VI. Analysis of Release Products

2. Sky Coverage

In this section we describe the depth-of-coverage of the Preliminary Data Release area and explain some of the issues with the pathologies of low coverage areas. For the purposes of this section, the term "Sky Coverage" refers generically to the topic of the WISE dataset spatial surveying features; when appropriate, the term "depth-of-coverage" is used to refer specifically to the number of observations at a specific point on the celestial sphere, and the term "Catalog source density" is used to refer to the number density of Catalog sources extracted in a small region.

The WISE survey strategy was designed to provide at least 8 frames of coverage on at least 99% of the sky in the 6-month minimum all-sky survey interval. This minimum coverage was required in order to achieve sensitivity limits, and includes coverages lost due to proximity to the Moon and the SAA, and further allows for recovery period in the event of satellite anomaly (although no such anomalies caused the loss of any data). As a result, the expected whole-mission data downlink produce an ab initio depth-of-coverage as shown in Figure 1; the cryogenic portion of the mission produces the coverage shown in Figure 2 for bands W1 and W2. Because of the gradual degradation of performance at cryogen loss, the depths-of-coverage of W3 and W4 vary slightly from that shown in Figure 2; these are shown in Figure 3 and Figure 4.

Figure 1 - During the entire WISE operations period, covering more than a year, a total of 2,784,184 framesets were taken and downloaded. The depth of coverage across the whole sky in Galactic coordinates shows the buildup near the ecliptic poles.	Figure 2 - The cryogenic portion of the mission, those framesets which will be included in the Final Data Release, lasted until cryogen depletion in October 2010. Up to that time, all W1 and W2 data cover the sky at a depth of more than 8 using 18,844,474 framesets.

Figure 3 - As WISE approached cryogen loss, the telescope began to warm up. W3 integration times were reduced to prevent saturation once the telescope had reached 45 K. The actual coverage at full integration time is somewhat shallower for W1 and W2.	Figure 4 - As WISE approached cryogen loss, the telescope began to warm up and the W4 detectors saturated. The actual coverage at 22 microns is therefore shallowest of all the WISE bands. Even so, the nominal survey coverage allowed 99.92% of the sky to be seen 8 times or more.

a. Preliminary Data Release Area

The Atlas and Source Catalog of the WISE Preliminary Data Release are drawn from data taken during the first 105 days of WISE survey observations, as described in V.2. Because of WISE's survey strategy, this region is oriented in ecliptic coordinates and covers great circles in the ranges of:

This totals approximately 23,658 deg², or 57.35% of the sky. On online tool is available to determine if a particular location or object is in the Preliminary Data Release area, and if so, provides the address of the Tile in which it is located.

WISE Preliminary Data Release Tile Lookup Service

The WISE Preliminary Data Release area is comprised of 10,464 Atlas Tiles, each Tile spanning 1.564° × 1.564° in 4095 × 4095 pixels at a resolution of 1.375′′ per pixel. These Tiles are delivered in FITS format image sets, consisting of an intensity image, the corresponding uncertainty image, and a depth-of-coverage map at each of the four WISE bands. The full sky is tessellated with a grid of 18,240 such Tiles on an equatorial projection for the purpose of combining the WISE single-exposure images and extracting final source information. The Tiles are distributed in 119 iso-declination bands with 238 Tiles on the celestial equator decreasing to six Tiles in the |δ|=89.35° declination band. Tiles are designed to overlap by 180′′ in RA and Dec on the equator, increasing in RA overlap towards the equatorial poles.

The ecliptic longitude boundaries of the Atlas and Catalog are pulled in from the original boundaries because of minimum coverage requirements. The Release does not cover the ecliptic poles because those Tiles are not fully contained in the survey boundaries from the first 105 days. The actual boundaries of the Atlas coverage area are not smooth in ecliptic longitude because the Atlas Tiles are laid out in an equatorial grid.

Figure 5 - Equatorial aitoff projection sky map showing the area covered by the WISE Preliminary Data Release. The colors encode the average number of single 7.7/8.8 sec WISE exposure frames covering 15´ × 15´ bins. The legend on the left gives the cumulative area in square degrees as a function of frame coverage depth. (Galactic and Ecliptic projections are available in I.4.a.ii).	Figure 6 - Differential area as a function of average frame depth-of-coverage in the Preliminary Release, computed in 15´ × 15´ bins.

Below is a full-sky Aitoff equa-area projection of the WISE Preliminary Data Release Atlas Images in equatorial coordinates. The slight differences in coverage at the four wavelengths can be seen, as can the strips of enhanced background due to the proximity of the Moon (seen as segments running slightly clockwise of vertical). The plane of the galaxy runs prominently throughout, as does a diffuse zodiacal light component.

Figure 7 - Aitoff equal-area projection of the WISE Preliminary Data Release intensity images, covering 23,658°2.

W1	W2
W4	W4
Figure 8 - Aitoff equal-area projection of the WISE Preliminary Data Release intensity images in each band.

b. Sky Coverage Statistics

WISE survey depth-of-coverage varies across the sky because of the survey scanning strategy, as described in III.4. There are typically 12 independent exposure frames contributing to each point on the sky near the ecliptic plane. The depth increases towards the ecliptic poles, reaching a maximum of ~260 frames at the highest ecliptic latitudes in the Preliminary Data Release (Figure 5). Visible in Figure 5 are some small patches with decreases in frame coverage caused by filtering out exposures considered to be of lower quality because of contamination by scattered moonlight (within 20° of the ecliptic), image quality degradation due to flight system motion, or other events. Pixel-level frame coverage information is provided in the WISE Image Atlas Depth-of-Coverage Maps. Here we present ensemble statistics of the coverage achieved for this data release.

Each Atlas FITS image contains in the header some high-level quantification of the depth-of-coverage for that Tile. This information may also be accessed in the Atlas Inventory metadata table. (Full depth-of-coverage information can readily be derived from the Atlas Depth-of-Coverage maps, for those who wish to determine this on a per-pixel level). The header keywords are listed in Table 1.

The median depth-of-coverage across the full Preliminary Data Release Area is 13.79 in W1, 13.78 in W2, 13.26 in W3, and 13.55 in W4. Below, we show more complete statistics of the depth-of-coverage distributions in this data release.

W1	W2
W3	W4
Figure 9 - Calculated depth-of-coverage vs. ecliptic latitude for each band; the yellow region indicates the dispersion of the distribution in each Tile.

W1	W2
W3	W4
Figure 10 - FITS header values of the median coverage in each band vs. ecliptic latitude; the yellow region indicates the minimum and maximum coverage in each Tile.

W1	W2
W3	W4
Figure 11 - Histogram of per-pixel depth-of-coverage in each Tile for each band for the entire Preliminary Data Release. Note that this is summed over Tiles, and so there is a slight overlap resulting in a double-counting of certain spatial pixels on the sky.

W1	W2
W3	W4
Figure 12 - Cumulative histogram of the area in square degrees resulting from integrating the curves in Figure 11, above.

W1	W2
W3	W4
Figure 13 - Same as Figure 12, but in a linear scale.

W1	W2
W3	W4
Figure 14 - The ordinate shows the percentage of the sky covered in the Preliminary Data Release Tiles to a depth of at least that indicated on the abscissa, for each band.

W1	W2
W3	W4
Figure 15 - Same as Figure 13, but in log-log scale.

c. Caveats About Low Coverage

The WISE scan strategy was designed to allow for repeat viewings of at least 8 times on every point on the sky, with accommodations allotted for planned survey motions and margins for unplanned survey interruptions. The achieved characteristic coverage is slightly higher than this because the survey was not interrupted. However, there are some regions that have anomalously low effective coverage in the Atlas Images and Source Catalog because some Single-exposure framesets were rejected from Multiframe processing due to poor assessed quality (i.e. see V.2). Some of the reasons for low coverage are summarized below.

i. Torque Rod Gashes

Comparison of the achieved Atlas tile coverage (Figures 5, 7, and 8) with the survey frame coverage (Figures 1-4, also via interactive comparison) illustrates several areas with a notable loss of coverage within the general boundaries of the release. Most significant are the horizontal bands at Ecliptic λ,β= 100°,+45° and 290°,-45° (Equatorial α,δ= 110°,+70° and 310°,-70°). Early in the survey, the spacecrafts' magnetic torque rods were enabled to dump accumulated momentum when scans approached within 45° of the ecliptic poles. Activating the torque rods resulted in a small jump in the telescope pointing and smearing of the resulting images. Because the smearing occurred near the same point on each orbit, and the smeared images were flagged as having degraded image quality in the QA process, low-coverage "holes" developed at those locations. Later in the survey (2010 May 02), torque rod enabling was staggered between 45, 57.5 and 70° latitude on alternating orbits so that any image smearing would not occur at the same point on the sky on each orbit.

ii. Moon Contamination

When WISE observes near the moon, stray light can contaminate images significantly enough that source detection sensitivity is degraded and spurious detections are triggered by the structured scatter light. Moreover, spatially-varying scattered light artifacts are problematic for the background-matching portion of the Multiframe pipeline image coaddition process.

The Moon crosses the scan circle twice a month. This would imply that a large amount of data would be corrupted; this would leave gaps in the sky coverage. To counter this, the WISE survey strategy uses a modified scan pattern where the scan circle gets slightly ahead before the Moon interferes and then drops slightly behind to recover the region the Moon obscured. The Moon moves 13° per day in ecliptic longitude, so with a 15° nominal exclusion zone (30° diameter) WISE needs to be 1.2° ahead just before the Moon crosses the scan circle, and then drops back to 1.2° behind just after the Moon crosses the scan circle. This "Moon avoidance" maneuver produces the "spokes" of enhanced coverage that are visible in the nominal sky coverage maps seen in Figures 1-4, for example.

Moon avoidance helps to fill in the coverage, but does not solve the problem fully because scattered light artifacts affected frames taken as far as ~30° away in W3 and W4, and ~20° in W1 and W2. (e.g. see II.4.a.ii). To minimize the impact of scattered moonlight in Single-exposure images on the coadded Atlas Images, frames suspected to be contaminated are flagged if they fall within the area of a static "moon-mask", and filtered out from the coadding if the spatially-varying portion of the moonlight produces a pixel RMS in excess of a threshold defined by frames not within the static moon-mask area, as described in IV.5.a.vi.

There are a few cases where most or even all of the available input frames touching parts of an Atlas Tile are within the masked region resulting in incomplete rejection of the scattered light artifacts, or, in the worst cases, zero-coverage holes in the Atlas Images and Catalog. An example of such an Atlas Image, 0333p181_aa11, is shown in Figures 16 and 17. Because the extent of scattered moonlight is larger in W3 and W4, the loss of coverage is more severe in those bands.

W1	W2
W3	W4
Figure 16 - Atlas Intensity Images for Tile 0333p181_aa11, showing residual scattered light and loss of coverage because of Moon contamination. The extent of scattered moonlight is larger at longer wavelengths, so the resulting loss of coverage is greater.

W1	W2
W3	W4
Figure 17 - Corresponding depth-of-coverage map for Atlas Image 0333p181_aa11 showing obvious Moon contamination; the depth-of-coverage ranges from zero near the upper left to ~12 towards the lower right.

To assist in the proper identification of moon-contaminated Tiles, there are FITS keywords in the Atlas Image headers that describe the Moon contamination mitigation process. These are detailed in Table 2. These parameters are also available in the Atlas Inventory metadata tables. As an example, for Tile 0333p181_aa11 in W1, the value of MOONINP indicates that 47 frames (out of NUMFRMS=64 initial frames) are flagged as "suspect" for moon-glow; of these, MOONREJ=45 were rejected, leaving only 19 frames to make up the final Tile. With such extreme ratios, care should be taken with the portions of the Tile nearest to the uncovered area. The WISE outlier detection relies on median absolute deviation as a robust measure of the dispersion of the pixel intensity distributions at any particular location. This technique becomes unreliable below a depth-of-coverage of five, so outliers may be incorrectly flagged and removed.

iii. Source Density Suppression Around Bright Stars

Bright stars have a noticeable impact on depth-of-coverage, but also a more subtle change in sky coverage as measured by the Catalog. Very bright stars effectively obscure background sources with their their scattered light halos and diffractions spikes, and they elevate the surrounding background, thus increasing the source detection limits. This can be illustrated visually in an extreme case for a very bright star, Betelgeuse. As a roughly -4^th mag star, Betelgeuse saturates thoroughly in each of the WISE bands, making it an easy example for the effect. A by-eye inspection reveals the suppression of source density in the vicinity of bright stars down to a magnitude of ~5, and it is likely that this effect would be statistically significant down to fainter levels.

We show in Figure 18 the intensity in W1 for a 54′×54′ region around Betelgeuse from Tile 0884p075. Overlaid on this are all the Catalog sources, including those flagged as being contaminated by diffraction spike objects, halo objects, etc. It is clear that the the distribution of sources is spatially nonuniform. One can infer that the source distribution is the result of a high density of potentially spurious sources within the halo area, a suppression of real sources in the halo near the bright star, and the more-or-less-uniformly distributed real sources at a greater distance from the star. The suppression of Catalog source density is a subtle effect of bright stars that can lead to, for example, incorrect two-point correlation function determination due to an unknown windowing. This lack of sky coverage is not reflected in the depth-of-coverage information provided in the Atlas Image sets.

The example in Figure 18 shows the distribution of all Catalog sources, but many of them are flagged to enable Catalog-query-based removal of potentially contaminated sources. If one removes these potentially contaminated and spurious sources by selecting only those with cc_flags="0000," the entire region is denuded of sources, as shown in Figure 19.

Figure 18 - Atlas Intensity Image in W1 for a 54′×54′ region around Betelgeuse from Tile 0884p075_aa11. Overlaid on this are all the Catalog sources, including those flagged as diffraction spike objects, halo objects, etc.	Figure 19 - Intensity in W1 for a 54′×54′ region around Betelgeuse from Tile 0884p075_aa11. Overlaid on this are only the fully reliable Catalog sources where cc_flags="0000." A ~1.5° region near Betelgeuse is effectively blanked out.

Refined selection criteria can be used to eliminate the spurious sources without filtering our all real objects. One way is to also use the multiple-band detection bit flag, det_bit, since spurious sources should not necessarily be associated with the same Catalog entry. Taking the above field and selecting according to:

results in an improved likelihood of the reliability of sources (Figure 20), which makes it more evident that the Catalog source density is low in the region of bright stars.

Figure 20 - Intensity in W1 for a 54′×54′ region around Betelgeuse from Tile 0884p075. Overlaid on this are selected Catalog sources chosen to be detected in W1 and W2, removing many spurious sources and highlighting the Catalog source density suppression near Betelgeuse and other bright stars.

As one final note, we mention that the depth-of-coverage is low in the immediate vicinity of bright stars. This is reflected in the Atlas Image Depth-of-Coverage Maps files and is visible in the depth-of-coverage histograms as the regions of anomalously low coverage. It should be noted there are also low depth-of-coverage groups of pixels associated with the ghosts and latents of bright stars, as is shown for W2, W3, and W4 in Figure 21. This shows the intensity and coverage images near Betelgeuse for bands W2 (top), W3 (middle), and W4 (bottom). The ~zero depth-of-coverage right at the position of the star is evident in all bands, as are the ~zero depth-of-coverage latent images in W3 and W4 that extend for several degrees.

Figure 21 - Intensity (left) and depth-of-coverage (right) images near Betelgeuse for bands W2 (top), W3 (middle), and W4 (bottom). The ~zero depth-of-coverage right at the position of the star is evident in all bands, as are the ~zero depth-of-coverage latent images in W3 and W4 that extend for several degrees.

A quantitative analysis of the Catalog source density suppression has been done for a large set of randomly-selected stars brighter than 9th magnitude. As the plots in Figure 22 show, there is a lack of sources near the bright star that suppresses other sources as a function of radius, with fainter stars suppressed to greater radii.

W1	W2
W3	W4
Figure 22 - Flux vs. separation

iv. High Flux/Source Density Regions

In regions with a large number of sources or particularly bright sources (either diffuse or point-like), the depth-of-coverage can vary between wavelengths in a spatially-varying way dependent on the intensity structure. Hence, performing tasks such as aperture photometry on extended sources must take into account these variations across the source. As an example of the coverage variations, the four-color composite intensity and depth-of-coverage maps for Tile 0450p605_aa11 are shown in Figure 23. There is a pronounced decrease in coverage (in magenta color) across a several arcminute-wide region just north of center and in the southeast in W3, but not in the other bands. Similarly, some regions to the east of center exhibit low coverage in W4 (cyan). These decreases are not severe; the total stretch depth-of-coverage is roughly 12 to 19, while the low coverage areas in those bands amount to a change in depth of only a few. What is interesting about the spatial variations is that the low depth-of-coverage is not coincident with the brightest regions in the relevant bands, nor in the region of highest flux density. Such decreased depths-of-coverage are to be anticipated and do feature in this Tile.

Intensity Image	Depth-of-Coverage Map (stretched)
Figure 23 - Four-color composite intensity and depth-of-coverage images for Atlas Tile 0450p605_aa11, showing the reduced coverage in regions with high background and around bright stars. Also note the pair of W4 latent images near the top center of the image.

Last update: 2011 April 26