I. Comparison of the 2MASS All-sky Release with the SDSS Early Data Release
1. Introduction
Mike Skrutskie has correlated point sources in the 2MASS All-sky Release with non-detections in the Sloan Digital Sky Survey Early Data Release. Taking the SDSS EDR as "truth", we can study the 2MASS-only objects to learn about the reliability of the 2MASS data.
Specifically, Mike took all objects in the 2MASS PSC with (23h28m < RA < 24h or 0h < RA < 3h40m) and (-1deg < DEC < +1deg) -- an area completely covered by both releases and representing 126 square degrees -- and looked for objects which had no corresponding SDSS source within a 1.5" radius. Total numbers of sources are shown in Table 1:
| TABLE 1: Statistics for the Coverage Area | |
| Note | Number |
| All 2MASS PSC entries in coverage area | 208766 |
| Those with no matching SDSS source | 4371 |
| Subtract those w/ a USNO-A or Tycho match | -1219 |
| Subtract those w/ mp_flg=1 | -191 |
| Subtract remainder w/ use_src=0 | -728 |
| Subtract remainder not having an 'A' in ph_qual | -1787 |
| Total number of "catalog quality" non-matches | 446 |
A large fraction (27.9%) of the 4371 SDSS non-matches comes from problems or shortcomings of the SDSS data themselves since these objects do have optical counterparts in USNO-A or Tycho. Another sizeable chunk (4.4%) are bona-fide 2MASS asteroid detections, which is not surprising since the survey area straddles the ecliptic (at the vernal equinox). Only 10.2% of the 4371 non-matches are "catalog quality" 2MASS sources, i.e., they have use_src=1 and have an 'A' in at least one column of their ph_qual flag. The breakdowns on these final 446 catalog-quality sources are given below.
2. "Catalog Quality" Non-matches
We have visually inspected all 446 of these "catalog quality" non-matches by painting the R-band DSS, R-band XDSS, and J-band 2MASS images side by side. In cases where no source was seen on any of these images, we also inspected the 2MASS H- and Ks-band images. The results of those inspections are given in Table 2:
| TABLE 2: Results of the Visual Inspection | ||
| Description | Number | Breakdown of ph_qual and cc_flg |
| a) Optical source found at position: | 140 | - |
| b) Single-band 2MASS source on images: | ||
| - J-band only sources | 3 | AUU 000 (3) * |
| - H-band only sources | 224 | UAU 000 (224) * |
| - Ks-band only sources | 6 | UUA 000 (6) * |
| c) Bright star chaff: | ||
| - Diffraction spikes | 21 | AUU c00 (20), UUA 00c (1) |
| - Real, confused sources | 3 | AAA ccd (1), AUU c00 (1), CBA ccc (1) |
| - Glints | 2 | AUU c00 (2) |
| - "Spray" artifact | 1 | AUU c00 |
| - Nothing visible | 6 | UUA 00c (2), UAU 0c0 (4) |
| d)Galaxy confusion: | ||
| - Piece of galaxy w/ gal_contam=2 | 6 | AUU c00 (5), UAU 0c0 (1) |
| - Real src w/ gal_contam=2 | 4 | AAA 000 (2), ABB 0c0 (1), ACD 000 (1) |
| - Real src w/o gal_contam=2 | 1 | AAB 000 |
| e)Persistence sources: | ||
| - Marked with 'p' in cc_flag | 12 | AUU p00 (11), UUA 00p (1) |
| - Not marked with 'p' | 3 | AAC 000 (2), BAC 000 (1) * |
| f) Line-128 "stripe" artifacts: | ||
| - Unflagged "star echo" | 3 | AUU 000 (2), UUA 000 (1) * |
| - Flagged as source on stripe | 1 | UUA 00s |
| g) Real 2MASS-only detections?: | ||
| - Confirmed uncataloged asteroids | 2 | ABD 000 (1), AAD 000 (1) |
| - Suspected uncataloged asteroids | 8 | AAA 000 (2), AAA ccc (1), AAC 000 (1) |
| ABD 000 (1), ABU 000 (2), ACU 000 (1) | ||
| Total: | 446 | |
a. Optical sources
For 140 of the sources, there was an obvious optical DSS or XDSS source at the 2MASS position. In most cases, these were objects buried in the extended haloes around bright stars or were close double stars. Both of these cases cause problems with the digitizations of the photographic material (hence no USNO-A match). The depth of the SDSS survey means that close doubles and fainter objects in bright star haloes are much more difficult to disentangle as well (hence no SDSS match).
b. No obvious 2MASS sources
Most of these sources were H-band only 2MASS detections and appeared spurious. There are 140659 2MASS sources (out of the 208766 total from Table 1) that have use_src = '1' and ph_qual LIKE '%A%', so these 224 H-band only sources represent 0.16% of the "catalog quality" 2MASS sources in this region. This is obviously of concern since our specifications are for a 99.95% reliability at high galactic latitudes. In addition to the 224 H-band only sources, 6 Ks-only and 3 J-only sources were also found.
c. Bright star chaff
Thirty-three sources were found to lie very close to a bright star in the 2MASS images. Three of the 33 appeared by eye to be real sources confused with the bright star, but the remainder appeared to be artifacts. (See Table 2 for details.) As noted in Table 2, all 33 sources were labelled with a 'c' in the cc_flag field and are thus already flagged as suspicious.
d. Galaxy confusion
Eleven sources appeared to be confused with a galaxy on the 2MASS images. All but one of these was marked as contaminated by a galaxy (gal_contam=2). The final source was cleanly separated from the galaxy in 2MASS even though SDSS (and the USNO-A) did not have the same success.
e. Persistence sources
Fifteen sources were actually persistence artifacts. Three of these were not marked with a 'p' in the cc_flg field. Two of these unmarked persistence artifacts -- 2MASS J01464247+0006121 and 2MASS J01464247+0007344 -- are from the same bright star 2MASS J01464248+0011413 (J mag = 7.32), which falls in the first few frames of 980923s s094 but not in the region that contains 6-deep frame coverage. Tracey is investigating why the persistence artifacts failed to be flagged in this case.
f. Line-128 "stripe" artifacts
Four sources were found to be spurious detections along the line-128 stripes. Only one of these was marked with an 's' in the cc_flag field. The other three were star echoes. The two unmarked J-only echoes had J-band PSF chi-squared values of 2.38 and 8.19, and the Ks-only echo had a Ks-band PSF chi_squared value of 2.41.
g. Real 2MASS-only detections?
The final ten sources appeared by eye to be real 2MASS sources with no optical counterparts. Seven of these were three-band sources and three were two-band. Because all sources lie along the ecliptic, we believe that many of these will turn out to be uncatalogued asteroids. Table 3 lists two objects whose fields were imaged twice by 2MASS:
| TABLE 3: Objects that are probably uncataloged asteroids | |||
| Designation | Image showing no source at posn | Image from which source was drawn | Conclusion |
| 2MASS J01490530-0058490 | 980923s s104 i056 | 980923s s105 i221 | asteroid |
| 2MASS J23500291+0003569 | 001120n s038 i009 | 980925s s083 i267 | asteroid? |
2MASS J0149-0058 is presumably an uncatalogued asteroid as it moved 1.4" between the two consecutive scans. 2MASS J2350+0003 is probably also an uncatalogued asteroid since it appears as a real source on 980925s s083 but is missing from the scan of the same region acquired 26 months later.
The other eight objects are listed in Table 4. All have been imaged at z-band by Adam Burgasser on 2002 Nov 18 (UT) and 2003 Jan 25 (UT) from the Palomar 60-inch telescope and results are given in the table. Postscript finder charts are available by clicking on the source name:
| TABLE 4: Follow-up of the 2MASS-only detections | |||||
| Source Designation | J mag | H mag | Ks mag | ph_qual | Note |
| 2MASS J00092257+0037558 | 15.110+/-0.041 | 14.434+/-0.051 | 14.117+/-0.045 | AAA | seen at r and z; colors indicate bkg star; missed by SDSS because object lies in wings of much brighter star |
| 2MASS J00102857+0013397 | 14.544+/-0.035 | 14.328+/-0.067 | 14.144+/-0.050 | AAA | not seen at z; likely asteroid |
| 2MASS J00342321-0037257 | 16.523+/-0.092 | 16.095+/-0.164 | (not detected) | ACU | not seen at z; likely asteroid |
| 2MASS J00413096-0027299 | 15.632+/-0.058 | 14.659+/-0.048 | 14.600+/-0.083 | AAA | not seen at z; likely asteroid |
| 2MASS J01091776-0008469 | 16.237+/-0.083 | 15.875+/-0.139 | (not detected) | ABU | not seen at z; likely asteroid |
| 2MASS J01564569-0042566 | 16.179+/-0.091 | 15.757+/-0.152 | 15.587+/-0.212 | ABD | not seen at z; likely asteroid |
| 2MASS J02055159-0029080 | 16.198+/-0.073 | 15.815+/-0.132 | (not detected) | ABU | not seen at z; likely asteroid |
| 2MASS J23585570+0026538 | 16.150+/-0.086 | 15.607+/-0.099 | 15.599+/-0.201 | AAC | not seen at z; likely asteroid |
3. Conclusions
A summary of problems is given in Table 5. This analysis has shown that the H-band only sources seriously compromise our level-1 specifications for catalog reliability. Several other flavors of problem sources were also uncovered, but there are so few of these that they do not jeopardize our reliability requirements. Specifically, there are only 15 of these, or 0.011% of the total number of catalog quality sources, which is well below the 0.050% threshold for reliability:
| TABLE 5: Types of sources that impact reliability | |||
| Problem | Number | Percentage of all catalog quality sources | Note |
| J-band only sources | 3 | 0.002% | ok |
| H-band only sources | 224 | 0.162% | needs corrective action |
| Ks-band only sources | 6 | 0.004% | ok |
| Unflagged persistence sources | 3 | 0.002% | ok |
| Unflagged striped artifacts | 3 | 0.002% | ok |
Addendum
Roc Cutri's recipe for removing sources caused by a couple of known H-band hot pixels has led to a dramatic improvement of the numbers quoted above. The number of H-band only sources under part (b) of Table 2 has been reduced from 224 to just 12, and the number of "Nothing visible" sources under part (c) of the same table has been reduced from 6 to 3.
The bottom line is that the number of spurious sources in all five categories listed in Table 5 has been reduced from 239 to only 27. The reliability for the 2MASS All-Sky PSC, as measured by the comparison to SDSS EDR data, thus stands at 99.981% (27 bad sources out of 140659), well within the level-1 specification of 99.950%.
[Last Updated: 2003 Jan 27; Davy Kirkpatrick, Brant Nelson]