I. Introduction

6. Cautionary Notes

The data acquisition strategy, data processing system, rigorous quality assurance procedures and extensive data product validation for 2MASS were all designed to generate products of the highest reliability and quality to enable far-reaching research using the properties of the near infrared sky on unprecedented scales and precision. However, validation of a dataset as large as that produced by 2MASS presents a great challenge, and it is expected that there will be problems both known and unforeseen that persist into the release data products.

The following sections contain a summary of known caveats and cautionary notes that pertain to the 2MASS All Sky Release as a whole, and to each of the release products. Section I.6.a presents general Cautionary Notes, and I.6.b, I.6.c and I.6.d address essential information users must be aware of when using the PSC, XSC and Atlas Images, respectively. Sections II.2.b and II.3.b are designed to provide you with guidance and suggestions on how best to use the 2MASS All-Sky Catalogs to maximize their scientific return, and to minimize any potential pitfalls due to their known limitations.

a. General Survey Notes

Sky Coverage
Reliability of Survey Products
Photometric Considerations
Astrometric Considerations
Differences between All-Sky Release and Incremental Data Release Products

i. Sky Coverage

The effective fractions of the sky covered by the 2MASS All Sky Release Data products are:

Image Atlas: 99.998%

Point Source Catalog: 99.65% (J band); 99.51% (H-band); 99.56% (K_s-band)

Extended Source Catalog: ~98%

The majority of the lost coverage area comes from the masking influence of bright stars. Approximately one square degree of sky was not covered by scans included in the All Sky Release area.

Physical Gaps

The 2MASS telescopes scanned the sky in units of Survey Tiles which are one camera frame (8.5´) wide in the Right Ascension direction and 6° long in Declination. As laid out on the sky, each Tile overlaps adjacent Tiles by ~51´´ in RA and 8.5´ in Dec. Small telescope pointing errors sometimes resulted in misplacement of the scans with respect to the predefined Tile positions, leading to possible gaps in coverage between Tiles. These gaps were monitored during the Survey, and special scans were conducted to fill them. At the completion of the Survey, 21 physical gaps between Tiles remained. These gaps encompass an area of 0.71 deg². Their location is described in III.2.c.

Effective Gaps in the Catalogs

The 2MASS PSC and XSC coverages contain larger effective gaps than the imaging area because sources were required to lie >10´´ and >15´´ away from Tile edges, respectively (V.3). This Tile-edge safety boundary was designed to minimize edge-effects that include non-three-band coverage due to slight misalignments of the focal plane arrays, distortion at the edges of focal planes, and partial coverage of extended sources close to Tile edges. For Tile pairs that overlap by less than <20´´ there will be an effective coverage gap in the PSC. There are 43 such gaps, and the total effective missing area in the catalog, including the physical gaps, is 1.26 deg². Similarly, for Tile pairs that overlap by <30´´, there will be an effective gap in the XSC. There are 89 such gaps in the XSC, for a total combined missing area of 1.65 deg². The location of these gaps is described in III.2.c.

Area Lost Due to Bright Stars

Bright stars effectively mask detection of nearby fainter sources because of the influence of their bright image wings, ghosts images, diffractions spikes, latent images and other image artifacts (IV.7). These regions are masked out during source extraction and characterization in the data processing to minimize spurious detections and confusing photometric effects triggered by the artifacts, resulting in further loss of sky coverage. The area lost due to bright stars is much larger than any of the physical or effective gaps in the Tile sky coverage.

The effective areas lost to bright stars in the PSC are 106.16 deg² in J band, 156.21 deg² in H band, and 177.87 deg² in K_s band. Approximately 825 deg² of coverage in the XSC is lost due to the confusing effects of bright stars. Detailed sky coverage maps for the XSC are available in II.6.f.

Effective Loss of Faint Source Coverage Due to Confusion in the Galactic Plane

The PSC and XSC lose effective coverage area due to the increasing effects of confusion in high source density regions in and around the Galactic Plane. This loss of coverage is not as simply quantified as those due to the Survey gaps or the influence of bright stars because the loss is a varying function of Catalog source brightness with field star density.

The point source extraction threshold is regulated by the point-source-filtered noise levels on Atlas Images (IV.4.a). The point source detection brightness limit rises in response to increasing confusion noise. Thus, the PSC completeness limit increases in brightness as the source density increases. This is illustrated in this GIF "movie," which shows the differential PSC source counts as a function of source brightness (see VI.1n). Coverage loss begins at K_s~12.7 mag, and the areal loss increases with decreasing brightness.

The extended source detection thresholds are also automatically increased in response to increasing confusion noise, using the extracted density of point sources brighter than K_s<14.5 mag as a proxy for confusion noise (IV.5). As the star density increases, the source processor requires a source to be increasingly brighter and larger to be operated on as a true extended object. This is because of the increasing incidence of close multiple stars masquerading as extended sources, and because smaller, fainter galaxies are more heavily contaminated by foreground stars. The growth of the excluded area with decreasing extended source brightness is clearly illustrated in this movie showing differential XSC source counts as a function of brightness. Extended source coverage loss begins at K_s~11.5 mag, and the areal loss increases with decreasing brightness.

ii. Reliability of the All Sky Release Products

The 2MASS All Sky Release data products have been demonstrated to meet or surpass the Level 1 Requirements for reliability (cf. VI.5a.ii and VI.5b.ii).

Every effort has been made to remove unreliable sources from the Catalogs, and to flag such entries that do remain. However, the sheer volume of the 2MASS image and source data products make it impossible to guarantee their perfection. For example, the target reliability for sources in unconfused regions of the sky in the 2MASS Catalogs is 99.95% (PSC) and 99% (XSC for |b|>20^o). Even at these strict limits, there may be >170,000 unreliable point sources and >12,000 unreliable extended sources.

Extensive discussion of the characteristics of the remaining unreliable sources in the Catalogs are presented in the PSC and XSC Cautionary Notes sections of the Explanatory Supplement. Examples to help users recognize artifacts that can be found in the over four million Atlas Images in the All Sky Release are discussed in the Image Atlas Cautionary Notes section.

Anomaly Lists

The 2MASS All Sky Catalog products are static. No deletions will be made after their official release, even for sources that are known to be artifacts. Anomaly lists identifying confirmed unreliable sources are provided in II.6.g. These lists will be updated periodically.

iii. Photometric Considerations

Photometric System

Magnitudes reported in the Point and Extended Source Catalogs are in the natural 2MASS photometric system. The system bandpasses are described in III.1.b, and the 2MASS calibration procedure is described in III.2.c and IV.8. Users should be aware that comparisons of 2MASS photometry with measurements made in other common near infrared photometric systems may show systematic differences. Preliminary transformations between the 2MASS and selected other photometric systems were presented by Carpenter et al. (2001, AJ, 121, 2851) using data from the 2MASS Second Incremental Data Release. The transformations have been updated using data from the All-Sky Release PSC, and are given in VI.4b.

Photometric Biases

The global uniformity of the 2MASS photometric calibration is demonstrated to be better than 1-2% over large spatial scales (c.f. VI.2.a.ii). The global uniformity is enforced by the Survey's photometric calibration procedure (III.2.d) that ties all photometry to measurements of a common network of standard stars observed hourly each night from both observatories.

Several sources of bias that affect the measurements reported in the PSC are discussed in detail in I.6.b.viii. These include brightness-dependent biases that exist between measurements of sources that fall in the different brightness regimes of the Survey observations: sources that are non-saturated in the 1.3 s "Read_2-Read_1" exposures, those that are non-saturated in the 51 ms "Read_1" exposures, and those that saturated the 51 ms exposures. These are known to cause features in global source count curves for the PSC. Spatially correlated biases in PSC photometry are known to be caused by mismatches between the instantaneous seeing and the PSF used during profile-fitting photometry measurements, and normalization errors of the profile-fitting measurements. These can produce coherent "jumps" in mean source color correlated with Tile and Atlas Image boundaries. Finally, point source photometry can be biased due to proximity to sources of equal or greater brightness because of gradients induced in the backgrounds by the nearby source.

Clouds

Data taken under non-photometric conditions due to clouds or other instances of decreased or variable atmospheric transmission were identified and rejected from the Survey. However, during final review of 2MASS data quality following the completion of observations, two scans were found to have a brief 10-25% decrease in transparency, presumably because the telescope scanned over a small cloud or contrail. Colors of sources in the affected regions are largely unchanged, however, because the extinction due to the clouds is grey. It was not possible to reobserve these regions following decommissioning of the observatories. Since the impact on photometry was relatively small, the total area affected by the clouds is <0.25 deg², and there were only 1,391 point and 3 extended sources in those areas, these Tiles were included in the Release.

Table 1 gives the description of the two scans affected by clouds, including the Tile number, observatory, observation date and scan number, scan_key, and the central right ascension and approximate declination range over which the transparency was diminished. PSC and XSC sources in the affected areas can be identified through the use of scan information (scan_key or date, hemis, scan) and declination range. The photometric uncertainties for the affected point sources in Tiles 29270 and 205438 were increased by adding in quadrature 0.11 and 0.25 mags to the original uncertainties, respectively, equal to the peak photometric offset induced by the obscuration in each scan. For the affected XSC sources in these Tiles, the aperture photometry uncertainties had 0.11 and 0.25 mags added in quadrature, and the isophotal magnitude uncertainties had 0.22 and 0.50 mags added in quadrature.

Table 1 - Information on Tiles Affected by Clouds

Tile Observatory Obs. Date (UT) Scan Number Scan_key Approximate RA (deg) Declination Range Affected (deg)

29270 north 2000-02-18 14 54806 217.65963 81.7 to 83.1

205438 south 2000-07-28 49 58256 298.12385 6.0 to 6.5

Tile	Observatory	Obs. Date (UT)	Scan Number	Scan_key	Approximate RA (deg)	Declination Range Affected (deg)
29270	north	2000-02-18	14	54806	217.65963	81.7 to 83.1
205438	south	2000-07-28	49	58256	298.12385	6.0 to 6.5

iv. Astrometric Considerations

The astrometric precision of 2MASS is outstanding. Comparison of PSC positions with those reported in the Tycho 2 and UCAC astrometric reference catalogs demonstrates accuracy of 70-80 mas with respect to the International Celestial Reference System (ICRS) over much of the brightness range for non-saturated sources.

The accuracy of position reconstruction will be slightly poorer near the declination ends of Survey Tiles, in regions with a low density of astrometric reference stars, and near the celestial poles where the telescope tracking was least stable. The degraded accuracy is reflected in the position uncertainties quoted in the PSC.

Focal plane distortion corrections were incorporated for point source position reconstruction, but were not used in the construction of 2MASS Atlas/Quicklook Images. Offsets ranging from 0.08" and 0.2" may exist between Catalog source positions and those measured near Tile edges (c.f. I.6.d.iv).

v. Differences Between the All Sky Release and Incremental Release Data Products

The 2MASS All Sky Release data products supersede those from the Incremental Data Releases

The All Sky Release products were constructed using the results of a complete, final reprocessing of all Survey data. The final processing software incorporated numerous improvements to the global photometric and astrometric calibrations, Atlas Image generation, and source measurement algorithms. Thus, the overall quality of data in the All Sky Release is believed to be superior to those in the Incremental Releases.

Differences in point and extended source extraction thresholds, handling of confused and blended sources, and the limits used to select sources for the release Catalogs mean that some sources in the Incremental Release Catalogs will not appear in the All Sky Release Catalogs, and vice-versa. This will be most apparent near the faint limits of the release Catalogs. For example, the All Sky Release PSC includes sources down to a signal to noise ratio limit >5 if they are detected in three bands. The Incremental Release PSC required sources to have SNR>7 in at least one detected band. The All Sky Release XSC contains fainter sources in the higher source density regions near the Galactic Plane than are found in the Incremental Releases because of improved handling of contamination by foreground stars in the final processing.

A number of scans of Tiles included in the Incremental Data Releases have been replaced in the All Sky Release with scans taken later in the Survey under better atmospheric conditions. The replacement scans will generally be more sensitive than the ones they replace, so may contain fainter sources.