PaperlayerView exact revision on arXiv

Source-linked AI summary

The Data Reduction Pipeline for the SDSS-IV MaNGA IFU Galaxy Survey

David R. Law, Brian Cherinka, Renbin Yan, Brett H. Andrews, Matthew A. Bershady, Dmitry Bizyaev, Guillermo A. Blanc, Michael R. Blanton, Adam S. Bolton, Joel R. Brownstein, Kevin Bundy, Yanmei Chen, Niv Drory, Richard D'Souza, Hai Fu, Amy Jones, Guinevere Kauffmann, Nicholas MacDonald, Karen L. Masters, Jeffrey A. Newman, John K. Parejko, José R. Sánchez-Gallego, Sebastian F. Sánchez, David J. Schlegel, Daniel Thomas, David A. Wake, Anne-Marie Weijmans, Kyle B. Westfall, Kai Zhang

arXiv:1607.08619v3astro-ph.IM

TL;DR

MaNGA addresses the challenge of mapping galaxies internally while meeting demanding calibration, astrometric, and 3-D reconstruction requirements. This paper describes its data-reduction pipeline and metadata framework, demonstrating nearly Poisson-limited sky subtraction and calibrated data cubes with quantified imaging and spectral performance.

  • Problem

    MaNGA must overcome substantial calibration, astrometric, flexure, and 3-D reconstruction challenges to produce high-quality imaging spectroscopy rather than point-source spectra.

  • Method

    The MaNGA Data Reduction Pipeline combines calibrated fiber spectra, astrometric solutions, flux-conserving cube building, and centralized metadata tracking to construct rectified 3-D data cubes.

  • Results

    For 1390 DR13 galaxy cubes, sky subtraction is nearly Poisson-limited below ∼8500 Å, while typical calibration, sensitivity, and spatial-resolution performance reaches 1.7%, µ = 23.5 AB arcsec−2, and 2.54 arcsec FWHM, respectively.

  • Takeaways & Limitations

    The resulting MaNGA data products provide sky-subtracted, spectrophotometrically calibrated spectra and rectified 3-D data cubes for studying galaxy structure.

  • Takeaways & Limitations

    DR13 instrumental line-spread-function estimates are systematically underestimated relative to values measured by many third-party routines, and correction efforts remain ongoing.

Abstract

from arXiv · show

Mapping Nearby Galaxies at Apache Point Observatory (MaNGA) is an optical fiber-bundle integral-field unit (IFU) spectroscopic survey that is one of three core programs in the fourth-generation Sloan Digital Sky Survey (SDSS-IV). With a spectral coverage of 3622 - 10,354 Angstroms and an average footprint of ~ 500 arcsec^2 per IFU the scientific data products derived from MaNGA will permit exploration of the internal structure of a statistically large sample of 10,000 low redshift galaxies in unprecedented detail. Comprising 174 individually pluggable science and calibration IFUs with a near-constant data stream, MaNGA is expected to obtain ~ 100 million raw-frame spectra and ~ 10 million reduced galaxy spectra over the six-year lifetime of the survey. In this contribution, we describe the MaNGA Data Reduction Pipeline (DRP) algorithms and centralized metadata framework that produces sky-subtracted, spectrophotometrically calibrated spectra and rectified 3-D data cubes that combine individual dithered observations. For the 1390 galaxy data cubes released in Summer 2016 as part of SDSS-IV Data Release 13 (DR13), we demonstrate that the MaNGA data have nearly Poisson-limited sky subtraction shortward of ~ 8500 Angstroms and reach a typical 10-sigma limiting continuum surface brightness mu = 23.5 AB/arcsec^2 in a five arcsec diameter aperture in the g band. The wavelength calibration of the MaNGA data is accurate to 5 km/s rms, with a median spatial resolution of 2.54 arcsec FWHM (1.8 kpc at the median redshift of 0.037) and a median spectral resolution of sigma = 72 km/s.

1. INTRODUCTION

MaNGA combines the statistical reach of multiplexed spectroscopy with spatially resolved integral-field observations, requiring a reduction pipeline that addresses demanding calibration and cube-construction challenges. The contribution describes the DRP and its DR13 release products.

  • Large multiplexed surveys provide statistical power but historically treat galaxies as point sources through small, biased fiber regions.
  • Integral-field surveys combine spatial and spectral coverage with larger samples to probe galaxies' internal structure.
  • Accurate spectrophotometry, flexure correction, micron-precision astrometry, and cube construction are central MaNGA calibration challenges.
  • The DRP uses a BOSS-derived 2d stage for fiber spectra and a custom 3d stage for spatially registered data products.
  • Version v1.4 of the MaNGA DRP corresponds to the first public SDSS DR13 science-data release.

2. MANGA HARDWARE AND OPERATIONS

MaNGA combines configurable fiber-bundle hardware, dual-camera BOSS spectrographs, repeated calibration, and dithered observations to obtain spatially sampled galaxy spectra. Operations and metadata handling are designed around accurate mapping, calibration, and coverage.

  • Hardware: Each cartridge contains 12 seven-fiber calibration IFUs and 17 science IFUs ranging from 19 to 127 fibers.
  • Hardware: Each fiber has a 120 µm glass core subtending 1.98 arcsec, while associated sky fibers sit at block ends to minimize crosstalk.
  • Hardware: The spectrographs provide combined wavelength coverage of approximately 3600–10,300 Å through blue and red cameras with a 400 Å overlap.
  • Operations: Calibration uses Hartmann focus exposures, quartz flatfields, and Neon-Mercury-Cadmium arc lamps, with throughput and wavelength calibrations repeated hourly.
  • Operations: Science observations use sets of three 15-minute dithered exposures to improve focal-plane coverage despite the IFU's 56% circular-aperture fill factor.
  • Operations: Weather can prevent uniform three-exposure dither sets, so scheduling assembles compatible exposures from similar hour angles across nights.

3. OVERVIEW: MANGA DATA REDUCTION PIPELINE (DRP)

The MaNGA DRP converts individual exposures into calibrated fiber spectra and astrometrically combined data cubes through automated 2d and 3d stages. A centralized, versioned metadata system supports reproducible processing, while DOS provides rapid exposure-quality feedback.

  • DRP organization: The DRP produces flux-calibrated individual-exposure spectra and stacked galaxy data cubes with associated inverse-variance and mask products.The 2d stage processes plate exposures; the 3d stage combines information from many exposures for each galaxy.
  • 2d reduction: The 2d stage preprocesses CCD frames, extracts and calibrates fibers, subtracts sky, and derives flux calibration from standard stars.It uses calibration exposures for fiber traces, flatfields, and wavelength solutions before processing science frames.
  • 3d reduction: The 3d stage uses fiber metrology, dither offsets, differential-refraction corrections, and imaging refinement to construct rectified data cubes.It also produces associated inverse-variance and mask cubes.
  • Quick reduction: DOS is a speed-optimized, pared-down pipeline that reduces calibration and science exposures through sky subtraction for real-time observing feedback.Unlike the DRP’s profile-fitting extraction, DOS uses boxcar extraction, trading accuracy and robustness for speed.
  • Metadata and quality control: The mangacore repository tracks changing survey metadata and versions it by observation dates and major data-release tags so pipeline versions can be rerun with appropriate metadata.The repository is synchronized daily between APO and the Utah reduction hub.

5. SKY SUBTRACTION

MaNGA’s sky-subtraction pipeline uses dedicated sky fibers and hybrid one- and two-dimensional models to handle continuum and skyline structure. DR13 tests show near-Poisson performance through approximately 8500 Å, with residual systematics at longer wavelengths.

  • Procedure: Sky subtraction targets Poisson-limited performance from 4000–10,000 Å using 92 dedicated sky fibers to model the background.The increasing read noise of the BOSS cameras prevents this performance beyond 10,000 Å.
  • Procedure: The pipeline builds a super-sampled sky spectrum from sky fibers, fits it with an iterative cubic B-spline, and evaluates it on each fiber’s native wavelength solution.Fiber-specific scaling accounts for relative sky backgrounds across harnesses.
  • Procedure: The final model uses the 1-d sky model in continuum regions and the 2-d model near bright skylines, trading some SNR for improved skyline-wing and line-spread-function fidelity.The transition is defined within 3 Å of wavelengths where the sky exceeds the interline continuum by more than 5σ.
  • Performance: Poisson-ratio distributions are broadly consistent with theoretical expectations in all four cameras, with slight average oversubtraction and a non-Gaussian blue-camera wing.The asymmetric wing does not correspond to particular wavelengths or fiber IDs.
  • Performance: Skyline-region performance is within 10% of Poisson expectations through approximately 8500 Å, worsening to 10–20% above theoretical expectations at longer wavelengths.The excess is attributed to systematic residuals associated with brighter skylines and increased detector-spectrum curvature.
  • Performance: Stacking performance approaches nominal expectations once fibers from both spectrographs are included, with the transition occurring at N > 621.Sequential stacking along one slit is initially limited by the 46 sky fibers on that slit rather than all 92 sky fibers.
  • Performance: DRP and skycorr produce nearly indistinguishable sky-subtracted spectra for a representative galaxy-fiber spectrum.The comparison indicates comparable performance between the two techniques.

6. FLUX CALIBRATION

MaNGA flux calibration is designed for spatially sampled IFU data rather than total point-source fluxes. Standard-star mini-IFUs provide aperture corrections and system-throughput estimates that yield percent-level relative and absolute calibration accuracy.

  • Calibration goal: IFU flux calibration accounts for seeing-convolved spatial sampling, rather than only recovering the total flux within a single fiber aperture.This distinction reflects the different calibration goal for spatially resolved IFU observations.
  • Calibration method: Twelve 7-fiber mini-IFU bundles assigned to standard stars provide the calibration data for each plate.Six calibration bundles are assigned per spectrograph.
  • Calibration method: A wavelength-dependent PSF model estimates each standard star’s total flux and relative flux distribution across its seven fibers.The model incorporates guider-based seeing, wavelength-dependent seeing, and focal-plane-to-plate shape mismatch.
  • Calibration method: System-response vectors are derived by comparing aperture-corrected standard-star fluxes with theoretical templates normalized to SDSS broadband magnitudes.Correction vectors from individual standards are averaged per exposure before application to science fibers.
  • Performance: Relative calibration is accurate to 1.7% between Hβ and Hα and 4.7% between [O ii] λ3727 and [N ii] λ6584.Absolute RMS calibration is better than 5% over more than 89% of MaNGA’s wavelength range.
  • Performance: The median g-band flux scaling factor improved from 0.98 to 1.01 after DR13 pipeline improvements to modeling the outer SDSS telescope PSF wings.The earlier galaxy-to-galaxy scatter was σ = 0.04.

7. WAVELENGTH RECTIFICATION

The pipeline combines flux-calibrated camera frames across the dichroic break onto fixed wavelength grids before subsequent coaddition. This introduces slight covariance and reduces effective spectral resolution by approximately 6%.

  • Wavelength combination: The camera-combination stage merges four flux-calibrated frames and all 1423 fibers onto a common fixed wavelength grid.It also combines spectra across the approximately 6000 Å dichroic break.
  • Wavelength combination: Camera combination introduces slight spectral covariance and degrades effective spectral resolution by approximately 6%.The common grid is required for ultimately coadding individual spectra.
  • Wavelength grids: The logarithmic grid spans 3621.5960–10353.805 Å with 10^-4 dex steps, yielding 4563 spectral elements.Its dispersion ranges from 0.834 Å channel^-1 to 2.384 Å channel^-1.
  • Wavelength grids: The native CCD wavelength grid varies by fiber, with approximately 1.0 Å pixels in the blue camera and 1.4 Å pixels in the red camera.Rectification therefore places fiber-dependent native sampling onto a common representation.
  • Wavelength grids: A cubic B-spline fit is evaluated on fixed wavelength solutions to rectify spectra around emission features including Hα, [N ii], and [S ii].The native spectrum and rectified fit are compared through their residual difference.
  • Data cleaning: A second-pass cosmic-ray mask grows along fiber and wavelength directions, flagging pixels more than 5σ from local sigma-clipped means.This reduces unflagged cosmic-ray features in the final products.

8. ASTROMETRIC REGISTRATION

MaNGA constructs wavelength-dependent fiber positions from hardware, observing, atmospheric, and optical metadata, then refines bundle shifts and rotations by registering spectra to SDSS imaging. This extended astrometry addresses substantial replugging uncertainties before exposures are combined.

  • Basic astrometry: The basic astrometry module combines fiber metrology, dithering, drilling offsets, atmospheric refraction, field distortion, and telescope-optics models.These inputs produce wavelength-dependent positions for each fiber.
  • Basic astrometry: The basic module outputs two-dimensional X and Y arrays giving each fiber’s tangent-plane position relative to the nominal IFU center.The arrays allow the on-sky location to be retrieved at any wavelength for any fiber.
  • Extended astrometry: Replugging introduces approximately 0.5 arcsec centering and 2–3° rotation uncertainties, exceeding the uncertainties from the other modeled effects.The changes cannot be measured directly and vary with cable-routing stresses.
  • Extended astrometry: The extended astrometry module registers synthetic fiber broadband fluxes against SDSS imaging while searching position and rotation offsets relative to the basic solution.An overall flux normalization is also permitted during the fit.
  • Extended astrometry: The calibration fit determines position, rotation, and flux offsets through χ2 minimization, with A centered near 1.00 and σ = 0.037 in i-band across approximately 25,000 IFU exposures.These coefficients indicate about 4% spectrophotometric accuracy relative to SDSS imaging.
  • Extended astrometry: Extended-astrometry precision improves for larger bundles on galaxies with significant azimuthal structure, including a measured approximately 0.5 arcsec IFU-center shift in one example.Smaller bundles on smooth, circular galaxies provide less favorable constraints.
  • Extended astrometry: Iterative fitting averages rotations within each plugging before refitting positional shifts, enabling exposures from different pluggings to share a common astrometric solution.The correction applies automatically because all MaNGA target galaxies lie within the SDSS imaging footprint.

9. DATA CUBE CONSTRUCTION

MaNGA constructs rectified 3-D data cubes by combining astrometrically registered fiber spectra slice by slice with flux-conserving weights, while propagating variance and covariance information.

  • Image reconstruction: 4563 wavelength channels for logarithmic data and 6732 for linear data are reconstructed one image slice at a time using a modified, flux-conserving Shepard method.The inputs contain intensity and variance vectors from all fibers and exposures.
  • Image reconstruction: Weights use fiber-to-pixel distance, exclude bad spectra, and vanish beyond rlim = 1.6 arcsec.The distance weighting has an exponential scalelength of σ = 0.7 arcsec, while binary masking removes zero-inverse-variance inputs.
  • Image reconstruction: Flux conservation is enforced by normalizing weights so their sum is unity for each output pixel.The normalized weights map irregular fiber measurements into regularly gridded image intensities.
  • Image reconstruction: The output intensity at each wavelength is formed from normalized weights and input intensities, with α = 1/(4π) converting fiber-area flux to spaxel-area flux.The resulting values are rearranged into the output image using the known pixel coordinates.
  • Uncertainty propagation: Variance is propagated through the reconstruction, but inverse-variance uncertainties do not determine combined flux values beyond masking zero-weight inputs.The covariance formalism describes correlations introduced between spatially adjacent output pixels.
  • Covariance: Adjacent reconstructed spaxels are strongly correlated, with ρ ≈ 0.85 at 0.5 arcsec separation and ρ < 0.1 at separations of ≥2 arcsec.This covariance affects inverse-variance estimates for spectra formed by coadding adjacent spaxels.
  • Covariance: DR13 provides a rough covariance calibration rather than correlation matrices, which are planned for future data releases.The calibration corrects nominal no-covariance noise estimates using rigorous covariance calculations.
  • Covariance: The true error in combined spectra exceeds estimates that ignore spatial covariance; for Nbin > 100, spaxels become effectively uncorrelated.The calibration is intended for adjacent binned spaxels and is accurate to about 30%.

10. DATA QUALITY

The DR13 data products show well-characterized spatial and spectral performance, accurate wavelength calibration, and strong relative agreement with independent spectra. The pipeline also identifies a systematic underestimate in reported instrumental LSF values that requires correction.

  • Angular Resolution: 2.2–2.7 arcsec g-band reconstructed PSF FWHM is typical across DR13 cubes, with a tail reaching about 3 arcsec.The distribution is based on 1390 galaxy data cubes.
  • Spectral Resolution: 1–2% rms spectral-resolution variability is typical within IFUs, while worst-case large IFUs can vary by 8–10% at blue wavelengths.Red-camera variation is 1% or less longward of 6000 Å, except near the 8100 Å detector discontinuity.
  • Spectral Resolution: 1.10× is a reasonable first-order correction to DR13 LSF estimates, which are systematically underestimated because of pixel integration and wavelength-rectification broadening.The estimated contributions are approximately 4% from pre- versus post-pixellization treatment and 6% from rectification broadening.
  • Wavelength Calibration: 2 km s−1 is the upper bound on systematic MaNGA–SDSS-I wavelength offsets, while individual galaxies show approximately 10 km s−1 Gaussian scatter.The comparison uses bright Hβ, [O iii], Hα, and [N ii] emission lines when quality criteria are met.
  • Wavelength Calibration: 5 km s−1 rms agreement is obtained between repeated MaNGA observations for a small sample with strong emission lines.The estimate is based on approximately 10 repeat observations in DR13.
  • Typical Depth: 23.5 AB arcsec−2 is the blue-wavelength 10-sigma continuum sensitivity in a 5-arcsec aperture, declining to about 20 AB arcsec−2 near strong OH skylines.The corresponding line-flux sensitivities are 5 × 10−17 and 2 × 10−16 erg s−1 cm−2, respectively.

11. SUMMARY

The MaNGA DRP produces calibrated spectra and rectified data cubes, with DR13 products demonstrating strong sky subtraction, calibration, astrometric, and resolution performance while retaining several areas for improvement.

  • Pipeline and products: The DRP operates in two stages, producing calibrated row-stacked spectra and rectified coadded data cubes on linear and logarithmic wavelength grids.The products cover 3622–10,354 Å.
  • Data quality: 1390 DR13 galaxy cubes achieve nearly Poisson-limited sky subtraction shortward of ∼8500 Å.The residual pixel distribution is nearly Gaussian with width set by detector and Poisson noise.
  • Data quality: 1.7% relative calibration between Hβ and Hα and better than 5% absolute calibration over more than 89% of the wavelength range are achieved.The cubes reach µ = 23.5 AB arcsec−2 at 10σ in a five-arcsecond g-band aperture.
  • Data quality: 5 km s−1 rms absolute wavelength accuracy and better than 1 km s−1 rms relative fiber-to-fiber accuracy are demonstrated.These measurements characterize wavelength consistency across the MaNGA data.
  • Data quality: 0.1 arcsec rms astrometric accuracy and 2.54 arcsec FWHM median spatial resolution characterize the reconstructed cubes.The effective reconstructed point-source profile is well described by a single Gaussian.
  • Data quality and limitations: The median spectral resolution is σ = 72 km s−1, while sky residuals, line-spread-function reporting, spatial covariance, and quality control remain improvement areas.The DR13 line-spread functions are effectively under-reported by about 10%, and quality control can be overly aggressive.

A. KEY DIFFERENCES BETWEEN mangadrp AND idlspec2d

mangadrp inherits substantial infrastructure from idlspec2d but introduces MaNGA-specific algorithms and implementations for its hardware, observing strategy, and data-quality requirements.

  • Shared framework: mangadrp and idlspec2d share much of the spectral preprocessing, fiber flatfielding, extraction, and wavelength-calibration framework.The MaNGA implementation modifies inherited methods for its survey design.
  • MaNGA-specific algorithms: MaNGA uses substantially different fiber tracing, including nominal-location cross-correlation and Gaussian-profile centroid refinement.The tracing routine has proven robust against all described hardware failure modes.
  • MaNGA-specific algorithms: The MaNGA scattered-light spline routine is entirely new relative to idlspec2d.It is implemented for bright-time data and flatfields.
  • Survey-specific implementation: Sky subtraction differs substantially because MaNGA’s grouped IFU hardware differs from BOSS’s nearly random distribution of 1000 fibers.Both pipelines use basis splines to build a super-sampled sky model.
  • Survey-specific implementation: MaNGA flux calibration addresses system throughput losses separately from geometric fiber-aperture losses, unlike BOSS’s combined correction problem.The stellar spectral-library comparison is shared, but implementation differs substantially.
  • Quality control: DRP2QUAL is entirely new infrastructure that evaluates frame quality and can stop reduction at multiple points.This provides a MaNGA-specific quality-control layer.

B. MANGA DATA MODEL

The MaNGA data model packages intermediate and final reduction products as structured multi-extension FITS files, with Figure B1 illustrating the principal individual-exposure products.

  • Data model: MaNGA DRP products are delivered as gzipped multi-extension FITS files containing image data and binary-table extensions.The online DR13 documentation provides detailed keyword and data-model definitions.
  • Intermediate products: Figure B1 shows extracted fiber flats, arc-lamp spectra, extracted science spectra, sky-subtracted spectra, and flux-calibrated science spectra.The examples are from the r2 camera and show wavelength-solution curvature along the spectroscopic slit.

B.1. Intermediate DRP data products

The intermediate 2d DRP products record outputs from calibration, extraction, sky subtraction, and flux calibration, including extracted arc frames used for wavelength calibration.

  • Intermediate products: Intermediate products are generated during calibration, flux extraction, sky subtraction, and flux calibration in the 2d DRP stage.Their structures and naming conventions are documented in Tables B1–B6.
  • Calibration products: Extracted arc frames are produced during wavelength calibration in a format similar to the BOSS spArc product.The MaNGA format uses a blank extension 0 and extension names instead of numbers.
  • Data model documentation: The data model documentation is organized under the MANGA/REDUX/DRPVER/PLATE4/MJD5 hierarchy.The referenced model includes the mgArc product description.

B.1.2. mgFlat

The DRP produces flatfield, sky-subtracted, and flux-calibrated frame products through successive processing stages. These products retain camera-specific science spectra and standardized data-model structures.

  • mgFlat: Flatfield frames are produced after fiber tracing, wavelength calibration, and removal of the global quartz lamp spectrum.Their format resembles BOSS spFlat files, except for a blank extension 0 and named extensions.
  • mgFlat: Sky-subtracted frames contain the science fiber spectra for each camera after sky subtraction is applied to mgFrame files.The “S” in mgSFrame denotes sky subtraction.
  • mgFlat: Flux-calibrated frames contain the science fiber spectra for each camera after flux calibration is applied to mgSFrame files.The “F” in mgFFrame denotes flux calibration.

B.1.6. mgCFrame

mgCFrame products combine camera-specific flux-calibrated spectra across the dichroic break and resample them onto common wavelength grids. Both logarithmic and linear wavelength versions are produced.

  • mgCFrame: mgCFrame combines individual-camera flux-calibrated spectra across the dichroic break and appends spectrograph 2 fibers after spectrograph 1 fibers.The resulting spectra are arranged in increasing fiberid order.
  • mgCFrame: Both LINEAR and LOG mgCFrame versions are produced with linear or logarithmic wavelength sampling, respectively.The files use a common wavelength grid across the MaNGA survey.
  • mgCFrame: 4563 spectral elements are provided for logarithmic sampling, while 6732 spectral elements cover the linear wavelength grid.The logarithmic grid spans log10(λ/˚A)=3.5589 to 4.0151; the linear grid spans 3622.0–10353.0 ˚A.

B.2. Final DRP data products

The final DRP products provide row-stacked spectra and regularly gridded three-dimensional cubes in both linear and logarithmic wavelength formats. Associated metadata, quality arrays, exposure records, and reconstructed images support scientific use and traceability.

  • Final products: The DRP provides RSS files and regularly gridded combined data cubes with both logarithmic and linear wavelength solutions.These are alternative final summary products suited to different science cases.
  • Final products: RSS files store row-stacked spectra, whereas cubes use regular 0.5 arcsec square spaxels with wavelength as the third dimension.RSS rows aggregate fibers across exposures; cubes are three-dimensional spatial-spectral arrays.
  • Final products: Final files include inverse variance, pixel masks, OBSINFO exposure metadata, and wavelength- and spectral-resolution information.OBSINFO records exposure number, integration time, hour angle, seeing, and other details for combined exposures.
  • Final products: RSS products include wavelength-dependent fiber positions, while cubes include reconstructed broadband images.Chromatic differential atmospheric refraction gives each wavelength of a fiber a slightly different effective position.
  • Final products: The DRPall FITS file aggregates metadata from individual reduced cubes, spectrophotometric standard stars, and the NSA targeting catalog.Each row corresponds to an individual observation and provides a convenient summary of reduced-data information.

B.4. DRP Data Quality Bitmasks

MaNGA quality bitmasks encode problems in reduced frames, composite RSS files, and final data cubes. They support pixel- and spaxel-level screening as well as object-level quality assessment, with some flags reflecting conservative QA decisions.

  • DRP quality bitmasks: 2D pixel bitmasks identify quality issues affecting individual reduced frames and composite RSS files.They can flag entire fibers or individual pixels, including broken or unplugged fibers, cosmic rays, and sky-subtraction failures.
  • DRP quality bitmasks: 3D spaxel masks summarize the overall quality of individual cube spaxels after combining multiple exposures.They indicate conditions such as no coverage, low edge coverage, and dead fibers.
  • DRP quality bitmasks: The DRP3QUAL integer summarizes the quality of an entire galaxy cube, ranging from low average depth to critical failures.Critical flags mean the data should be treated with great caution or conservatively omitted from science analyses.
  • DRP quality bitmasks: Some critical quality cases may reflect an overly vigorous QA algorithm rather than an intrinsic data problem.Quality flags therefore require interpretation rather than automatic rejection in every case.
  • DRP quality bitmasks: Quality-control bitmasks may gain additional bits over the survey lifetime and can vary across data releases.The documentation provides an online DR13 version and analogous locations for future releases.
Loading 1607.08619v3…