Spectra

Up: Data Products Sections: Images - Object lists - Spectra - Tiling

Down: Quality of spectra - About SDSS spectra - Caveats

DR7 contains completely reprocessed spectra.

In DR7, there have been improvements to the algorithms which photometrically calibrate the spectra, and all spectra have been re-reduced. The new spectro "rerun number" is 26.

About spectroscopic samples

Normal survey plates

The spectroscopic survey is predominantly a survey for

• Bright galaxies, the "main" sample, with Petrosian r-band magnitude < 17.77 (to get an idea of their distribution, see this pie plot showing the redshift distribution of 78275 DR1 galaxies with -15° < DEC < 20°).
• Luminous Red Galaxies (LRGs) with Petrosian r-band magnitude down to 19.5 (see this redshift distribution of DR2 "main" galaxies and LRGs), and
• Quasars, targetting quasar candidates as faint as i < 20.2 (see this DR2 quasar redshift distribution).

There is also a large number of spectra of stars. Brown dwarfs, ROSAT and FIRST counterparts are targetted as well. Refer to the target selection quality or target selection algorithm for details about the spectroscopic survey targets.

Extra and special plates, including SEGUE

New since DR4 is the publication of "extra" and "special" plates. Extra plates are repeat observations of normal survey plates. These have the same plate number as the original observation, but a different MJD and usually a different mapping between fiber number and ra,dec. Special plates are spectroscopic observations made under a variety of special programs which are described on the special spectra page.

Note the caveat below about querying special-plate data in the CAS (including SEGUE).

Getting and using spectra

The spectra distributed by the SDSS have been sky subtracted, corrected for telluric absorption spectrophotometrically calibrated by the Spectro2d pipeline. Spectral classification, redshift determination, and emission and absorption line measurements are carried out by the "1D" pipeline. DR7 contains the outputs from two independent 1D pipelines: the spectro1d code that has also been used in all previous releases, and the specBS code whose outputs ("Princeton reductions") were previously available only from spectro.princeton.edu. The CAS contains only the spectro1d outputs as before; the specBS outpus are available through the DAS.

The algorithms page contains details about spectroscopic data processing for all pipelines.

Since DR1, the Data Archive Server has been providing the same spectrum in two files:

• Spectro2d spPlate*.fits: all calibrated spectra and supporting data (signal-to-noise, resolution, quality flags and similar) from a single plate, without any parameters or continuum fits. In DR7: the spPlate files now also contain the sky spectra that have been used for sky subtraction of each spectrum.
  See the spPlate*.fits data model for details of the file contents. There is one 2d directory for every plate number in the DAS that contains all spPlate files for that plate, including those for different MJDs where repeats are available; e.g. http://das.sdss.org/spectro/2d_26/0644/
• Spectro1d spSpec*.fits: multiple fits extensions (images and tables) with the calibrated spectrum, a continuum-subtracted spectrum, and all measured parameters (redshift, line fits, line indices, per-pixel resolution). See the spSpec*.fits data model for details of the file contents, and how to read an spSpec file. There is one spSpec file per plate/MJD/fiber combination, found in a parallel 1ddirectory tree with one directory per plate, e.g. http://das.sdss.org/spectro/1d_26/0644/1d/
• In addition, spectro1d produces spPlot*.gif,.ps plots of the spectra with line locations for the best-fitting redshift, again in one directory per plate, e.g. http://das.sdss.org/spectro/1d_26/0644/gif/

All of the above files are available through the DAS query form.

DR7 provides additional spectroscopic data:

• spectro2d spCFrame*.fits: SDSS spectra are typically combined from 3 or more individual exposures of 15 minutes each. The individual flux-calibrated spectrograph exposures are available in spCFrame*.fits files. They contain spectra in the spectrograph's native wavelength mapping, which is neither linear in wavelength nor log-wavelength.
  
• specBS outputs ("Princeton reductions"; see https://web.archive.org/web/20161021194947/https://web.archive.org/web/20161021194947/http://spectro.princeton.edu/):
  • spZbest*.fits: best-fit classification and redshift for a given spectrum
  • spZall*.fits: Same structure aas spZbest, but contains all attempted fits
  • spZline*.fits: emission-line measurements
  There is one file of each type for every plate/MJD combination.
• sspp stellar parameter pipeline outputs: Most of these outputs are available only through two new tables in the CAS:
  • Line index equivalent widths for stars: sppLines
  • Stellar atmospheric parameters ([Fe/H], log g, Teff) for stars: sppParams for all the SEGUE spectra and a large fraction of the main database spectra classified as stars (not galaxies or quasars).
  In the DAS, we provide bulk versions of the stellar parameter outputs in one directory per plate/MJD combination, e.g. spectro/sspp_26/0644-52149, which include .csv versions of the sppParams and sppLines tables and some additional information and plots not available through the CAS.

The specBS outputs are available through the DAS query form. The spCFrame files are not available through the DAS query form, but can be found in the same 2d directories as the spPlate files in the DAS; these also contain the specBS outputs. E.g., http://das.sdss.org/spectro/2d_26/0644/.

Bulk download of spectra, spectroscopic parameters, and corresponding imaging data

Very often one wishes to obtain all the SDSS photometric (imaging camera) data associated with each spectrum. One can obtain this information most easily for nearly all plates through the SQS interface by selecting imaging parameters to return with your selected spectra.

For those who prefer working with the full fits files of spectra and imaging catalogs, make available a special set of files (spObj-*.fit) containing a bundled version of the spectroscopic data and all the corresponding imaging data.

These files are available via the DAS by requesting the data product tsObjFromMap (or spObj) for the plates you wish. To get these files in bulk for the entire survey, see the page on redshift catalogs and SDSS spectra and corresponding imaging data.

More on data access

The data access page contains various query forms to get spectra by coordinates, or to search for spectra by redshift, object magnitude, color etc., and to retrieve them from the archive. In particular, the Catalog Archive Server provides a fast search capability for object lists and spectroscopic parameters as well as pointers to the files in the Data Archive Server, whose structure is explained on the Data Archive Server structure and contents page. The Spectro Query Server query form is dedicated to the search of the spectroscopic database. You can also download fits tables with the classification, redshifts and other information from the page with redshift catalogs and spectro+imaging data.

Please note the caveats below, which are essential to obtain meaningful scientific results from analysing SDSS spectra, in particular the redshift status caveat.

Quality of spectra

• spectroscopic signal-to-noise as function of magnitude
• Details about the quality of spectrophotometry
• Accuracy of stellar radial velocities improved substantially in DR7

About the spectra

For details about the spectrographs, see the spectrograph page.

Further details

• Error and mask arrays, and the sky spectrum subtracted from each object spectrum, are available (see the data model.)
• Spectroscopic observations are undertaken in non-photometric conditions when the imaging camera is not in use. At least three fifteen-minute exposures are taken until the cumulative mean S/N per pixel exceeds 4 for a fiducial fiber magnitude of g = 20.2 and i = 19.9.
• We provide the cross-correlation templates used to obtain cross-correlation redshifts.

Caveats

DAS only plates

 plate-mjd      Why not in CAS platex table:
 -----------------------------------------------------------------------
 0356-51779    S/N too low -- plate not done (not enough exposure)
 1112-53180    S/N too low -- plate not done
 1858-53271    SEGUE test plate, high DEC, never made it to CAS
 2309-54441    S/N too low -- SEGUE plate
 2333-53676    M71 very bright plate, lots of saturation, only 320/640 done
 2535-54632    SEGUE bright, faint matching plate not done 
 2640-54474    BAOTest plate, test plate for SDSS-III 
 2716-54629    SEGUE bright, faint matching plate not done
 2813-54650    SEGUE bright, insufficient S/N, plate not done
 2851-54485    SEGUE dup pointing of plate 2045.
 2962-54578    don't have correct tsObj file, Low S/N, plate not done

Note on Radial Velocities

Redshift status

Only 1% of the objects have an "unknown" classification, usually because of low signal-to-noise ratio or completely featureless spectra. The redshifts of all but a few tenths of a percent of the remainder are believed to be correct. To identify the few objects with unreliable redshifts, be sure to consider the confidence we have assigned to each redshift (z_conf in the spSpec*.fits primary header, and the status zStatus and zWarnin of the redshift measurement, which may have failed.). A useful cut on the redshift confidence is z_conf > 0.35 (or zConf > 0.35 in the SQL database).

Zero equivalent width of emission lines, especially H alpha

There is a bug in the line-measurement code that has been in use since DR3 which gives some emission lines an equivalent width of zero, even though there is a significant line detection. The aim of the change introducing the bug had been to determine the equivalent width by integrating the spectrum, instead of using the parameters of a fitted Gaussian. The Gauss-fit equivalent width can be recovered from the fit parameters using the usual expression EW = 2.5066 * sigma * height / continuum.

Galactic extinction correction

In the EDR and DR1, the spectroscopic data were nominally corrected for galactic extinction. The spectrophotometry since DR2 is vastly improved compared to DR1, but the final calibrated spectra in DR2 and beyond are not corrected for foreground Galactic reddening (a relatively small effect; the median E(B-V) over the survey is 0.034). Users of spectra should note that the fractional improvement in spectrophotometry from DR1 to DR2 and beyond was much greater than the extinction correction itself. As the SDSS includes a substantial number of spectra of galactic stars, a decision has been taken not to apply any extinction correction to spectra, since it would only be appropriate for extragalactic objects, but to report the observational result of the SDSS, namely, the spectrum including galactic extinction.

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.

Plates with not-quite-perfect spectrophotometry

A small number of plates, given in the list of not-quite-perfect plates, suffered from a variety of minor problems. The CCD frames for several plates suffered from a transient electronic problem in the red camera in Spectrograph 2, causing the columns of the CCD to be misaligned on readout. This was fixed in software, and we believe the data to be reliable. Another set of plates labeled "Spectrograph Collimation Problem" suffered from having the spectrograph collimator improperly focused. This problem caused a mismatch between the flatfields and the science exposure instrumental profile shapes on the CCD in both the spatial and wavelength directions, causing 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. Other plates have individual problems as noted. E.g., during the exposure of one plate, light from an LED somewhere on the telescope found its way to the spectrographs, resulting in an artificial excess of light centered roughly at 6500Å; the spectrophotometry of this plate is quite poor.

Mismatches between spectra and photometric data

In a few cases, the fiber mapping failed which identifies which fiber has been plugged into which hole. When this happens for two or more objects on the sample plate, there is the possibility of wrong matches between spectra and photometric objects. There are 123 objects for which the mapping between object and spectrum cannot be established ("unmapped fibers"); their ra/dec is listed as -9999 in the spectroscopic data set but synthetic fiber magnitudes (mag_0, mag_1, mag_2 in the specObj tables in the CAS, mag_g, mag_r, mag_i in the spSpec*.fit files) greater than 0 (i.e., use the synthetic magnitudes to distinguish the unmapped fibers from the 2221 fibers which were broken completely at the time of observations and did not yield a spectrum at all).

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 best imaging reductions with its improvements to the deblender. In other cases, the photometric pipeline timed out during the best imaging reductions in fields for which target imaging proceeded without problem, so that the best photometry is missing for bona-fide objects. This predominantly happens in fields close to a few very bright stars. We expect to recover objects from these ``timeout holes'' in future data releases.

The special plates have special issues regarding photometric matches. Some of the special plates were targeted using photometry that is not part of DR7, or not even SDSS photometry. Therefore, the SDSS photometry for objects on plates 797, 1468, 1471, 1472, 1665, and 1666 is only available through DRsup.

In addition, special plates with SDSS photometry are different from regular SDSS plates in that they have no tiling information associated with them. Therefore, the specPhotoAll table is not populated with all entries from special plates. Querying SEGUE data is explained in a SEGUE sample SQL query.

To obtain photometric information for non-SEGUE special-plate spectra, perform the following join:

select columns
from specObjAll as so
inner join photoObj as po on so.bestobjid = po.objid

Galaxy velocity dispersion measurements

The velocity dispersion measurements distributed with SDSS spectra use template spectra convolved to a maximum sigma of 420 km/s. Therefore, velocity dispersion sigma > 420 km/s are not reliable and must not be used. The figure below shows the quality of velocity dispersion error estimates.

Error distribution of the velocity dispersion measurements from spectro1d DR6 (thin black solid line), spectro1d DR5 (dotted red line), specBS (dashed blue line), and B03 (dotted-dashed green line). The thick solid line was obtained by comparing repeated measurements.

We recommend the user to not use SDSS velocity dispersion measurements for:

• spectra with median per-pixel S/N < 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 relative circular aperture.

See the velocity dispersion algorithm for details.

"Bonus" plates beyond the survey limits

A few plates target objects beyond the survey limits for a particular survey stripe, stripe 10 (see survey coverage page). These plate/MJD combinations are:

Their corresponding targetimaging data does not contain any PRIMARY objects. Objects from this region are therefore not available via a normal search of the target object lists. To find the target data, select on status & 0x402 (both GOOD and OK_SCANLINE) instead of selecting on the PRIMARY flag being set in status. The information is also contained in the spObj-*.fit files, which are available for all plates. These files are available via the DAS by requesting the data product tsObjFromMap (or spObj) for the plates you wish. They are also available for bulk rsync or wget download via DAS through http (in spectro/ss_SPRERUN/PLATE/spObj-*).

These objects have been declared primary in the best data set. The CAS correctly records the positional matches between these spectra and their counterparts in best. However, since the corresponding targets are not primary, the spectra are not included in the specObj and specPhoto views in the CAS, but must be searched explicitly in specObjAll and photoObjAll.

Accuracy of stellar radial velocities

The accuracy of stellar radial velocities in DR2/DR3 and beyond is described on a separate page.

Clipped Spectral Lines

The spectroscopic pipeline combines observations of a given object on the red and blue spectrographs, and between the separate 15-minute exposures on the sky, by fitting a tightly-constrained spline to the data, allowing discrepant points such as cosmic rays to be rejected. This spline requires as input the effective resolution of the spectra.

As described in the DR6 paper, it did not do a perfect job; occasionally, very strong and sharp emission lines were erroneously rejected by this algorithm. This turned out to be due to the fact that the spline code did not adequately track the changing resolution of the spectra as a function of wavelength and fiber number. Including this effect much improved the behavior of this algorithm. The figure below shows an example spectrum of an object affected by this problem in DR6, and its improved counterpart in DR7, as is apparent by the correct 3:1 ratio of the 5007 and 4959 Å lines of [OIII]. The total number of objects that were affected by this problem is quite small; affecting less than 1 percent of galaxies with Hβ equivalent width >25 Å.

There is another problem, unfortunately not fixed in DR7, which has a similar effect. If the line is so bright that it is saturated in the individual 15-minute exposures of the spectrograph, it will also appear clipped. Unfortunately, such saturated pixels are not flagged as such, although usually such lines are recognizeable as having 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.