VI. Analysis of the Release Catalogs
1. Comparison of Achieved Performance of the All-Sky Release Catalogs
with Level 1 Science Specification
h. XSC Sensitivity
The level-1 specification for sensitivity to extended sources are 15.0, 14.2, and 13.5 at J, H, and Ks, respectively. 2MASS clearly exceeds this specification at J, but is probably borderline at H and Ks.
This specification is intended as a measure of the repeatability of flux measurements to better than 10% (0.109 mag) at the stated magnitude limits. (This does not include calibration variations, which are covered by the uniformity specification.) The meaning of a flux measurement for an extended source is not as straightforward as for a point source, since there are so many different ways of measuring the magnitude (and no such thing as a "total" magnitude can be reliably measured at any wavelength). A fixed aperture test gives a good measure of the actual photometric repeatability, but a more commonly-used measure is the flux within an ellipse fit to an isophotal level. This latter method is more challenging, since uncertainties in the ellipse fitting to a faint isophote (20 mag/arcsec2 at Ks) introduce additional sources of scatter.
We used the duplicate measurements of sources in the overlaps between scans to test the repeatability of the magnitudes. This test will likely overestimate the uncertainties, because at least one of the measurements is made close to the edge of the frame. This means there is greater uncertainty in the background subtraction, and the outer parts of larger galaxies may be cut off. Figure 1 shows the magnitude differences (Catalog source mag - duplicate source mag), as a function of the Catalog source magnitude. The error bars show the rms population dispersion in each magnitude bin. The differences shown are all for measurements with overlapping scans on the same night, to avoid differences due to varying calibration solutions.
|
| Figure 1 |
To determine how well the magnitudes met the Level-1 specification, we look at the galaxies within 0.1 mag of the magnitude limit, and divide the resulting standard deviation by 21/2 (because we are considering two independent measurements). There are between 3000 and 4200 pairs of sources meeting these requirements in each band, yielding the following estimates for the magnitude uncertainties at the magnitude limits:
| Mag Range | Ks-band 20 mag/sq. arcsec isophotal magnitude uncertainty | 7´´ circular aperture magnitude uncertainty |
| J = 14.9-15.0 | 0.102 | 0.096 |
| H = 14.1-14.2 | 0.115 | 0.116 |
| Ks = 13.4-13.5 | 0.141 | 0.108 |
These values are all quite close to the 10% repeatability level. The Ks-band elliptical isophotal magnitudes exceed the specification, but this appears to be due primarily to uncertainties in the elliptical fit, since the fixed aperture magnitudes for the same set of galaxies meet the specification.
Lastly, as a reminder that comparing sources at the edge of the frame is a particularly difficult test to pass, we note that sources close to the edge of a frame exhibit some complicated magnitude-dependent biases. In Figure 1, the magnitude of the source closer to the edge is subtracted from that of the Catalog source. For sources slightly brighter than the Catalog limit, there is a small negative bias for sources nearer the edge, while for fainter sources the bias becomes strongly positive. In the 0.1 mag bins in the table above, the biases ranged from 0.012 to 0.042 mag--the measurements of sources closer to the edge were dimmer, probably because of problems in determining the backgrounds near the edge. The shift to a positive bias at dimmer magnitudes is probably a selection effect, since duplicate sources that are randomly scattered to fainter magnitudes are unlikely to be detected, so our comparisons only include one side of the distribution. Because the great majority of catalog sources are well away from the edge of a frame, these kinds of biases should be less problematic, and the Catalog magnitudes more consistent than these tests indicate.
[Last Updated: 2003 Jan 20; by S. Schneider]
Return to VI.1.