However, the reliability of this diagnostic breaks down in fields with high source densities since confusion can make the background difficult to model. We have not applied this Hcn4 diagnostic to fields with source densities in excess of 4.2 (log of Ks souce counts per sq. degree) because of this confusion problem, since high Hcn4 values can just as easily crop up in scans with no airglow problems.
In Fig. 1 below, the scans currently falling within our airglow checks
are black (no airglow) and red (downgraded for airglow). The blue &
magenta point indicate scans at densities higher than those actively downgraded
for airglow.
Fig 1. Hcn(4) vs. source density diagnostic. Automatic airglow downgrades are only applied for densities < 4.2 since the coadd noise estimates begin to break down after this. At densities > 4.7 the Hcn(4) diagnostic blows up as the coadd noise estimates become completely dominated by source confusion.
As a result we have produced a list of ~70 scans which have likely airglow problems. They comprise only about 30% of the total number of scans inspected. Of these, about half have unquestionably bad airglow variations based soley on the erratic bkg plots. The other half are marginal but quite possibly bad. Representative bkg plots follow:
Fig. 2: Two sample "bad" aiglow bkg plots
Fig. 3: Two plots with "marginal" airglow bkg plots
Following are two sets of coadd images for scans picked arbitrarily from the list of bad and marginal tiles. Both show obvious airglow. While only 3 coadds are displayed in this quick priview, T. Jarrett has inspected both scans in their entirety and found significant airglow throughout.
Fig. 4a: "Moderate" airglow scan 990610s 056
Fig 4b: "Bad" airglow scan 990726s 007
We have also inspected 3 scans from 980624s including 2 "moderate" (110-111) and 1 "bad" (112) scans. All show airglow effects throughout, with the "bad" field consistently worse with sharp changes in background. The severity of airglow in all of the inspected scans correlates well with the new diagnostic discussed below.
Tile DateHemi Scan Kdens Hcn4
311055 980624s 112 4.25 11.1 324473 990620s 016 4.27 17.4 324475 990620s 018 4.32 4.7 322965 990626s 037 4.51 11.9 322966 990626s 038 4.5 8.7 322967 990626s 039 4.48 5.5 322985 990626s 069 4.31 6.2 322987 990626s 071 4.3 13 322988 990626s 072 4.28 5.7 322989 990626s 079 4.28 7.4 322990 990626s 080 4.3 9.3 322997 990626s 093 4.21 5.7 323022 990724s 050 4.23 5.7 205399 990724s 094 4.28 12.3 324444 990725s 020 4.26 4.6 324415 990726s 007 4.28 6 324416 990726s 008 4.3 16.1 324417 990726s 009 4.28 11 324418 990726s 010 4.29 12.1 324442 990726s 013 4.27 6.1 324446 990726s 014 4.26 12.1 324467 990726s 021 4.3 13.5 324478 990726s 022 4.29 9.7 202412 990801s 089 4.24 6.5 202413 990801s 090 4.23 10.6 202330 990807s 075 4.57 10.3 305308 990820s 014 4.67 42.3 21266 991121n 012 4.29 7.3 324381 000126s 103 4.35 5.2 11160 000422n 092 4.61 11 326065 000423s 028 4.63 20.2 322990 000611s 047 4.26 9 322991 000611s 048 4.27 6 302368 000611s 074 4.64 7.2 322990 000717s 021 4.29 4.9
Tile DateHemi Scan Kdens Hcn4 11110 970614n 098 4.28 5.3 11112 970614n 100 4.32 7.7 11167 970617n 072 4.59 9.9 11168 970617n 073 4.57 11 11187 970617n 104 4.35 5.7 308234 980416s 121 4.22 6.3 16496 980430n 085 4.33 7.3 311053 980624s 110 4.31 8 311054 980624s 111 4.28 7.3 305217 990426s 089 4.4 5.3 320840 990610s 056 4.53 5.9 320842 990610s 058 4.56 5.2 320844 990610s 060 4.59 5.9 320846 990610s 061 4.59 5 324495 990620s 051 4.32 6.7 324497 990620s 053 4.31 4.9 322976 990626s 054 4.41 5.2 322978 990626s 056 4.35 4.8 322979 990626s 057 4.38 6.8 322982 990626s 066 4.35 12.7 322983 990626s 067 4.34 5.2 322984 990626s 068 4.34 5 322986 990626s 070 4.31 4.8 322991 990626s 081 4.25 5.1 322992 990626s 082 4.24 6.2 322994 990626s 084 4.23 4.7 320962 990715s 086 4.5 5.6 205396 990721s 101 4.3 11.9 205397 990721s 102 4.28 8.4 205398 990721s 103 4.3 11.8 202409 990801s 086 4.32 4.6 205363 990804s 073 4.68 17.6 205366 990804s 076 4.62 14.7 202326 990807s 071 4.54 5.8 202328 990807s 073 4.54 9.6 202332 990807s 077 4.61 5.6 21295 991022n 017 4.27 5.6 21267 991121n 013 4.27 7.3 21268 991121n 014 4.29 7.9 21273 991121n 022 4.29 7.2 21275 991121n 024 4.27 7.5 326142 000502s 041 4.67 4.8 302367 000611s 073 4.65 18.5
In order to isolate H-band airglow problems using the Airglow Banding Diagnostic (ABD) values, we construct a new indicator of the form:
airglow banding diagnostic = 2(Hbl + Hbr) - (Jbl + Jbr + Kbl + Kbr)where the term represent the "l"eft and "r"ight banding counts for each band. Each term can be in the range of 0-23. Banding found evenly across all bands (i.e. not likely related to airglow) will have ABD = 0.
By subtracting the J & K banding contributions from the H contributions,
only H-strong banding seen in both halves will create large values.
The correlation between high ABD values and scans inspected for airglow
problems can be seen in the following histogram:
Fig. 5: Airglow banding diagnostic values for 3 populations of high density scans. Red bins are for scans with confirmed bad airglow, yellow bins show marginal evidence for airglow, and green bins are for scans clearly not suffering from airglow. The distribution exhibits consistency between the diagnostic and the visually-selected airglow sample.
This diagnostic is offered as food for thought on another possible
way to identify airglow problems. No statistics have yet been accumulated
for low density scans.