Spectroscopic Caveats

Incorrectly-labeled Hydrogen Balmer line

In SDSS spectra released in DR8 – DR13, the Balmer Series line Hζ (H-zeta, 3889.049 Å) was incorrectly labeled as Hε (H-epsilon, 3970.072 Å), and the real Hε was not included in the analysis of spectral lines. This affects line measurements tabulated in spZline files. These files are only available on the SAS. These measurements are not loaded into any version of the CAS.

SAS-only plates

If one browses the directory trees containing all of the spectra (see the spectroscopic data access page) one will find files associated with a certain number of plates not listed in the DR8 list of plates and not loaded into CAS. In essentially all cases, it is best to ignore such files and plates. For DR8, we went through some effort to include all reasonably good plate observations; any plate observations found on SAS but not in CAS are likely to be disastrously bad.

Note on stellar radial velocities

The standard redshift z from idlspec2d is available unaltered in the specObj and sppParams tables. These redshifts, primarily for galaxy work, have no offsets or corrections applied.

For stars, a better redshift to use is the ELODIE-matched template redshifts, stored as elodie_z in the specObj file and the specObj table in CAS. The CAS also records this as the quantity elodierv in the sppParams table, but with a correction term:

   elodierv = c*elodie_z+7.3 km/s

The 7.3 km/s is an empirically derived offset putting the elodierv of all stars on a system consistent with that of other literature measures of known radial velocity standards.

Redshift status

The quality flags for the redshift fitting procedure is stored in the zwarning bit mask. Most redshift warnings indicate a likely substantial problem with the data. An exception is MANY_OUTLIERS, which flags when many pixels are poorly explained in a statistical sense by the best-fit redshift model. This bit is typically set for very high signal-to-noise ratio stars (where errors are small, so χ² is high), or galaxies with broad lines (the redshift fitting model includes only narrow lines); in such cases, the redshift is usually fine.

About 2% of non-sky spectra have some warning set other than MANY_OUTLIERS. The redshifts of the remainder are virtually always correct. Many of the spectra flagged with problems also have correct redshifts and classifications, but we recommend care before using them.

Note that this set of flags to check has changed since DR7.

Galactic extinction correction

The spectra released in DR8 has not been corrected for Galactic extinction, because the SDSS includes a substantial number of spectra of Milky Way stars whose extinction would differ from that given in the dust maps. This policy has been the standard since DR2; in the EDR and DR1, the spectroscopic data were corrected for galactic extinction. The extinction is a relatively small effect, since the median E(B-V) over the survey is around 0.04; however, for some SEGUE pointings the reddening can be substantially larger.

Night sky emission lines

The night sky emission lines at 5577Å (when there is auroral activity) at 6300Å, 6363Å, and in the OH forest in the red can be very strong, and leave significant residuals in the spectra whose amplitude is occasionally underestimated by the noise model. Be cautious about interpreting the reality of weak features close to these lines.

Not-quite-perfect plates

A small number of plates suffered from a variety of minor problems. For plates that we deem that the data is unreliable, they have had their platequality set to bad, and some terse comments put into the qualityComments status.

Plates with comments about collimation problems refer to hardware problem causing a mismatch between the flatfields and the science exposure instrumental profile shapes, in both the spatial and wavelength directions. This problem caused the optimal extraction process to reject an excessive number of pixels. This problem was fixed in software, and comparing overlapping objects from adjacent plates confirms that the redshifts from these problematic plates are unbiased. However, the spectra themselves should not be used for precision work or spectrophotometry.
Plates in the apbias program used multiple, very slightly offset pointings, but the reductions do not properly combine them. They should have valid redshifts in these spectra, but the spectrophotometry will be very inaccurate.
For some plates the software had issues with rejecting cosmic rays, because there was only a single exposure to work with. These are all marked as bad plates (though again in many cases the redshifts and spectrophotometry are fine, except for the cosmic rays).
Plates located in regions with extended diffuse emission (like in Orion or Taurus) often have sky-subtraction errors and issues, because there is no truly blank sky available. In these cases, the emission lines from the nebula are partially, but not wholly. subtracted and hard to interpret.
Because of time-variability in the dichroic throughput, occasionally the spectrophotometry has "kinks" at the transition between the red and blue spectrographs; we have identified some, though perhaps not all, of the worst cases of these.
Occasionally the second spectrograph electronics caused serious issues for fibers 321 through 640.
One plate had substantial contamination from Pollux because of light scattered through clouds.
A number of other plates are simply low signal-to-noise ratio for a variety of reasons, but because they were special plates, needed to have their quality values set by hand. That is, they targeted deeper than we normally do, and so would have passed the surveys signal-to-noise criteria at the standard magnitude for which it is defined.

Mismatches between spectra and photometric data

There are occasional "mismatches" between the spectra and the photometry, both due to problems on the spectroscopic side in identifying the location associated with every fiber, and due to problems on the photometric side in finding an associated photometric object given a location.

With some frequency, the fiber mapping failed which identifies which fiber has been plugged into which hole. There are around 7200 such cases in DR8, which are marked as UNPLUGGED in the ZWARNING bitmask. The vast majority of these cases occur because the fiber was actually not plugged or was broken (in such cases, essentially no signal is detected in the fiber, and snMedian is reported as zero). In around 200 cases, there is measureable signal down the fiber. In cases where there is more than one such fiber on plate, there is a possibility that the fiber location associated with the spectrum is incorrected (and thus that the photometric and spectroscopic information is mismatched). This problem occurs for around 70 objects in the survey.

Other mismatches can occur due to problems in the photometry. Errors in the deblending algorithm in the target reductions caused spectroscopy to be carried out occasionally on non-existent objects (e.g., diffraction spikes of bright stars or satellite trails). Many of these objects no longer exist in the current imaging reductions, with its improvements to the deblender over the years. We have in fact tried to mitigate this problem in this data release, as described in the spectroscopic-photometric matching documentation.

Galaxy velocity dispersion measurements

We recommend not to use SDSS velocity dispersion measurements for:

spectra with median per-pixel (S/N)² < 10
velocity dispersion estimates smaller than about 70 km s^-1 given the typical S/N and the instrumental resolution of the SDSS spectra

Also note that the velocity dispersion measurements are not corrected to a standard aperture size.

See the velocity dispersion algorithm for details.

Clipped Spectral Lines

Some emission lines are erroneously clipped because they were identified as cosmic rays. If an emission line is so bright that it is saturated in the individual 15-minute exposures of the spectrograph, it can suffer this effect. Unfortunately, such saturated pixels are not flagged as such, although usually that region of the spectrum has an inverse variance equal to zero.

Luckily, objects with such strong emission lines are very rare, but the user should be aware of the possibility of objects with extremely strong emission lines and unphysical or unusual line ratios.

Spectrophotometric calibration induces artificial Balmer lines

Very occasionally, the spectrophotometric calibration procedure induces redshift-zero Balmer lines that are apparently due to mismatches between the calibration stars and the template library. This is noticeable in particular on some fibers in plate 274. This problem existed in DR7 and earlier as well as in DR8.

Known missing or corrupted spectra files on SAS

There are some spectra-related files on SAS which are known to be missing. These are documented in the "knownMissing.txt" files in each subdirectory. Most of these are logs and diagnostic plots, but a few spZline (redshift fits to individual lines) and spCFrame (calibrated individual exposures) files are missing. There are no known cases of missing coadded spectra.

In addition, the individual spectrum exposure SPECTRO_REDUX/26/2639/spCFrame-b2-00042347.fits is missing HDU 6 (sky), but the other HDUs are fine.