PaperlayerView exact revision on arXiv

Source-linked AI summary

The Seventh Data Release of the Sloan Digital Sky Survey

K. Abazajian

arXiv:0812.0649v2astro-ph

TL;DR

The release addresses limited photometric performance in crowded, low-latitude fields and gaps in survey coverage. It presents SDSS-DR7’s completed imaging and spectroscopy with recalibrated astrometry and improved reductions, meeting the original SDSS science goals.

  • Problem

    SDSS photometry was not optimized for crowded low-latitude fields, where dense stars degraded PSF measurement and deblending.

  • Method

    The paper presents the final SDSS-II data release, combining expanded imaging and spectroscopy with revised photometric, astrometric, and spectroscopic processing.

  • Results

    The release meets the original SDSS science goals, including complete spectroscopy across roughly 7500 deg2 of contiguous Northern Galactic Cap coverage.

  • Takeaways & Limitations

    SDSS-DR7 provides the completed SDSS-II survey dataset for studies using its expanded imaging, spectroscopy, repeat photometry, and improved calibrations.

  • Takeaways & Limitations

    Changes in u-filter effective wavelengths may significantly affect measured colors of extremely colored objects and remain under investigation.

Abstract

from arXiv · show

This paper describes the Seventh Data Release of the Sloan Digital Sky Survey (SDSS), marking the completion of the original goals of the SDSS and the end of the phase known as SDSS-II. It includes 11663 deg^2 of imaging data, with most of the roughly 2000 deg^2 increment over the previous data release lying in regions of low Galactic latitude. The catalog contains five-band photometry for 357 million distinct objects. The survey also includes repeat photometry over 250 deg^2 along the Celestial Equator in the Southern Galactic Cap. A coaddition of these data goes roughly two magnitudes fainter than the main survey. The spectroscopy is now complete over a contiguous area of 7500 deg^2 in the Northern Galactic Cap, closing the gap that was present in previous data releases. There are over 1.6 million spectra in total, including 930,000 galaxies, 120,000 quasars, and 460,000 stars. The data release includes improved stellar photometry at low Galactic latitude. The astrometry has all been recalibrated with the second version of the USNO CCD Astrograph Catalog (UCAC-2), reducing the rms statistical errors at the bright end to 45 milli-arcseconds per coordinate. A systematic error in bright galaxy photometr is less severe than previously reported for the majority of galaxies. Finally, we describe a series of improvements to the spectroscopic reductions, including better flat-fielding and improved wavelength calibration at the blue end, better processing of objects with extremely strong narrow emission lines, and an improved determination of stellar metallicities. (Abridged)

1. OVERVIEW OF THE SLOAN DIGITAL SKY SURVEY

SDSS’s seventh public data release realizes its original goals of five-band CCD imaging over 10,000 deg^2 of high-latitude sky and spectroscopy of one million galaxies and 100,000 quasars over the same region.

  • Survey goals: 10,000 deg^2 was the original imaging goal for five broad bands over high-latitude sky.The survey also aimed to obtain spectroscopy of one million galaxies and 100,000 quasars over the same region.
  • Survey goals: The seventh public data release marks realization of these original imaging and spectroscopic goals.The survey facilities were additionally used for comprehensive imaging and spectroscopic work exploring structure and composition.

2. SURVEY FOOTPRINT

DR7 expanded the imaging and spectroscopic footprints while closing the previous North Galactic Cap spectroscopy gap. It also included supplementary and repeated imaging, plus SEGUE and cluster spectroscopy targeting diverse stellar populations.

  • Imaging and spectroscopy: 22%: the imaging footprint increased since DR6, mostly outside the contiguous North Galactic Cap area.The number of spectra increased by 29%.
  • Imaging footprint: Less than 10 deg2: the Legacy imaging footprint grew in DR7 through small gap-filling and repeat observations of poor-seeing regions.Legacy imaging was already substantially complete with DR6.
  • Supplementary imaging: Supplementary imaging covered M31, Perseus, M71, the South Galactic Pole, the Sagittarius Tidal Stream, and Orion.Additional scans crossed the regular Legacy or SEGUE stripes at perpendicular or oblique angles.
  • Imaging and spectroscopy: 8032 deg2: the Legacy spectroscopy footprint, including three Southern Galactic Cap stripes, increased 26% over DR6.Spectroscopy became complete for principal galaxy and quasar targets across roughly 7500 deg2 of contiguous North Galactic Cap coverage.
  • Stellar spectroscopy: SEGUE spectroscopy used target-selection algorithms to identify stars across a wide variety of types from SEGUE and Legacy imaging.The survey also observed stars in 12 open and globular clusters to calibrate SEGUE stellar-parameter measurements.
  • Repeated imaging: 2.5°: Stripe 82, centered on the Celestial Equator, was imaged multiple times throughout SDSS and SDSS-II.The data include scans contributing to a deep coaddition and scans taken under non-ideal conditions.

3. ADDITIONAL IMAGING PRODUCTS AND DATABASES

DR7 provides databases for unique and repeat imaging detections, including a runs database spanning SDSS-I/II and scans beyond regular survey footprints. Stripe 82 repeat imaging is coadded into a deeper catalog, with improved star–galaxy separation but unreliable photometry for very bright objects.

  • Imaging databases: Primary objects define unique detections, while Secondary objects represent repeat observations in overlapping strips.The CAS flags photometric measurements as Primary or Secondary to distinguish unique detections from duplicates.
  • Imaging databases: 530 complete runs from SDSS-I and SDSS-II are included, along with scans outside the regular DR7 Legacy and SEGUE footprints.Primary status in this database is assigned strictly from geometric limits within each scan.
  • Stripe 82 coaddition: 122 Stripe 82 runs cover more than 250 deg2, with each location observed 20–40 times for coaddition.The resulting coaddition is available in the Stripe82 database and uses separate North- and South-strip designations.
  • Stripe 82 coaddition: Roughly two magnitudes deeper than single scans, the coadded photometry also improves star–galaxy separation by allowing a cut closer to the stellar locus.The coadded images are processed through the SDSS photometric pipeline to produce the Stripe82 catalog.
  • Stripe 82 coaddition: Photometry for objects brighter than roughly r = 15.5 is suspect because saturated pixels are not properly propagated through the coaddition.No processing flag identifies these data, so a simple magnitude cut is recommended.

4. IMPROVEMENTS IN PROCESSING OF IMAGING DATA · 4.1. New Reductions of SEGUE Imaging Data and Crowded Fields

The DR7 imaging reductions address failures of the high-latitude-optimized photo pipeline in crowded, low-latitude fields by combining PSPhot with revised photo processing. They also define the preferred SEGUE reductions and document a latitude-dependent aperture-correction bias.

  • 4.1. New Reductions of SEGUE Imaging Data and Crowded Fields: Above 5000 deg−2 stars brighter than r = 21, photo fails in crowded low-latitude fields because it cannot measure an accurate PSF.The pipeline was designed for high Galactic latitudes and performs adequately across the Legacy footprint.
  • 4.1. New Reductions of SEGUE Imaging Data and Crowded Fields: PSPhot improves crowded-field photometry by assuming every object is unresolved and modeling spatially varying Gaussian PSFs plus pixel-based residuals.Its PSF representation allows both analytical parameters and residuals to vary across each image field.
  • 4.1. New Reductions of SEGUE Imaging Data and Crowded Fields: PSPhot processes each frame and filter independently, outputs positions and PSF magnitudes, and matches photometry between filters within a 1′′ radius.The results are stored in the PsObjAll table in the CAS.
  • 4.1. New Reductions of SEGUE Imaging Data and Crowded Fields: photo is rerun at PSPhot-detected positions, with less aggressive overlap detection and deblending behavior to distinguish objects in crowded regions.This supplements positions independently detected by photo itself.
  • 4.1. New Reductions of SEGUE Imaging Data and Crowded Fields: 7.43′′ aperture photometry anchors PSPhot magnitudes for bright stars to the calibration system, despite isolation difficulties in crowded regions.The large-aperture measurements were used by ubercalibration to tie the photometry together.
  • 4.1. New Reductions of SEGUE Imaging Data and Crowded Fields: |b| < 25◦ SEGUE runs were processed with photo and PSPhot, and rerun=648 was designated their Best reduction in the DR7 CAS and DAS.The Target version instead used photo alone for SEGUE spectroscopic-plate design.
  • 4.1. New Reductions of SEGUE Imaging Data and Crowded Fields: 3.00′′ aperture corrections were used when runs dipped below |b| = 8◦, underestimating corrections by typically 0.03–0.06 mag depending on seeing.The issue arose because sufficiently isolated stars were unavailable for the intended large-aperture measurement.

4.2. Comparison of photo and PSPhot Photometry

PSPhot and photo produce broadly comparable stellar photometry, but PSPhot has tighter scatter across Galactic latitudes and is preferred for stellar-locus studies. Quality control flags approximately 2% of fields using photo outputs and 3.6% using PSPhot outputs, primarily near the Galactic plane.

  • Photometry comparison: 0.014 mag rms characterized the within-field median difference between PSPhot and photo PSF magnitudes for stars with 14 < u, g, r, i, z < 20.Fields exceeding 0.08 mag are suspect and require investigation of the responsible pipeline.
  • Photometry comparison: Up to 0.1 mag lower Qgri and Qriz colors occurred at low Galactic latitudes and were corrected by subtracting a fitted constant-plus-Lorentzian model.The medians used stellar objects satisfying the stated magnitude and color cuts.
  • Quality control: 2% of fields were flagged bad from photo outputs and 3.6% from PSPhot outputs when either |Qgri| or |Qriz| exceeded 0.12 mag.The vast majority of flagged fields lie within 15° of the Galactic plane.
  • Quality control: PSPhot showed tighter scatter than photo at both high and low Galactic latitudes, so its stellar photometry was preferred for stellar-locus studies.Bad fields should be identified in the fieldQA table and culled; robustness to outliers was not fully assessed.

4.3. Further Assessments of Imaging Quality

The section adds field-level imaging-quality flags for focus, rotator, astrometric, and miscellaneous problems, marking affected fields across all five bandpasses. These flags identify only 0.3% of CAS fields, while photometric-locus diagnostics compare the photo and PSPhot pipelines across Galactic latitudes.

  • Field-quality flags: 60 µm focus motion over three frames triggers a focus-problem flag, indicating an automated telescope-focus issue.The criterion uses the median telescope focus over three frames and assigns Problemchar = ’f’.
  • Field-quality flags: 25′′ rotator motion between adjacent fields triggers a flag corresponding to a 0.55′′ image shift at the camera edge.This condition assigns ProblemChar=’r’.
  • Field-quality flags: 4 pixels (1.6′′) of astrometric-solution shift from smooth interpolation triggers an astrometric-problem flag.This condition assigns ProblemChar=’a’.
  • Flag propagation and prevalence: 0.3% of CAS fields are marked for these problems, with most flagged fields attributed to focus problems.Focus problems mark 11 preceding and three following fields, while rotator and astrometric problems mark nine preceding and one following field.
  • Photometric-locus diagnostics: PSPhot produces a tighter stellar locus than photo at high Galactic latitudes, while within the Galactic plane it is more concentrated but has more systematic departures.The Qgri and Qriz parameters measure stellar-locus position, with zero indicating uniform photometry.

4.4. Astrometric Recalibration

DR7 recalibrated all astrometry against UCAC2, incorporating proper-motion information from UCAC2 and SDSS+USNO-B. The release also documents changed object names and a prior proper-motion coding error affecting DR3–DR6.

  • Astrometric Recalibration: All DR7 astrometry was recalibrated against UCAC2, whose systematic errors are thought to be less than 20 mas.UCAC2 provides proper motions for stars with δ < +41°, while higher-declination stars use proper motions merged from SDSS+USNO-B and UCAC2 positions.
  • Astrometric Recalibration: DR7 object names may differ slightly from DR6 because of subtle astrometric changes.Users comparing objects between the two releases should account for these naming differences.
  • Astrometric Recalibration: 1–2 mas year−1 systematic errors in right-ascension proper motions affected Data Releases 3 through 6 because of an error in the proper-motion code.The CAS provides proper motions in the ProperMotions table, derived by combining SDSS astrometry with recalibrated USNO-B positions.

4.5. SEGUE Target Selection

SEGUE target-selection algorithms evolved during SDSS-II, with major changes to K-giant selection based on color separation. Version 4.6 was applied across the imaging catalog to support completeness and efficiency analyses.

  • 4.5. SEGUE Target Selection: SEGUE target-selection algorithms evolved during the course of SDSS-II.The most significant changes affected the K-giant algorithm.
  • 4.5. SEGUE Target Selection: g −r > 1.1 identified the reddest giant candidates for color-based luminosity separation using deviations from the main-sequence locus in the ugr color diagram.This approach requires accurate u-band photometry.
  • 4.5. SEGUE Target Selection: v4.6 was applied to all stellar imaging-catalog objects with g < 21 or z < 21 across the entire sky.The application enabled users to analyze completeness and efficiency of SEGUE stellar target-selection samples.
  • 4.5. SEGUE Target Selection: SEGUEPrimTarget and SEGUESecTarget fields received the appropriate target-selection bits in the DR7 CAS photoObjAll table.The bits and algorithms were documented in Yanny et al. (2009).

4.6. Photometric Redshifts

DR7 improves both SDSS galaxy photometric-redshift offerings by adding per-galaxy redshift probability distributions and substantially revising the template-based code with a hybrid calibration method. The revised algorithm reduces reference-set rms error from 0.044 in DR6 to 0.025 in DR7, while p(z) aggregation improves redshift-distribution estimation and lensing calibration.

  • Photometric-redshift methods: Two photometric-redshift determinations are provided: neural-net solutions and a template-fitting approach, with improvements to both in DR7.The neural-net solutions remain unchanged since DR6, while the other code receives a substantial methodological revision.
  • Neural-net redshifts: A value-added catalog provides each galaxy’s p(z), calculated with the weights method from spectroscopic-redshift distributions.The neural-net photometric-redshift solutions and errors are listed as Photoz2 and do not use ubercalibrated magnitudes.
  • Neural-net redshifts: Summing galaxies’ p(z) gives a better estimate of their true redshift distribution and significantly reduces photometric lensing calibration bias versus one redshift estimate per galaxy.These findings are attributed to Cunha et al. (2008) and Mandelbaum et al. (2008).
  • Template-based redshifts: DR7’s revised Photoz code combines template fitting with empirical calibration from objects having observed colors and spectroscopic redshifts.The method uses a k-d tree to find the 100 nearest neighbors in ubercalibrated four-color space and fits a local hyperplane after outlier rejection.
  • Template-based redshifts: 0.025 is the DR7 reference-set rms redshift error, down from 0.044 in DR6; after removing outliers beyond 3 σ, the corresponding errors are 0.020 and 0.028.The quoted errors are also substantially more reliable with the improved algorithm.

4.7. SDSS Filter Response Functions

SDSS griz response functions remained stable apart from negligible seasonal variations, while the u-band response changed substantially in amplitude and shape. Photometric calibration corrected the resulting zero-point change near standard-star colors, but the u-filter effective wavelength shifted roughly 30 Å redward over the survey.

  • The griz response functions were stable, with seasonal temperature variations well below typical photometric errors.
  • The u-band response changed relatively greatly in both amplitude and shape, likely because of degradation of the UV-enhanced u-band CCD coating.
  • Photometric calibration effectively corrected the u-band zero-point change for objects near the mean color of standard stars.
  • Repeat photometry in Stripe 82 remained stable over time for stars with −0.5 < g −r < 1.5.
  • The u-filter effective wavelengths shifted roughly 30 Å redward over the survey lifetime.

5. PHOTOMETRY OF BRIGHT GALAXIES

The paper revises the estimated impact of SDSS sky-subtraction errors on bright-galaxy photometry using more appropriate galaxy size–magnitude relations. Simulations show substantial underestimation of galaxy scale sizes, while the effect on typical galaxies is less severe than previously reported.

  • Systematic photometric errors: Bright galaxies with r < 16 have underestimated fluxes and scale sizes, while neighboring-object counts are suppressed by systematic sky-estimation errors.These effects arise from errors in estimating the sky near bright galaxies.
  • Revised simulations: The revised simulations used observed magnitude–half-light-radius relations and modeled n = 1 and n = 4 galaxies with axis ratios b/a = 0.5 and 1.The earlier simulations used an incorrect size–magnitude relation that overstated the problem for typical galaxies.
  • Simulation results: 20% and 30% are the maximum scale-size underestimations for simulated galaxies with Sersic indices of 1 and 4, respectively.The simulations included disk-like and elliptical-like profiles with axis ratio 0.5.
  • Simulation results: The most massive elliptical or cD galaxies may experience larger effects because their more extended envelopes are not represented by the typical-galaxy simulations.This caveat is attributed to the greater extent of their outer light profiles.

6. IMPROVEMENTS IN PROCESSING OF SPECTROSCOPIC DATA

DR7 improves spectroscopic processing by correcting shifting blue-camera flat-field interference, strengthening blue-end wavelength calibration, preserving sharp emission lines, and improving stellar metallicity estimates. These changes address known DR6 processing errors while retaining some documented radial-velocity systematics.

  • Flat-field correction: DR7 models and corrects wavelength-shifting interference patterns in blue-camera flat fields using repeat exposures of each plate.The unstable component shifts between exposures, so extracted-spectrum ratios are compared across the typically 45-minute plate observation.
  • Wavelength calibration: 20 km s−1: bright moonlight could systematically skew the blue-end wavelength solution by this amount; DR7 fits the affected Hg line with a stellar template.Bright moon affected 10 of 410 SEGUE plates.
  • Known limitations: The DR7 pipeline retains a 7 km s−1 systematic error in SEGUE stellar radial velocities, with a correction applied in the SSPP outputs but not elsewhere in CAS or DAS.Plate-to-plate SEGUE velocities also have systematic errors of about 2 km s−1 in the mean.
  • Spectral combination: DR7 improves spectral combination so very strong, sharp emission lines are no longer erroneously rejected as discrepant points.The revised processing prevents strong [OIII] lines from being clipped as cosmic rays.
  • Stellar parameters: DR7 improves stellar metallicity estimates by adding super-solar synthetic spectra, recalibrating several methods, and retraining ANNRR on improved stellar parameters.DR6 under-estimated metallicities by about 0.3 dex for stars approaching solar metallicity.

7. LOOKING AHEAD TO SDSS-III

SDSS-II met the original SDSS science goals and added extensive stellar spectroscopy and repeat imaging. SDSS-III begins a new operational phase with four surveys targeting the distant halo, baryon oscillations, exoplanets, and Milky Way evolution.

  • SDSS-II completion: SDSS-II met its original science goals and added spectroscopy of close to half a million stars plus repeat imaging over 250 deg2.The original goals included five-band imaging and spectroscopy of 106 galaxies and 105 quasars.
  • SDSS-III: SDSS-III will operate four surveys with the 2.5m telescope through 2014.Its data will be made public through a series of data releases following the SDSS and SDSS-II pattern.
  • SDSS-III: SEGUE-2 targets fainter stars in the distant halo and will increase the number of distant halo stars by a factor of 2.5.It uses the same instrumentation and data processing pipelines as SEGUE.
  • SDSS-III: BOSS will survey 1.5 million luminous red galaxies to z ≈0.7 and 160,000 quasars with 2.3 < z < 3.The survey aims to measure the baryon oscillation scale in the correlation function as a function of redshift.
  • SDSS-III: MARVELS will monitor the radial velocities of 11,000 bright stars for planets with periods from several hours to two years.It searches for the radial-velocity signature of planets.
  • SDSS-III: APOGEE will obtain R ≈20, 000 H-band spectroscopy of 105 giant stars to H = 13.5 for Milky Way radial-velocity and chemical studies.The survey is designed for detailed studies of Galactic evolution.
Loading 0812.0649v2…