Overview of 2MASS Catalog Astrometric Quality
H.L. McCallon 12-12-02
I. Quality of 2MASS Positions:
Comparison of the 2MASS catalog with a variety of other catalogs indicate that 2MASS positions in the International Celestial Reference System (ICRS) are accurate (in-scan and cross-scan separately) to 70-80 mas (RMS) over most of the magnitude range,far better than the Level 1 requirement of 500 mas. This conclusion is based on comparisons of positions and uncertainties with the Tycho-2 and UCACr10 catalogs as well as 2MASS scan overlap data as shown in Figure 1. These catalogs will herein be referred to as the "comparison catalogs" (CCat). For details of these comparisons, the reader is referred to the 2MASS:Tycho-2 and 2MASS:UCACr10 comparisons as well as a consistent independent UCACr10:2MASS comparison by Norbert Zacharias of the U.S. Naval Observatory. This can demonstrated by a single figure, Figure 1 - 2MASS Astrometric Quality, which plots a set of derived 2MASS RMS values as a function of Ks magnitude.
|
First, consider what is meant by a "derived 2MASS RMS". Since the comparison catalogs (Tycho-2 and UCACr10) also have errors their contribution to the RMS of the 2MASS:CCat differences needs to be removed to obtain a measure of the 2MASS RMS. The "derived 2MASS RMS" values are obtained by subtracting in quadrature the mean quoted comparison catalog sigma from the 2MASS:CCat difference RMS value computed for each bin. Since the quality of the Tycho-2 astrometry for stars previously included in Tycho-1 [referred herein as Tycho-2(1)] are considerably better than those which were not [Tycho-2(2)] they are considered separately. Note that an underestimate of comparison catalog errors results in an overestimate of the 2MASS RMS and vice-versa. The 2MASS overlap differences were divided by two before the RMS values were computed. This assumes the two positions are independent and will tend to underestimate the RMS if they are not. More on this later.
In the aforementioned figure, data taken from the northern telescope are presented in the left panels and from the southern telescope in the right panels. Cross-scan results can be found in the upper panels and in-scan in the lower. In each of the four panels 2MASS RMS values derived from 2MASS differences with Tycho-2(1) [solid black], Tycho-2(2) [solid blue], UCACr10 [solid red] and scan overlaps [solid green] are shown. For comparison, the mean quoted 2MASS sigma is shown as a dotted black line.
With the notable exception of the Tycho-2(2) results, these form a reasonably consistent set of measurements. The UCACr10-derived 2MASS RMS values tend to be somewhat higher that the Tycho-2(1)-derived values. A certain amount of that difference might be expected, since the Tycho-2(1) was the primary reference for the 2MASS reconstruction and the UCACr10 was not used at all. Given that the overlap positions are not totally independent, the overlap RMS values can be expected to be low. But since any correlations between overlapping scans are not expected to be magnitude dependent, the change in overlap-derived 2MASS RMS with magnitude, however, is meaningful. Raising the overlap-derived curve up to between the Tycho-2(1) and UCACr10 derived curves should provide a fair measure of how fast the 2MASS RMS climbs after a Ks magnitude of 15, where other comparison catalog data are not available. Given the discrepancy of Tycho-2(2)-derived 2MASS RMS values with all other measures, the most likely explanation is that the Tycho-2(2) quoted errors are underestimated. As can be seen in the 2MASS:Tycho-2 comparison there are no significant systematic differences between Tycho-2(1) and Tycho-2(2).
II. Quality of 2MASS Uncertainties:
Comparing the mean quoted 2MASS sigmas [dotted black line] to the derived 2MASS RMS values in Figure 1 and ignoring the Tycho-2(2) data [solid blue], it can be seen that quoted 2MASS sigmas track reasonably well except at the bright (Read1) end. For Ks magnitudes between 4.0 and 8.5 (typically unsaturated Read1's) the quoted 2MASS sigmas are too low but for brighter magnitudes (typically saturated Read1's) they are very conservative.
Although of lesser amplitude (2MASS RMS<100 mas) than the magnitude dependence, the detailed comparisons referred to earlier
also show an increase in 2MASS errors hear the poles and very near scan edges.
http://spider.ipac.caltech.edu/staff/hlm/2mass/overv/overv.html
Comments to: Howard McCallon
Last update: 17 Dec 2002