VI. Analysis of the Release Catalogs

1. Comparison of Achieved Performance of the All-Sky Release Catalogs
with Level 1 Science Specification

b. PSC Photometric Uniformity

i. Executive Summary

The 2MASS Level 1 Science Requirements specifies that point source photometry will exhibit <4% deviations or biases in spatial uniformity. The All-Sky Release PSC easily meets the Level specification. This specification is driven by the concern that drifts in calibration at small spatial frequencies might lead to inference of spurious features using star count analyses.

We estimate that the PSC exhibits <1-2% global photometric variation in calibration, which will make it a major resource for studies of Galactic structure.

This is demonstrated:

Each of these tests are discussed briefly below.

ii. Zero Point Stability

Spatial uniformity is imposed on 2MASS photometry on the largest scales by the Survey's photometric calibration strategy. Nightly photometric transformations are derived from hourly measurements of one of 35 calibration fields. The fields are distributed on approximately two hour centers in RA at declinations of approximately -30°, 0° and +30°. The equatorial fields were often observed from both observatories on the same night. Thus, the photometric system is tied between the 2MASS observatories by referencing all photometry, in part, to measurements of the same standard stars.

The nightly photometric transformations are derived by fitting the difference between the instrumental and "true" brightnesses of a networks of 958 standard stars contained in 35 calibration fields, as a function of time. With the Survey observations calibrated against the nightly transformations derived from the spatially distributed network of calibrators, the achieved spatial uniformity is limited primarily by the internal consistency of magnitudes of stars in the network.

The standard star "true" magnitudes were derived using a procedure that derived the set of photometric solutions to each night's calibration measurements that minimizes the overall variance of the solutions for all observations (Nikolaev et al. 2000, AJ, 120, 3340). This global 2 minimization procedure used the database of between 600 and 3500 standard star measurements accumulated over the Survey. It is sensitive to small (<1%) systematic changes photometry around the sky, and provides a test of the internal stability of the derived calibrations. Figures 1, 2 and 3 show the magnitude residuals between standard stars observed from both observatories on the same night, extracted during the 2 minimization process, plotted as a function of brightness, right ascension and declination.

We observe the self-consistent tie between the North and South to be at the <1.5% level (which is close to the limits of our analysis), well within the 4% photometric spatial uniformity requirement. There are a few measurement differences that fall outside the 4% tolerances, but all are related to random contamination of individual measurements, rather than systematic trends in the photometry.

Figure 1
N/S mag Differences vs. Magnitude
Figure 2
N/S mag Differences vs. RA
Figure 3
N/S mag Differences vs. Declination

iii. Calibration Stars Measured in Survey Observations

Each of the 958 2MASS standard stars was observed during normal Survey scans, and they are listed in the PSC. Thus, they can be used as test particles to examine the uniformity of the calibrations that were applied to the PSC. This analysis is discussed in detail in this memo. The results are summarized here.

Figure 4 shows histograms of the differences between the "true" magnitudes (Mcal) and PSC magnitudes (Mpsc) for all the calibration stars. The mean and RMS magnitude differences (Mcal-Mpsc) in each band are given in Table 1. Statistics are given for the full ensemble of stars, and also broken out for stars observed from the northern and southern observatories.

Figure 5 shows the magnitude differences for each star as a function of "true" magnitude. The small points are the differences for individual stars, while the large black points are the mean and RMS differences in 0.5 mag bins. Figures 6 and 7 show the mean and RMS magnitudes differences for all stars in each calibration field as a function of the field right ascension and declination. Figure 8 shows the magnitude differences as a function of survey day number. In Figure 8, the small points are the differences for individual stars, and the large colored points are the mean and RMS differences for all stars in two-day bins.

There is no measureable net difference between true and PSC magnitudes for the full ensemble of standards, or from the observations from each hemisphere. There are no biases larger than 1-2% between the "true" and PSC photometry as a function of source brightness or time. The most significant outliers are the associated with the standards in the two fields which lie deep in the Galactic Plane. Their influence is diluted on the net calibrations, since they were always observed with several other fields on any given night. As with the global calibration analysis, measurements of individual stars sometimes fall outside the 4% range, but these are always associated with contamination of the individual measurements, rather than with systematic effects.

The one standard deviation levels of the mean photometric difference distributions in all three cases is <0.02 mags, well within the Level 1 Requirement for spatial uniformity.

Table 1 - Mean and RMS Magnitude Differences For Calibration Stars
 All   North South 
Band< Mcal-Mpsc >RMS< Mcal-Mpsc >RMS< Mcal-Mpsc >RMS

Figure 4
Mag Residual
Figure 5
Mag Residuals vs. Brightness
Figure 6
Mag Residuals vs. RA

Figure 7
Mag Residuals vs. Dec
Figure 8
Mag Residuals vs. Time

iv. Tile Overlap

Each 2MASS survey Tile overlaps adjacent Tiles by ~1 arcmin in right ascension and ~8.5 arcmin in declination. Sources falling in the overlap regions are observed multiple times during the Survey, and photometry is extracted from each of these measurements. Because the time separating observations of adjacent Tiles ranged from ~7 min up to ~3 yr, and because adjacent Tiles were sometimes observed from different observatories, the comparison of photometry from each apparition of these multiply-detected sources probes photometric uniformity both on a variety of timescales and between observing systems.

During the process of selecting which apparition of multiply-detected sources that would be used in the PSC (duplicate source resolution), sources from adjacent scans in the Tile overlap regions were positionally associated and their extracted magnitudes were compared. The mean photometric offsets for all high signal-to-noise (SNR>20) sources common to each scan pair were computed. The mean and standard deviation photometric differences for scan pairs that contain at least 100 SNR>20 sources in the overlap regions are presented in Table 2. Statistics for three different sets of scan pairs are presented. The column labeled FSky contains the statistics for all scan pairs in the Survey. The column labeled N2N gives the statistics for only scans pairs in which the adjacent Tiles were observed on different nights. Finally, the column labeled Hemis gives the statistics for all scan pairs in which the adjacent Tiles were scanned from different observatories.

The mean photometric offsets between Tiles for all sets are <0.01 mag. Moreover, the dispersion in the offsets is <0.02 mags, indicating that the photometric precision of the point source photometry is excellent.

Table 2 - Mean and Standard Deviation Photometric Offsets of SNR>20 Sources in Tile Overlaps Regions (Counts >100)
Band N2N HemisFSky
Mean Delta MagStdMean Delta MagStdMean Delta MagStd
J +0.00040.019+0.00860.019+0.00250.014
H -0.00030.017+0.00050.018+0.00400.015
Ks -0.00020.017-0.00380.017+0.00010.014

v. Global Color Variations

The science driver for the 2MASS point source photometric uniformity requirement is the objective to use star count and color analysis to measure structure in the Milky Way, study stellar populations and variations thereof, and to infer estimates of extinction over small spatial scales. The spatial variations intrinsic to the data acquisition and processing ultimately limit the precision with which such studies can made.

An analysis has been carried out to probe the intrinsic color variations of the PSC; full details are available in this memo. In this analysis, the color changes in for sources having a narrow J-H color range (0.30J-H0.34), chosen empirically from the 2MASS color-color diagram to be blueward of the giant branch. This population, therefore, emphasizes the extended solar neighborhood. The center of this color range corresponds to an early G dwarf (e.g., Bessell & Brett 1988, PASP, 100, 1134). Sources were selected from the PSC that have the highest photometric quality (ph_qual="AAA"), no photometric confusion or association with image artifacts, and that were in the non-saturated regime of the 1.3 s R_2-R_1 exposures (rd_flg="222"). Figure 9 shows the mean source counts in equal area spatial bins for this sample, in Galactic aitoff projection, and demonstrates the effect of low and moderate latitude reddening and extinction on number counts.

Figures 10, 11 and 12 show the spatially-binned mean J-Ks, H-Ks and (J-H)-(H-Ks) colors of the sample in Galactic Cartesian projections, respectively. The latter is roughly perpendicular to the reddening vector, so should best reveal any systematic non-reddeding color changes in the PSC. Figure 13 shows the root sample variance in J-Ks color in an Equatorial Cartesian projection. Analysis of these figures show:

Figure 9
Number Counts
Figure 10
Mean J-Ks Color
Figure 11
Mean H-Ks Color

Figure 12
Mean (J-H)-(H-Ks Color
Figure 13
Std Dev in J-Ks Color

[Last Updated: 2003 January 27; by R. Cutri, M. Weinberg and S. Wheelock]

Return to VI.1.