I. Introduction
6. Cautionary Notes
d. Atlas/Quicklook Images
2MASS Atlas Images derive from the combination of the six "Read 2-Read 1" frames which cover any given piece of the sky. The images are in FITS format, with full WCS information and photometric zero points in their headers. Please see Section IV.3 of the Explanatory Supplement to the 2MASS All Sky Data Release for a description of the procedure used to construct the Atlas Images.
~20:1 lossy-compressed version of the Atlas Images, known as "Quicklook" Images, are also available. These images are suitable for finding charts and visual inspection of the near-infrared sky. Because brightness information is lost during compression, especially at low signal levels, the Quicklook Images are not recommended for photometric measurements of sources or regions. Users should always defer to the Catalogs for photometry of point and extended sources, and use only the full-fidelity Atlas Images for direct measurements from the images.
For more information on the degradation of photometry due to compression on the Quicklook Images, please see Aperture Photometry from 2MASS Compressed/Uncompressed Images.
At certain pixel locations sky coverage is reduced from the nominal six apparitions by either noisy or bad pixels, cosmic rays, or meteor trails. If zero apparitions are available for a given pixel location, a zero flux value is inserted for this pixel. These pixels are apparent in the images, because the natural sky background in the frames is much larger than zero.
The Atlas Images preserve the observed background sky levels measured relative to camera dark frames with the shutter closed. This background is normally largest in the Ks band, although it can be even larger in the H-band, due to atmospheric OH airglow emission. The only background compensation that is made during Atlas Image construction is to adjust the frame backgrounds by a constant to produce seamless combined images. Because the OH airglow (especially at H-band) often contains structure on scales at or below the 2MASS frame size, the resulting Atlas Images show large background variations. Also, if the background is changing shape from frame to frame due to a bright star or a time-variant background, discontinuities may be seen at the frame edges.
Correction for optical distortion in the focal plane of the 2MASS cameras has been applied to the extracted source positions for optimal astrometric reconstruction. However, distortion corrections are not applied to the 2MASS Atlas/Quicklook Images. The effects of focal-plane distortion appear primarily in the east-west (cross-scan) direction in the Atlas Images. North-south distortion is minimal, because the combination of multiple frames offset in the in-scan direction averages down the distortion residuals. Offsets between Catalog source positions and positions measured off the images are zero near Image center line and increase towards the Image right ascension edges. The maximum distortion at the frame edges ranges from approximately 0.08´´ to 0.2´´; the greatest distortion is seen in the northern observatory H-band data. (See VI.6c).
An effort is made to remove transient "sources" such as intermittent noisy (or "hot") pixels, isolated cosmic ray hits, and meteor trails during the generation of the Atlas Images. For small, point-like transients, such as hot pixels and cosmic rays, this is accomplished by identifying point source detections above a specified SNR threshold seen on only a single frame, and masking them in the offending frame before combining the six frames. When the transient effect is confused with a source, it will not be identified as a single frame event (i.e., solo) and will remain in the Atlas Image.
Meteor trails are removed from single frames by identifying linear patterns in the solos identified in the frames and blanking the corresponding portion of the frame in all three bands prior to coaddition. This process allowed removal of most of the meteor trails from the images, but fainter trails and some anomalous cases were not identified by this technique, so users will occasionally see remnant trails on the frames either as solid streaks or broken streak segments. This process also serendipitously removed some artifacts due to bright stars, satellites, and other anomalies, but a number of these were incompletely blanked. Examples of incomplete blanking may be found in the gallery of Anomalies in Atlas/Quicklook Images.
The 2MASS Atlas and Quicklook Images can contain a number of artifacts produced by the optical and electronic effects of bright stars, and by transient phenomena such as meteor trails, aircraft, and insects crawling on the camera dewar windows. Compact artifacts that can be mistaken for point sources are marked on the Image overlays provided with the IRSA 2MASS Image Server. Extended and diffuse artifacts are usually apparent to the user when viewing full Atlas/Quicklook Images. As an aid to discriminating between real diffuse structure and artifacts, please see the gallery of Anomalies in Atlas/Quicklook Images.
The 2MASS Atlas and Quicklook Images are constructed with no attempt to correct or identify saturated data. The "Read 2-Read 1" frames from which the Atlas Images are constructed saturate at approximately J<9.0, H<8.5, and Ks<8.0. Any pixel in an Atlas Image brighter than a few thousand DN is potentially contaminated by saturation. At levels J<5.5, H<5.0, and Ks<4.5, the "Read 1" exposure also begins to saturate, and this will be apparent in the images as a depression in the center of the saturated image. The precise saturation limit varies, depending on the atmospheric seeing conditions and the sky background. Saturation is handled correctly during source photometry in pipeline processing. Therefore, users are once again advised to defer to the Catalogs for photometry of point and extended sources.
viii. Photometric Error Estimation Using the Atlas Images
The 1´´ pixel 2MASS Atlas Images are derived from 2´´ pixel image frames, as described in IV.3. When making photometric measurements directly on the Atlas Images, compensation must be made for both the correlated noise in the resampled pixels and the smoothing from the co-addition kernel or photometric uncertainties will be systematically overestimated. The appropriate noise estimation formulae for the images are given in VI.8.a.
[Last updated: 2006 May 4, by R. Cutri, M. Skrutskie and E. Kopan]
Previous page. Next page.
Return to Explanatory Supplement TOC Page.