Redshift Quality Codes

Q=4

Rock-solid z value with at least two clear features. Distinctly double [OII] 3727Å counts as two features.

Q=3

Good z value, probability > 90% that it is correct. At least two features still required, but ID is more doubtful than Q=4. Covers instances in which [OII] 3727Å is fuzzy and merged; identification of [OII] 3727Å is very likely, but not 100% sure. Also covers cases when one (or both) of the two required features has lower S/N.

Q=2

Not usable for the redshift catalog at this time, but SPECTRUM SHOULD BE REVISITED, either because a redshift MIGHT be possible with further looking, or because there is a bad reduction that might be fixed up. All Q=2 cases should have a comment flag. The most common reasons for Q=2 include:

• Marginal S/N. The signal-to-noise ratio (S/N) is MARGINAL and a redshift might be extractable with more work, but that is for a second pass. Comment: marg
• One feature only. ONLY ONE feature is visible. Use Q=2 for ALL such cases. Further info such as photo z's may be able to decide between leading candidate z's (usually Hbeta vs. Halpha), but do not try to use astrophysical logic to decide this at this time! Comment: sngl
• Superimposed serendip. Will need to be revisited and disentangled by hand. Comment: supser
• Poor reduction. The reduction or extraction is so bad that the redshift is presently no good for use in the catalog (estim. error more than 100 kms; if estimated error < 100 kms, then Q=3 or 4). Note: the error is an eyeball estimate of the error, not the program's. This estimate should take into account systematic errors (such as an offset extraction window) that the program does not know about. Comments: use bext, bs, bcont to give reason why bad; add fix if problem can be fixed with an improved reduction. Note: there are also cases in which a redshift has been ruined by a "catastrophically" bad reduction. This is case Q=-2. See below under Q=-2 for rule to distinguish between Q=2/bad-reduction (this case here) and Q=-2.
• Forced z. A forced redshift (z) is one where the program could not find a z by cross-correlation, even in z-refit, and a redshift was entered by hand. Comment: z=xx

Q=1

The redshift is not usable and the data will never yield one. The data are of such poor intrinsic quality that a redshift is hopeless. Q=1 means that the data will never be looked at again for any reason (as far as we can foresee now). The most frequent reason for Q=1 is low S/N. Note: if the redshift has been ruined by a bad reduction and there is some hope of rehabilitation through re-reduction, use Q=2. If the redshift has been ruined by a "catastrophically" bad reduction, oftimes unfixable, use Q=-2. See further discussion below on difference between Q=2 and Q=-2.

Q=0

The default value. Means spectrum has not yet been examined.

Q=-1

Star

Q=-2

The reduction and/or instrumental effects are so bad that, no matter how bright the object is, we cannot obtain a redshift for it; effectively, this object was never observed. Such cases can ONLY be diagnosed by checking the 2d spectrum. Q=-2 is needed to keep proper statistics on successful vs. failed redshifts. The division between cases Q=-2 and cases Q=2/bad-reduction (above) is clearly fuzzy, but a helpful list of guidelines for Q=-2 has been written by Jeff Newman. The main criterion is catastrophic vs. minor problems in the reduction: if it's catastrophic, use Q = -2; if not catastrophic, use Q=2. "Catastrophic" means that we are prevented from getting any significant shot at deriving a redshift, and this would have been true no matter how bright the object was. Most of the reasons in Jeff's list for Q=-2 are NOT FIXABLE through better reductions, though occasionally one might be. Even though the division might not be perfect, the fixable vs. non-fixable criterion might be a useful guideline to decide between Q=-2 and Q=2/bad-reduction, in addition to the basic notion that any galaxy would have been wiped out regardless of magnitude.

See Also

Comment Flags

Explanation of comment codes which can be assigned to spectra.