2MASS Final Catalog Scan Selection -
Sky Coverage Iteration

---

---

1. Introduction

First-pass scan selection for the final 2MASS Release is made by selecting a scan of each unique tile with the default quality score from processing quality assurance. This results in a preliminary list of 59,975 scans for the release:

Preliminary Scan List for Final Release - This is not the final scan selection list!

This list includes 325 gap-fill scans which may or may not be useful for the final list. The columns in this table are as follows:

• Atile - Adjusted tile number (north: Atile = tile; south Atile = tile - 200000)
• Qb - Best quality score achieved for this tile
• No - Number of observations of this tile at any quality in latest SIT file
• Nb - Number of scans of this tile with quality = Qb
• date - UT observation date of scan with highest NRS
• h - hemisphere of scan with highest NRS
• scn - Scan number of observation with highest NRS

2. Sky Coverage Iteration

2.1 Preliminary scan list

Laurent Cambresy's scan overlap analysis run on the final processing Working DB revealed 10 real gaps in survey sky coverage, where no acceptable scan covered the sky. Laurent Cambresy has repeated the scan overlap analysis restricting it to the preliminary scan list. Because the criteria for Catalog source selection include a boundary of 10" for point sources and 15" for extended sources, there will be effective sky coverage gaps in the Catalog if scan pairs overlap by less than 20" and 30" for point and extended sources, respectively.

There are 127 scan pairs with <30" RA overlap and 90 scan pairs with <30" DEC overlap in the preliminary scan list, including gap-fill tiles. The following two lists give the Scans with gaps in RA and DEC, respectively, in decreasing order of overlap. Each line in the tables is referenced to a "primary" scan of a Tile taken from the preliminary scan list, and "secondary" tiles that overlap the primary on either the eastern edge (RA) or southern edge (Dec).

The columns in these tables are as follows:

H/Date/Scn - Hemisphere, date and scan of the primary Tile.

Atile - Adjusted Tile number of the primary scan.

dx1/dx2 - RA overlaps (in arcsec) between the primary and secondary Tiles at the top and bottom corners of the scans. Since Tiles converge towards the poles, gaps may exist only for part of the scan.

dy1/dy2/dy3 - DEC overlaps (in arcsec) between primary and secondary Tiles at the east, center and west corners of the scans. Values of -999 indicate gaps of >999".

Atile1/Atile2/Atile3 - Adjusted Tile numbers of overlapping tiles. Atile1=Atile2 for the RA overlaps. There can be up to three different overlapping tiles considered for the DEC overlaps.

s/l - Flag indicating of there are alternate scans of the primary and secondary Tiles having the same quality or lower quality than the scans selected in the preliminary scan list.

Alternates - Alternate scans of the primary and/or secondary Tiles in the cases where there there are alternates with the same quality as the scans in the preliminary scan list. There are sometimes >1 alternate per Tile.

This analysis always finds the best overlaps available, so when the overlapping Tile is shown to be a gap-file Tile (80000 < Atile < 90000 or 180000 < Atile < 190000), the gap-fill is better than the available normal survey scans. In practive, every RA gap should also produce a DEC gap. However, the associated DEC gap will only be captured in this analysis if it falls in the corners or RA-midpoint of a Tile. This happens for only one RA gap. Thus, it is necessary to carry both the RA and DEC gap lists.

The largest linear gaps are in declination with the worst being the gap south of Tile 108268 (survey Tile 308268) The pointing for the scan of this tile was known to be off nearly 0.8 deg in DEC. There are three alternative scans of this Tile, but all have lower quality. Another large DEC gap is in the south-center of Tile 10812. This Tile lies below Tiles 7965 and 7966 which have an RA gap between them - hence the large DEC gap.

2.2 First Iteration

There are alternate scans for 17 of the 127 scan pairs with RA gaps, and 28 of the 90 scan pairs with DEC gaps. For this pass, we consider alternate scans of Tiles only when the quality of the alternate is equal to that of the original scan selected for the release. Laurent examined the resulting overlaps using the alternates and generated the following summary tables. The code "OK" indicates that the alternate scan produces overlaps >30" and therefore fills the effective gaps.

RA Alternates
DEC Alternates

All 17 of the RA alternates produce improvements in the sky coverage, with the gaps being filled in 12 cases and improved-but-not filled in 5 cases. The improvement is minimal in the cases of Tiles 29571, 29586, 26792 and 26794. The alternate for Tile 29602 improves the eastern gap but actually creates a slightly smaller RA gap on the western edge.

25 of the 28 DEC alternate scans effectively filled the gaps. The 000423n scan 87 alternate for Tile 7987 produces only 1" better coverage, so is a negligible improvement for this real gap. In the other three cases, the alternate scans do not fill the gaps since they are alternates of scans that do not contribute to gaps.

The worst DEC gap, associated with primary tile 108268, is a special case since it is so large. There are three alternate scans of this tile, all with quality 3 (the originally selected scan has quality 10). All three of the lower quality scans of this tile are positioned properly and fill the gap. For this case only, we recommend replacing the default selected scan with one of the lower quality scans: 001110n s014, 001114n s014 or 001123n s014.

The net result of the first iteration is that if the alternate scans described above are selected, there will be 115 RA gaps and 65 DEC gaps in the final scan selection set. The total area of these gaps is ~1.5 deg².

2.3 Actions after 3/26/02 Science Team Telecon

The preference of the Team was to not introduce any lower quality data to fill gaps at this time. We will work this in parallel as long as it does not delay the main FPG tasks, and there may be an option to add the low quality fillers immediately before the final catalog draw.

The first iteration actions are as follows:

• Start with the preliminary scan selection list dated 3/13/02 containing 59975 scans.



• Scans of 41 Tiles were replaced by alternate scans having the same quality to improve sky coverage where the RA or DEC overlap with adjacent tiles was <30".



• 11 "obsolete" scans of survey Tiles were removed where a gap-fill scan was used to replace a regular survey scan. In 6 of those cases, the gap fill scan had Q=10 and replaced a Q=1 survey scan. In no cases did the gap fill scan have lower quality than the survey tile. The following table contains a list of the obselete Tiles, and their best quality, followed by the replacement gap-fill Tiles and their quality. They are all northern Tiles:




   Survey Tile  Qb  Gap Tile  Qb    
          7961  10     89007  10
          7963  10     89008  10
         10831   1     89014  10
         10850   1     89015  10
         10878   1     89017  10
         28606   1     89036  10
         27552  10     89550  10
         27554  10     89551  10
         28608  10     89529  10
         28569  10     89561  10
         29328  10     89570  10




• 233 gap-fill scans that do not actually fill gaps or improve sky coverage were removed. A list of the Tile numbers for these scans is given here.

This leaves a list of 59731 scans for the release as a result of Iteration 1. This list is given below and has the same format as the Preliminary SCan List given at the top of this page. An additional column flagging the 41 tiles with scans replaced for sky covereage reasons has been added. For these tiles, the 8th column will contain "o". If there was no scan replacement, the 8th column is blank. If all Tiles had been covered perfectly, there would be 59650 scans in the release.

Iteration 1 Scan List for Final Release

2.4 Analysis on the First Iteration Selection

• R. Stiening - 3/27/02:

I agree that your list has 59731 scans.  There are 41 where the alternate
was chosen "o".  The adj_tile entries are distinct. (No repeats of the
same value.)  24,946 scans were observed at Mt. Hopkins. 34,785 were
observed from CTIO. (with a year less observing as well)

The working survey database has 1,116,583,386 rows in the selected scans.

• Laurent repeated the overlap analysis for the Iteration 1 scan list and found no new gaps. This confirms that no tiles were accidentally omitted from the list. The new lists of known scan pairs with <30" RA and DEC overlaps are listed below. As before, negative overlaps indicate real gaps between the sky coverage of scans. Overlaps <20" will produce effective gaps in the point source coverage; there are 46 such gaps in RA and 7 gaps in DEC. Overlaps <30" will produce effective gaps in the extended source coverage; there are 66 such gaps in RA and 45 gaps in DEC.

Tiles with <30" RA overlap
Tiles with <30" DEC overlap



• Figure showing the remaining gap positions in galactic coordinates. Large diamonds : overlap <0''; medium diamonds : overlap <20''; small diamonds : overlap <30''. The overlap size is taken from the edge where it is the worst. However, it is not a good number to characterize triangular polar overlaps. (L. Cambresy).



• The effective area of the gaps in the final release for different levels of overlap are:

Overlap	Area(RA)	Area(Dec)	Total (deg2)
<30"	1.579	0.068	1.648
<20"	1.194	0.064	1.258
<0"	0.646	0.060	0.706

The RA gap also detected in Dec was removed, and the correct value for the biggest Dec gap was used (1150" instead of 999").

There may be other small gaps resulting from scan wander that will not necessarily be diagnosed by this analysis.



• R. Stiening - 4/6/02:

I have analyzed your new selection of 59,731 scans.  The number of order 
11 htm triangles that are empty has decreased from 70 to 66.  So this scan 
selection is somewhat better than its predecessor.

However, the large hole at ra=287.9, dec=-17.8 remains.  I attached the 
table of empty triangles for you to review.

Rae

NB: Each triangle covers from .000965 to 0.00203 square degrees.

3. Possible Ways to Minimize Remaining Gaps

1. Selectively filling gaps with lower quality scans. Cost is additional work to minimize effective area impacted, and testing modified software. Risk is primarily to the schedule schedule, and secondarily the introduction of very low quality data into piecemeal areas. This option may be pursued in parallel with normal duplicate resolution processing, only if it does not impact schedule. If time permits, will analyze results and consider filling in gaps as possible during final catalog construction.



2. Eliminating 10"/15" "safety" boundaries in source selection. Duplicate source resolution removes most edge sources. Cost is additional runtime for duplicate source resolution, DBMAPCOR and Galaxy analysis. Risk is associated with allowing small number of non-3-band coverage sources at unbounded scan edges next to real gaps into the Catalog. However, these sources can be immediately identified using the existing dist_edge_?? parameters in the source records. This option was discussed during 3/26/02 Science Team. Team was not in favor of pursuing for fear of introducing problems with edge sources sneaking into catalog.

---

Last Updated: 6 April 2002
R. Cutri & L. Cambresy - IPAC