Image quality flags

The image quality of each SDSS field is tracked through a number of flags and other quantities. These values are stored in the photoField files on SAS as well as in the field table in the CAS database. When using SDSS images or catalogs, these values can be useful in determining whether to use data from a particular field. This page describes these flags, some of which are bitmasks and others of which are not. The most important quantity for each field is score, a combination of the image quality parameters, described below.

As noted below, some of these flags are relevant for both the images and the catalogs, whereas some are flagging problems with the catalogs alone (that is, the corrected frames may still be of good quality).

Note that there are some small nomenclature differences between SAS and CAS for these values; in particular, any underscores in the names below are omitted in the CAS naming convention.

• image_status: This bitmask indicates any issues regarding the data-taking or telescope status.
  • bit 0: CLEAR (indicates clear conditions)
  • bit 1: CLOUDY (indicates cloudy conditions)
  • bit 2: UNKNOWN (indicates unknown conditions)
  • bit 3: BAD_ROTATOR (rotator error renders this field unusable)
  • bit 4: BAD_ASTROM (bad astrometry renders this field unusable)
  • bit 5: BAD_FOCUS (bad focus renders this field unusable)
  • bit 6: SHUTTERS (shutter error renders this field unusable)
  • bit 7: FF_PETALS (petals error renders this field unphotometric)
  • bit 8: DEAD_CCD (CCD error renders this field unphotomeric)
  • bit 9: NOISY_CCD (CCD noise renders this field unphotometric)
  If any bit other than 0 is set, it indicates a serious problem with the data. Neither the catalogs nor the corrected frames ought to be used (calib_status, as described below, takes this consideration into account).
• psp_status: This flag indicates any issues found by the Postage Stamp Pipeline (which makes some initial measurements of the PSF and other items before the photo code runs). In detail, the PSP sets these values on a per-band basis, and what is reported here is the maximum value (worst case) among the bands for each field.
  • 0: PSF_FIELD_OK (Everything OK)
  • 1: PSP_FIELD_PSF22 (Forced to take linear PSF across field)
  • 2: PSP_FIELD_PSF11 (Forced to take constant PSF across field)
  • 3: PSP_FIELD_PSF11 (Forced to take default PSF)
  • 4: PSF_FIELD_ABORTED (aborted processing)
  • 5: PSF_FIELD_MISSING (missing field)
  • 6: PSF_FIELD_OE_TRANSIENT (odd/even bias level transient present)
  In practice we have not released any runs that have fields flagged PSF_FIELD_ABORTED or PSF_FIELD_MISSING. One should not consider any fields with psp_status > 2 as photometric in all bands. (calib_status, as described below, takes this consideration into account).
• photo_status: Also known in some uncalibrated files as frames_status. This value indicates what happened when the photo code tried to run on this field. It can take the following values:
  • -1: UNKNOWN (indicating catalogs from the field ought not be used)
  • 0: OK
  • 1: ABORTED (indicating the resulting catalogs should be ignored)
  • 2: MISSING (indicating the resulting catalogs should be ignored)
  • 3: TOO_LONG (the software timed out, probably during deblending, indicating the resulting catalogs should be ignored)
  In practice, we have not released any runs with fields of UNKNOWN, ABORTED or MISSING status.
• calib_status: This bitmask (one for each band) indicates whether a field is well calibrated or not, which depends on the quality of the photometric reductions, the image conditions, and the availability of overlapping fields for ubercalibration. See the ubercalibration documentation for full details.
• skyflux: This measurement (one for each band) is the median sky level in each field, given in nanomaggies per square arcsecond. The sky level is obviously important in setting the photometric flux limit in each field. The median sky fluxes in the ugriz bands across all primary photometric fields in the survey are 1.5, 1.7, 4.2, 8.1, and 24.9 nanomaggies respectively.
• psfwidth: This measurement (one for each band) is the FWHM of the PSF in arcsec (as determined using a double-Gaussian fit). It is obviously important in setting the photometric flux limit for unresolved sources in each field. The median seeing for primary photometric fields in the survey in the r band is 1.3 arcsec, with a standard deviation of around 0.2 arcsec. Around 1% of fields have seeing worse than 2 arcsec.
• score: The information in all of these flags is combined into a single number, the score. This score is used in deciding which field is primary in any given location of sky, and it is described in full detail in the resolve documentation.

As noted above, for using the catalogs, the score (and thus all these other flags are of importance and must be checked. However, if you are using corrected frames, not all of the above are relevant. In particular, the photo_status quantity can be ignored; typically that indicates a very bright object is in the field, but not that the data itself is intrinsically bad (though parts of it might be highly saturated).

Copyright © 2010-2013 SDSS-III