Redshift Quality

The redshift quality rating is the most important field in visual inspection. It indicates your confidence that the redshift measurement is correct, enabling downstream science to appropriately weight or filter the catalog.

The Five Quality Levels#

Quality	Color	Confidence	When to Use
Not Inspected	Gray	N/A	Default state, no one has reviewed this object
Impossible	Red	0%	Impossible to determine a redshift from available data
Tentative	Orange	~50%	Uncertain but feasible
Probable	Yellow	~80%	Likely correct, but some doubt remains
Secure	Green	>95%	Highly confident, multiple features confirm

Detailed Guidance#

Not Inspected (Gray)#

This is the default state for all objects before visual inspection. You should never explicitly select this (and in fact, you can't); instead, always assign one of the other four levels.

Impossible (Red)#

Use this when you cannot determine any redshift from the available data. Common reasons:

No spectral features - Featureless continuum or pure noise
Severe contamination - Another source dominates the extraction
Data quality issues - Chip gaps, stuck shutters, or other problems affecting key wavelengths

Note: "Impossible" means you can't measure a redshift, not that the automated fit is wrong. If you can identify features but they suggest a different redshift, override the value and mark as Tentative or higher.

Tentative (Orange, ~50% confidence)#

Use this when you have some evidence for the redshift but significant uncertainty remains, e.g. single emission line, very weak emission features, or an ambiguous spectral break.

Example of a tentative spectrum: only has G395M observations, with a single emission line identified as Hα, but spec-z matches photo-z.

Probable (Yellow, ~80% confidence)#

Use this when the redshift is likely correct but you have some remaining doubt:

Single line at high SNR with other supporting evidence - e.g., broad line, continuum shape, robust photoz
Strong break with some confirmation - Lyman break with weak emission
Multiple weak lines - Several features align but individually marginal
Reasonable but not perfect fit - Need to followup with a more careful interactive spectral fit

Example of a probable spectrum: multiple low-SNR lines consistent with [OIII] and Hα, though inconsistent with photo-z.

Secure (Green, >95% confidence)#

Use this when you are highly confident in the redshift:

Multiple emission lines - Two or more clearly identified lines at consistent redshift
Strong break with confirmation - Clear Lyman/Balmer break plus emission lines
Absorption features match - Stellar absorption lines at consistent redshift

Example of a secure spectrum: many high-SNR emission lines

Overriding the Automated Redshift#

If you identify a different redshift than the automated fit, enter that value in the "Override" field. You'll have to also select a quailty rating. It may be useful to add a comment explaining why you applied a manual override.