_QUAL = A,B,C,D, or E in the appropriate band JHK_cmsig < 0.5 (probably not necesssary)
All magnitudes	J   H   K	J   H   K	N(J) = 448,297,218 N(H) = 427,557,426 N(K) = 361,932,020	PH_QUAL = A,B,C,D, or E in the appropriate band
"Bright" stars (mag <= 11)	J   H   K	J   H   K	N(J) = 6,920,862 N(H) = 12,983,412 N(K) = 16,386,998	J <= 11 for J maps H <= 11 for H maps K <= 11 for K maps PH_QUAL = A,B,C,D, or E in the appropriate band
Single band detections (no mag limits)	J   H   K	J   H   K	N(J) = 26,327,109 N(J) = 4,545,614 N(K) = 3,361,090	For J map, RDFLAG=0 for H and K, RDFLAG!=0 for J For H map, RDFLAG=0 for J and K, RDFLAG!=0 for H For K map, RDFLAG=0 for J and H, RDFLAG!=0 for K PH_QUAL = A,B,C,D, or E in the appropriate band
Single band detections (J<=15.8, K<=14.3)	J   H   K	J   H   K	N(J) = 4,081,295 N(H) = 1,295,450 N(K) = 1,837,721	For J map, RDFLAG=0 for H and K, RDFLAG!=0 for J, J <= 15.8 For H map, RDFLAG=0 for J and K, RDFLAG!=0 for H, H <= 15.1 For K map, RDFLAG=0 for J and H, RDFLAG!=0 for K, K <= 14.3 PH_QUAL = A,B,C,D, or E in the appropriate band
Color maps
Level 1 specs	J-K   J-H   H-K   100um	J-K   J-H   H-K   100um	N(J-K) = 158,454,661 N(J-H) = 229,463,870 N(H-K) = 166,514,813	J <= 15.8, H <= 15.1, K <= 14.3 (applied to 2 bands) PH_QUAL = A,B,C,D, or E in the appropriate bands JHK_cmsig < 0.5 (probably not necesssary)
Astrometric uncertainties
Major axis	J   H   K		N(J) = 261,270,617 N(H) = 246,416,023 N(K) = 170,435,798	J <= 15.8, H <= 15.1, or K <= 14.3 PH_QUAL = A,B,C,D, or E in the appropriate bands
Minor axis	J   H   K		N(J) = 261,270,617 N(H) = 246,416,023 N(K) = 170,435,798	J <= 15.8, H <= 15.1, or K <= 14.3 PH_QUAL = A,B,C,D, or E in the appropriate bands

Notes

Single-band detections

1. While inspecting the single band detection images, I found a few regions of enhanced J-only and K-only detections. These are difficult to see on the gif images above, but are clearly visible on the fits images. One such region is at (ra=69.2, dec=-62.1).

   To investigate this further, I downloaded from gator sources within a 1 deg radius of this position that had a cc_flg of '000'. Figure 1a shows the spatial plot of the single band detections, and Figure 1b shows a similar spatial plot but for sources brighter than the Level 1 specs. These plots clearly show that these single band detections are artifacts along the diffraction spikes of a bright star. The brightest star in this field has a magnitude of (J=-2.652, H=-3.732, and K=-4.227).

2. As a second example, (Figure 2a, Figure 2b) show similar plots for a region around (ra=198.5,dec=-2.8). Again the single band detections are un-flagged diffraction spike artifacts. Note that this was not one of the more conspicuous examples on the singe-band detection all-sky map, so perhaps this problem is more prevelant than I first thought. The brightest star in this field has a magnitude of (J=-0.356, H=-1.606, and K=-2.003).

3. The K-band single detection plots clearly show that there are more single band detections in the north compared to the south. To investigate this, I made a spatial distribution plot of the K-only sources for a region search (ra=223.2, dec=56.6). As shown in Figure 3, many of the K-band single detections are aligned in declination. Prominent declination bands are located at RA=221.498, 222.42, 223.61, 224.02, etc, ... In these bands, the single band detections are not always separated by 82" (i.e. the frame separation), but they are separated in a round multiples of 82".

   Figure 4 shows a similar plot for a larger region of sky that straddles the north/south survey boundaries. This figure clearly shows that the north contains more single band detections, and that a number of sources are aligned in declination and are presumably artifacts. However, more most of the single band sources seem to be randomly distributed. Figure 5 shows the differential K-band magnitude histogram of the single band detections for the northern and southern regions for this same field. This plot clearly shows that the excess among the norther population is due to faint sources (K=15-16) in the north. Figure 6 expands on this figure by showing the magnitude histograms of all detected sources for both the north (solid histograms) and the south (dotted histograms). The red histograms in the K-band panel show the magnitude distributions of the K-band only sources. This plot shows that the north is more sensitive in this region of the sky, and that the single-band detections in the north occur where the south is losing sensitivity. It is conceivable that the increase single band detections result from increased sensitivity in the north.

4. The following table contains links to gif images in order to investigate the nature of the single band detections. For this exercise, I again picked the region between RA=160 and 90, and DEC=0 and 24. I analyzed sources with the following criteria
   1. Detected in K-band
   2. CCFLAG = 000, gal_contam=0, mp_flag=0
   3. Must be a read2 source in any detected band.
   4. Must be a blend flag = 1 source in any detected band.
   5. Must have PH_QUAL of A, B, C, D, or E in any detected band.

   Given the above selection criteria, I divided the sources into "north" (dec > 12) and "south" (dec < 12), and further into multi-band detections (i.e. 2 or 3 band detections). Multi-band detections should be solid detections, and I am assume that all of these are real detections. The following plots compare the K-band only detections with the multi-band detections to see how they are similar or different.

   Table 2

   Plot	North	South
   K-mag vs. PSF chi2	Multi-band    K-only	Multi-band    K-only
   Histogram N(3sigma) for K>=15	X	X
   Histogram N(frames) for K>=15	X	X
   K-band PH qual flag; K>=15	X	X
   K-band PSF chi2; K>=15	X	X

   Table 3: Spatial plots

   Selection Criteria	GIF
   K-only; K>=15; PSF chi2 <= 1.5	X
   K-only; K>=15; PSF chi2 >= 1.5	X

   Things to note in the above plots:

   1. K-mag vs. PSF chi2 plots
      This is the most distinguishing plot in the above table. The single band plot for the north shows roughly 3 populaton of sources:
      1. Relatively bright sources with high PSF-chi2. From visual inspection of the image atlas, these are double or extended sources but without any any associated error flags, but they do generally have higgh chi2.
      2. Faint (K > 15) with low PSF-chi2 (< 1.5).
      3. Faint (K > 15) with high PSF-chi2 (> 1.5).
   2. Spatial plots
      These plots clearly show that that the chi2 > 1.5 sources contain many of the apparent artifacts in the K-band only sources. That is, they contain sources that are aligned in declination. There are also randomly distributed sources with chi2 > 1.5 as well.
5. Analysis done by Roc Cutri

   Hi John,
   
   I've been doing a little follow-up on the Ks-only sources
   to see if there are any other parameters that could
   be used with chi-squared to ID the good and bad apples.
   Don't know if you want to add any of this to your page.
   
   The attached plot (Figure 7) 
   shows histograms of the chi-squared
   distributions for several high galactic latitude
   (|b|>30) samples of Ks-band only sources.  The black line
   shows the distribution for the full sample, and looks like
   the figures in your table 2.  The blue curve shows the
   distribution for Ks-only sources that have optical
   associations within 2", another reliability indicator.
   These are nicely concentrated to the unit-chi-squared peak.
   The green curve shows the distribution shows the
   Ks-only sources that are flagged as either contaminated
   by a galaxy (gal_contam >0) or identically an extended
   source (ext_key not null).  These peak slightly
   longwared of unity with a smooth tail.
   
   The red curve shows the Ks-only sources with no optical
   associations and no extended source contamination.
   Most of the high chi-squared peak is still there, but
   the peak near unit chi-squared is greatly diminished
   (but unfortunately not completely gone).  This suggests
   a way to identify some of the spurious sources at high latitude.
   I'm afraid at the lower latitudes, the leverage of the
   optical counteparts will be lost, though.
   
   Some statistics:
   
   Number of PSC sources with |b|>30               48163068
   
   Number of PSC sources with |b|>30
      and Ks detections				40727496
   
   Number of PSC sources with |b|>30
      and Ks-only detections                         113255 
   
   Number of PSC sources with |b|>30
      and Ks-only and no optical counterpart
      and no extended source contamination            73149
   
   Roc
   

6. Mike Skrutskie cross correlated the K-band only sources with Sloan between 352-55 degrees in RA. Figure 8 shows the PSF chi-squared distributed for K-band only sources with and without Sloan counterparts. Consistent with the above plots, the K-band only sources without Sloan counterparts tend to have a higher chi-squared distribution, although a simple chi-square cut is insufficient to identify the K-band only, no optical counterpart sources.