R. L. Hurt, J. D. Kirkpatrick (IPAC/Caltech), December 2000

These represent the current grading rules used for nightly QA evaluations. Older grading rules are archived elsewhere.

Survey Scan Grading

Each scan is scored using a base quality number multiplied by the minimum of a series of quality factors as detailed below. The "new" strategy of calibration is for observing a single calibration block at 1 hour intervals.

Base quality

fct1/Photometric quality

A photometric quality factor applies to each photometric interval during a night. It roughly corresponds to a probability in meeting Level 1 specs.

A photometric interval is defined as 2 or more cal sets spanning photometric survey sets. The standard photometric grading strategy follows the form:


A sensitivity quality factor is computed from a convolution of the seeing shape and background level (PSP or point-source probability), and corresponds roughly to the probability that a scan will meet the Level 1 specs for sensitivity. The grades are dependent upon hemisphere and date (reflecting differing sensitivities of detectors).
Probability North K PSP  (& 
H before 990701)
North H PSP
(after 990701)
South H PSP South K PSP fct2
>75% <= 10.85 <= 9.0 <= 9.6 <= 10.6 1.0
50-75% <= 11.11 <= 9.3 <= 9.8 <= 10.9 0.8
25-50% <= 11.35 <= 9.5 <= 10.3 <= 11.7 0.5
0-25% <= 11.85 <= 9.7 <= 11.7 <= 11.7 0.3
0% > 11.85 > 9.7* > 11.7 > 11.7 0.1

*Note that current Northern H PSP downgrades are only allowed to go down to 0.3, not 0.1, so this last condition does not apply.

fct3/Seeing Shape

Scans with poor seeing shapes are downgraded to fct3 = 0.1. Any of the following conditions warrants a downgrade: A downgrade override is also warranted if there is any indication that the seetracker score is in error and does not properly detect a known problem (i.e. many false extended sources seen across a coadd).

Note that in a related downgrade, a scan with a 2nd image moment < 0.81 receives a maximum final quality grade of 1.

fct4/Untracked Seeing

Scans with extensive runs of untracked seeing are downgraded to fct4 = 0.1.  The following conditions warrants a downgrade: All untracked seeing scans are inspected to verify the downgrade is warranted and not caused by a galaxy cluster.


Scans with strong residual structure in the coadds induced by OH airglow variations are downgraded to fct5 = 0.1. The following condition warrants a downgrade: All airglow scans are inspected to verify the downgrade is warranted . The cnoise4 statistic is the difference between the measured coadd background noise (after modelling large-scale gradients & structure) and the theoretical noise expected from the overall background level. Reviewers should note that this theoretical noise level varies with time and may need to be updated periodically by Tom Jarrett.

Other Downgrades

Individual scans should also be set to quality 0 for any of the following reasons: Single-frame electronics glitches (so far seen only in north H band) should not receive grades higher than 4 (however, there is currently no quality factor associated with this downgrade, but such cases will also be flagged with Bla0 = 1 in the SIT file).

Grades of scans may also be overridden on recommendation by UMASS (e.g. to force a single scan with a high grade to match the low grades of the rest of the block).

fct7/Untracked Seeing Eyeball Estimate

This final QA factor is carried along with the database transaction and SIT files for the night as a record of untracked seeing problems that have caused false extended source detections, even if the incident is not bad enough to warrant a downgrade. Reviewers inspect every scan with marginally high seetracker scores to look for obvious signs of point sources misidentified as extended. If such an incident is noted in at least two stars then the  fct7 score is set to 0.0 for the scan (otherwise it is 1.0). The default is for fct7 to be set equal to fct4.

Currently fct7 is not used for grading scans but may be a useful flag in selecting candidates for reobservation.

Calibration Scan Grading

Calibration scans receive grades of either 0 or 10. Conditions sufficient to fail a single calibration scan (grade 0) are: Conversely, any cal block that technically passes as photometric, even if it bounds non-photometric survey data or is not part of a solution spanning multiple blocks, should receive a grade of 10 for inclusion in the calibration working database. This is a change from previous rules in which cal blocks that bounded non-photometric data were themselves considered to be non-photometric.

Default Grading and Overrides

Much of the grading is computed automatically by the SCANSCI subsystem. This includes generation of the template QAreview and tile/database transaction files. If a default grade needs to be overridden the reviewer should enter that override into the YYMMDDH.qagrade file (found in the datequal directory) and use the /home/hurt/bin/run_scansci script to update the grading.

When overrides have been implemented they will show up on the second set of quality factors found on the "Scan Grading" section of the QA Report web pages.

Post-Processing Downgrades

A number of scans have been observed that do not overlap any adjacent tiles on the same night. To find potential photometric problems that would otherwise have been noticible in these overlaps, comparisons have been run between them and tiles from other nights to look for any photometric biases or evidence for clouds.

Scans in which clouds have been found are given retroactive downgrades to quality = 3, and they are also flagged with Bla1 = 1 in the SIT file.

Other scans have been found in which a clear, constant photometric bias is seen with respect to tiles on other nights. This indicates an incorrect zero-point determination, likely the result of weather-induced problems in these or nearby cal scans. If the bias is deemed sufficiently large to be a problem (~> 0.15 mag), the affected scans are flagged with Bla2 = 1 in the SIT file and their final quality scores are downgraded to 5 if they were higher than 5.

Last Updated 10/01/01 by R. Hurt