Observing Operations | Reviews | Survey Management

PERFORMANCE REQUIREMENTS FOR THE SDSS PHOTOMETRIC PIPELINE

Steve Kent (Fermilab) and Jill Knapp (Princeton)
January 29, 1998




The following document is an edited version of two parts of "Data Processing Requirements for SDSS Offline Software", written by Steve Kent (January 9 1995; sdss-sac #52).  The first part is the introduction, containing the overall requirements for the entire offline data processing system, and the second is the section dealing with the photometric pipeline. This section has been updated to take into account two changes.  First, we no longer expect to process any data taken by the Fermilab Drift Scan Camera on the 3.5 m telescope, and second, the postage stamp extraction is now done by a new pipeline, the serial stamp collecting pipeline, instead of by the DA system.

Photo now consists of three pipelines:

The SSC, which cuts and orders the five color data to make the frames, and cuts several sets of postage stamps: 65x65 pixel stamps at the positions of stars found in the astrometric chips, 200x200 pixel stamps at the positions of stars brighter than 14m to measure the wings (one star per frame), and every 30 frames or so, extracts an entire frame with a very bright star to measure the image ghosts.  Since SSC is basically a bookkeeping task, there are no requirements per se on this pipeline.

The PSP, which processes the quartile data from the DA and the postage stamp data from SSC and produces the run of sky, flat and PSF data for an entire run.

Frames, which processes the data frame by frame, finds and measures objects and writes a series of output files and images.

Apart from these changes and fixing the occasional typo, no changes have been made to the requirements as presented in the Jan 1995 document, though some outputs have been added to Photo and we could fill in quite a few of the TBDs; the goal is to compare these requirements with Photo's current performance.

The entire software document is included at the end for reference.

==============================================================================

1. DATA PROCESSING REQUIREMENTS FOR SDSS OFFLINE SOFTWARE - GENERAL

[All data processing requirements are subject to instrumental and astronomical limitations.]

[All level 1 software will be tested using simulated data and, if available, test data from existing telescopes and instrumentation.  All level 2 software will be tested using 2.5 m data]

The data processing software shall perform the following:

1. Maintain a survey strategy system to
   a. Define survey layout and all input catalogs needed for data processing
   b. Track survey progress
   c. Monitor quality of processed data and accept or reject

2. Process data tapes from photometric and astrometric imager cameras (excepting focus CCD's) and monitor telescope taken under photometric dark sky conditions with FWHM 0.8" to 1.5" in North Galactic Cap region (worst case l=0, b=30) to produce
   a. Excluding 5% of sky for bright stars, derive positions, parameters, and atlas images in 5 colors for
 i) All galaxies with diam < 2' to m(r')=19 completeness at >95%, contamination of <5%; enhanced goal of all galaxies to m(r')=21
 ii) All detected objects to m(u')=21.9, m(g')=22.5, m(r')=22.5,  m(i')=22.0, m(z')=20.4, with no completeness requirement. Enhanced
    goal: 0.5 mag fainter
 b. Atlas images only for all objects in select catalogs
   c. Corrected pixel map
   d. Systematic photometric errors outside the atmosphere of <2% rms all sky at m(r') and <2% rms in (g-r), (r-i) and <(u-g), (i-z)
       averaged over 1 degree patches.  Enhanced goal of 1% in magnitudes and colors.
 

3. Produce target lists for spectroscopy of
   a. Galaxies (density  100/sq deg), prob. of redshift >95%
   b. QSO candidates (density of 15/sq deg), prob of success >65%
   c. Additional high redshift (high-z) QSO candidates of arbitrary magnitude
   d. Brightest cluster galaxies up to 1 mag fainter than field galaxies
   e. Additional galaxies and QSO candidates for QA purposes
   f. Guide stars for spectroscopy
   g. Reddening standards
   h. Stars with some TBD criteria (treat as serendipitous objects)
   i. Other serendipitous objects
   j. Blank sky
   k. Spectrophotometric standards

4. Produce tiled target lists with coord for drilling machine, errors of
   <260 mas rms, with a turnaround time from delivery of data at Fermilab
   to delivery of drilling coord of 1 week (last 3 years of survey only).

5. Process spectroscopic data tapes and produce
   a. For all galaxies, redshifts with systematic errors of <30 km/s rms
 with success rate of 95% of maximum feasible.
   b. Enhanced goal: for galaxies with S/N per angstrom greater than 17 at
 5000 A, velocity dispersions with systematic errors of 20% rms for
 dispersions greater than 100 km/s
   c. For QSOs, redshifts and other parameters with success rate of 95%
 of maximum feasible with rms errors <1000 km/s.
   d. Enhanced goal: for all spectra, relative spectrophotometry with peak
 systematic error of 10% over range 3800-9000A for objects observed
 at airmass <1.4.
   e. Enhanced goal: For all spectra, all emission lines, peaks, widths,
 and positions.

============================================================================
2.  REQUIREMENTS FOR PHOTOMETRIC PIPELINE

Photometric/postage stamp pipeline will process postage stamp files, calibration files, and 5 color imaging data to produce object lists
and parameters.

INPUT

1. One column (1 to 5 colors) of CCD data interleaved on one or more tapes
2. Astrometric calibration with errors of no worse than 0.5" rms radius astrometric
3. MT photometric calibrations from one 27.3' square field per hour with stars brighter u'=16.3, g'=15.4, r'=15.6, i'=15.2, z'=TBD.
4. Extinction coefficients at approximately 1 set of values per hour
5. Dark drift, quartile, and bad column calibrations
6. Positions, magnitudes, diameters, PA's, and errors for all known globular clusters and known galaxies with diameters > 1'.
7. Positions, magnitudes, and errors for all stars brighter than V=9.
8. Positions and errors for objects in the following catalogs: IRAS FSC, FIRST, ROSAT, 2MASS.

Data will be obtained with the following observing conditions:

1. Seeing: 0.8" to 1.5" FWHM
2. Photometric zero-point fluctuations in one night of < 0.1 mag at r'
   (TBD: This spec needs refinement)
3. Max star density corresponding to l=0, b=30.
4. Sky brightness of nominal dark sky (APO) plus TBD extra (faint moon).
5. Defects: Double column defects may exist anywhere in u' CCD; double columns
   defects in outer 10% of any other imaging CCD except r'; single column
   defects anywhere.

SCIENCE GOALS

1. Excluding 5% of sky for bright stars, derive positions, parameters,
      and atlas images in 5 colors for
 i) All objects to 5 sigma above sky (smoothed over PSF area)
    with no completeness requirement.
 ii) All galaxies with diam <2' to m(r')=19 completeness of >95%,
    contamination of <5%.
 iii) Enhanced goal: All galaxies to 10 sigma smoothed over a galaxy
    diameter, completeness of <90%, contamination of <10%.
2. Positions for all galaxies with diam >2'
3. Enhanced goal: Atlas images for all objects in the following catalogs: IRAS FSC, FIRST, ROSAT, 2MASS.
4. Produce a Corrected pixel map

5. Systematic photometric errors outside the atmosphere of
   a. <2% rms all sky at m(r') and <2% rms in all colors
 (3% u-g and i-z) averaged over 1 degree patches.  Enhanced goal
 of 1% in magnitudes and colors.

   b. Relative stellar magnitudes, within a single frame, to an accuracy of 1% for isolated unsaturated stars brighter than u'=19.9, g'=20.9,
 r'=20.5, i'=20.0, z'=18.4.

   c. Systematic photometric error on individual galaxies of 0.1 mag rms to m(r') = 18.0 and 0.5 mag rms for fainter

   TBD: Spec on how well we can handle crowded objects

RETAINED OUTPUT

1. Postage Stamp Pipeline
   a. PSF parameters, one set per frame per color
   b. Sky brightness in each band
   c. Photometric calibration for each frame

2. Frames Pipeline

   a. Corrected frames.  For each frame, a noise mask (8 bit) and an object mask (8 bit).
   b. For each frame, the correction (FF) and bias vectors
   c. For objects present in a list of catalogues, Atlas Images
      TBD: Size of each image
   d. For each object:

 A unique identifier

 A single reference position, with errors
 Offsets in each colors w.r.t. reference position, with errors

 PSF magnitudes in each band, with errors.

 3" aperture magnitudes in each band, with errors, smoothed
 to largest canonical seeing (1.5 arcsec).

 Centroid, PA of major axis, and ellipticity, derived from weighted
 moments, in each band, within an isophote corresponding to 1 sigma
 (TBD: per pixel?) above sky for the brightest sky (TBD: gotta
 specify these numbers).  Also, parameters for the best fitting
 ellipse at that isophote.

 Radial profile in each band in logarithmically spaced circular annular apertures centred on the canonical position (see above). Each
 aperture will have a mean, a median, and a measure of the variance.  The number of annuli measured will depend on the brightness of the
 object, with the last point corresponding to TBD.

 Petrosian magnitude and radius, and a suitable measure of surface brightness for each band, for each band, inside the r' radius
 corrected for seeing.

 A parameter P that gives the likelihood of an object being a star,
 one per band.

 Morphology index in each band for all objects, with quality measure,
 from radial profile fit to DeVaucouleurs law and exponential profile.

 Atlas Images in 5 colors. (TBD: What is the maximum radius?)

 For all photometric quantities, additional parameters as needed
 to correct for -0.3 to +0.1 mag correction in photometric zero-point
 (reddening plus revised photometric calibration).

 For all parameters, errors

   e. `Mask objects', i.e. structures describing the regions masked and not
      searched for objects, or not searched so deeply, for LSS

   f. For each frame:

      Number of stars, number of galaxies, number of other classified objects,
      number of unclassified objects (`junks'?) per frame, using an
      a priori value of P for each color.

      Zero or one "blank sky" positions, with Atlas image, plus parameters.

   g. For each Big Galaxy (D>2'):

      Position in 5 colors

      Atlas image

   h. No atlas images will be cut for deblended images; rather children
      will be specified either as parameter lists or as 8-bit weight masks.
      No atlas images will be cut for saturated stars in the Bright Object
      catalog.

   i. For each frame:

      A small number of "blank sky" regions, with atlas images, which can be used to locate sky fibers during spectroscopy.

TECHNICAL PERFORMANCE

1. Reduce 8 hours worth of data in all of the 5 filters in 16 hours on an R4400 class cpu in 128 Mbytes or less.

2. Run pipeline on 8 hours of data without needing to be restarted.

3. Reject not more than 0.1% of frames for reasons other than the presence of bright objects  over the entire survey region.

==============================================================================

This is the science software requirements document written by Steve Kent, January 9 1995  The original is sdss-sac #52.  In that message, Steve had flagged changes from previous versions - these flags have been removed for easier reading, but otherwise the document is untouched.  Markup is HTML.
 

==============================================================================

         DATA PROCESSING REQUIREMENTS FOR SDSS OFFLINE SOFTWARE

[All data processing requirements are subject to instrumental and astronomical limitations.]

[All level 1 software will be tested using simulated data and, if available,
test data from existing telescopes and instrumentation.  All level 2 software
will be tested using 2.5 m data]

The data processing software shall perform the following:

1. Maintain a survey strategy system to
   a. Define survey layout and all input catalogs needed for data processing
   b. Track survey progress
   c. Monitor quality of processed data and accept or reject

2. Process data tapes from photometric and astrometric imager cameras
   (excepting focus CCD's) and monitor telescope taken under photometric
   dark sky conditions with FWHM 0.8" to 1.5" in North Galactic Cap
   region (worst case l=0, b=30) to produce
   a. Excluding 5% of sky for bright stars, derive positions, parameters,
      and atlas images in 5 colors for
 i) All galaxies with diam &lt; 2' to m(r')=19 completeness at of &gt;95%,
    contamination of &lt; 5%; enhanced goal of all galaxies to m(r')=21
 ii) All detected objects to m(u')=21.9 m(g')=22.5 m(r')=22.5
    m(i')=22.0 m(z')=20.4 with no completeness requirement. Enhanced
    goal: 0.5 mag fainter
   b. Atlas images only for all objects in select catalogs
      (none extant at present).
   c. Corrected pixel map
   d. Systematic photometric errors outside the atmosphere of &lt; 2% rms all
 sky at m(r') and &lt; 2% rms in (g-r), (r-i) and 3% (u-g), (i-z) averaged
 over 1 degree patches.  Enhanced goal of 1% in magnitudes and colors.

2a. Process data tapes from Drift Scan Camera and 3.5m telescope

3. Produce target lists for spectroscopy of
   a. Galaxies (density of 100/sq deg), prob. of redshift &gt; 95%
   b. QSO candidates (density of 15/sq deg), prob of success &gt; 65%
   c. Additional high redshift (high-z) QSO candidates of arbitrary magnitude
   d. Brightest cluster galaxies up to 1 mag fainter than field galaxies
   e. Additional galaxies and QSO candidates for QA purposes
   f. Guide stars for spectroscopy
   g. Reddening standards
   h. Stars with some TBD criteria (treat as serendipitous objects)
   i. Other serendipitous objects
   j. Blank sky
   k. Spectrophotometric standards

4. Produce tiled target lists with coord for drilling machine, errors of
   &lt; 260 mas rms, with a turnaround time from delivery of data at Fermilab
   to delivery of drilling coord of 1 week (last 3 years of survey only).

5. Process spectroscopic data tapes and produce
   a. For all galaxies, redshifts with systematic errors of &lt; 30 km/s rms
 with succes rate of 95% of maximum feasible.
   b. Enhanced goal: for galaxies with S/N per angstrom greater than 17 at
 5000 A, velocity dispersions with systematic errors of 20% rms for
 dispersions greater than 100 km/s
   c. For QSOs, redshifts and other parameters with success rate of 95%
 of maximum feasible with rms errors &lt; 1000 km/s.
   d. Enhanced goal: for all spectra, relative spectrophotometry with peak
 systematic error of 10% over range 3800-9000A for objects observed
 at airmass &lt; 1.4.
   e. Enhanced goal: For all spectra, all emission lines, peaks, widths,
 and positions.


               REQUIREMENTS FOR PHOTOMETRIC PIPELINE

Photometric/postage stamp pipeline will process postage stamp files,
calibration files, and 5 color imaging data to produce object lists
and parameters.

INPUT

1. One column (1 to 5 colors) of CCD data interleaved on one or more tapes
2. NOT: Binned data from 2.5m telescope
3. Also, Drift Scan Camera 3.5m data (1 color) on one or more tapes
4. Astrometric calibration with errors of no worse than
   0.5" rms radius astrometric
5. MT photometric calibrations from one 27.3' square field
   per hour with stars brighter u'=16.3, g'=15.4, r'=15.6, i'=15.2,
   z'=TBD.
6. Extinction coefficients at approximately 1 set of values
   per hour
7. Dark drift, quartile, and bad column calibrations
8. Up to 40 29x29 excised subframes of unsaturated bright
   stars and other objects per frame per color of CCD data.
9. Positions, magnitudes, diameters, PA's, and errors for all known
   globular clusters and known galaxies with diameters &gt; 1'.
10. Positions, magnitudes, and errors for all stars brighter than V=9.
11. Positions and errors for objects in the following catalogs: IRAS FSC,
 FIRST, ROSAT, 2MASS.

Data will be obtained with the following observing conditions:

1. Seeing: 0.8" to 1.5" FWHM
2. Photometric zero-point fluctuations in one night of &lt; 0.1 mag at r'
   (TBD: This spec needs refinement)
3. Max star density corresponding to l=0, b=30.
4. Sky brightness of nominal dark sky (APO) plus TBD extra (faint moon).
5. Defects: Double column defects may exist anywhere in u' CCD; double columns
   defects in outer 10% of any other imaging CCD except r'; single column
   defects anywhere.

SCIENCE GOALS

1. Excluding 5% of sky for bright stars, derive positions, parameters,
      and atlas images in 5 colors for
 i) All objects to 5 sigma above sky (smoothed over PSF area)
    with no completeness requirement.
 ii) All galaxies with diam &lt; 2' to m(r')=19 completeness of &gt;95%,
    contamination of &lt; 5%.
 iii) Enhanced goal: All galaxies to 10 sigma smoothed over a galaxy
    diameter, completeness of &gt;90%, contamination of &lt; 10%.
2. Positions for all galaxies with diam &gt; 2'
3. Enhanced goal: Atlas images for all objects in the
      following catalogs: (IRAS FSC, FIRST, ROSAT, 2MASS).
4. Produce a Corrected pixel map

5. Systematic photometric errors outsize the atmosphere of"
   a. &lt; 2% rms all sky at m(r') and &lt; 2% rms in all colors
 (3% u-g and i-z) averaged over 1 degree patches.  Enhanced goal
 of 1% in magnitudes and colors.

   b. Relative stellar magnitudes, within a single frame, to an accuracy of
 1% for isolated unsaturated stars brighter than u'=19.9, g'=20.9,
 r'=20.5, i'=20.0, z'=18.4.

   c. Systematic photometric error on individual galaxies of 0.1 mag rms.
 to m(r') = 18.0 and 0.5 mag rms for fainter

   TBD: Spec on how well we can handle crowded objects

RETAINED OUTPUT

1. Postage Stamp Pipeline
   a. PSF parameters, one set per frame per color
   b. Sky brightness in each band
   c. Photometric calibration for each frame

2. Frames Pipeline

   a. Corrected frames.  For each frame, a noise mask (8 bit) and an object
 mask (8 bit).

   b. For each frame, the correction (FF) and bias vectors
   b. For objects present in a list of catalogues, Atlas Images
      TBD: Size of each image

   d. For each object:

 A unique identifier

 A single reference position, with errors
 Offsets in each colors w.r.t. reference position, with errors

 PSF magnitudes in each band, with errors.

 3" aperture magnitudes in each band, with errors, smoothed
 to largest canonical seeing (1.5 arcsec).

 Centroid, PA of major axis, and ellipticity, derived from weighted
 moments, in each band, within an isophote corresponding to 1 sigma
 (TBD: per pixel?) above sky for the brightest sky (TBD: gotta
 specify these numbers).  Also, parameters for the best fitting
 ellipse at that isophote.

 Radial profile in each band in logarithmically spaced circular
 annular apertures centred on the canonical position (see above). Each
 aperture will have a mean, a median, and a measure of the variance.
 The number of annuli measured will depend on the brightness of the
 object, with the last point corresponding to TBD.

 Petrosian magnitude and radius, and a suitable measure of surface
 brightness for each band, for each band, inside the r' radius
 corrected for seeing.

 A parameter P that gives the likelihood of an object being a star,
 one per band.

 Morphology index in each band for all objects, with quality measure,
 from radial profile fit to DeVaucouleurs law and exponential profile.

 Atlas Images in 5 colors. (TBD: What is the maximum radius?)

 For all photometric quantities, additional parameters as needed
 to correct for -0.3 to +0.1 mag correction in photometric zero-point
 (reddening plus revised photometric calibration).

 For all parameters, errors

   d. `Mask objects', i.e. structures describing the regions masked and not
      searched for objects, or not searched so deeply, for LSS

   e. For each frame:

      Number of stars, number of galaxies, number of other classified objects,
      number of unclassified objects (`junks'?) per frame, using an
      a priori value of P for each color.

      Zero or one "blank sky" positions, with Atlas image, plus parameters.

   f. For each Big Galaxy (D&gt;2'):

      Position in 5 colors

      Atlas image

   g. No atlas images will be cut for deblended images; rather children
      will be specified either as parameter lists or as 8-bit weight masks.
      No atlas images will be cut for saturated stars in the Bright Object
      catalog.

TECHNICAL PERFORMANCE

1. Reduce 8 hours worth of data in all of the 5 filters in 16 hours
   on an R4400 class cpu in 128 Mbytes or less.

2. Run pipeline on 8 hours of data without needing to be restarted.

3. Reject not more than 0.1% of frames for reasons other than the presence
   of bright objects  over the entire survey region.


               REQUIREMENTS FOR IMAGING TEST DATA

Positions and magnitudes of artifial stars plus
maging data will be generated to simulate two 1-hour interleaved
strips at survey lat=long=0 degrees plus one 1-hour scanline at
l=0, b=30.

INPUTS

1. Positional standard stars from the Guide Star Catalog and the ACRS.

2. Realistic colors for Astrometric(/MT) standard stars
   (based on Bahcall-Soneira model).

3. Actual CCD dark images with SDSS electronics (with the same exposure time,
   real noise and gain).

4. Calculated ghost patterns (TBD: From where?)

5. Real galaxy images

OUTPUTS

1. Catalog (AKA the Mamoru/Weinberg catalog) of galaxy, star, and QSO positions,
   stars brighter than magnitude r'=25 (2 mag fainter than survey limit
   in each color), galaxies brighter than as above, QSOs
   brighter than as above, star densities corresponding to |b| &gt; 30 degrees.

2. Photometric CCDs, 12 scanlines, each 1 hour in length, 5 colors.

3. Astrometric frame data (22 CCDs per strip)

4. Monitor telescope overlap patches (Two sets of 6 fields each at beginning
   and end of above scans)
 

5. Monitor telescope primary standard star frames (10 stars total, each in
   5 colors)

SIMULATED EFFECTS

1. Bad columns
2. Variable sky levels and psf, up to limits required
3. Minor planets
4. For QSO: See section below on QSO TEST DATA
5. "Airplane trails" and meteor trails
 

TECHNICAL REQUIREMENTS

1. Headers: (need requests from pipelines).
2. Consistent observational parameters (Date&amp;time, sec z, telescope angle)
3. Code is distributable and documented; capable of generating "small"
   volumes of simulated data
4. Enhanced goal: Convert from Mirage to Dervish

OBSERVATIONAL TEST DATA

1. Actual CCD frames with
 a. Cosmic rays
 b. Ghosts, satellite trails, bleed trails
 c. Big bright galaxies &gt; 2', clusters of galaxies (redshift 0.03 to 0.2)
    merging groups, planetaries, etc.

2. Bright star images
  Ghosts, PSFs as a function of time for 3.5 telescope.
  CCD response including bleed trails, CCD hysteresis and (low) CTE.
  Big objects (nebulae, Galactic clusters, galaxies,
               clusters of galaxies, satellites, minor planets...)

3. DSC scans of 3 x 3 degree pieces of sky at at least 2 different directions
   in the survey area with different stellar densities. (Enhanced goal:
   one position at NGP, one position at l=0, b=30).  MT overlap data for
   above.

4. DSC scans of HST medium deep survey fields.

5. DSC scans of fields selected from the IRAS FSC


       REQUIREMENTS FOR SPECTROSCOPIC PIPELINE

The spectroscopic pipeline will accept as input:

For up to 10 fields per night,

1. For each field, one or more sets of CCD frames,
   each set consisting of two pairs of blue/red CCD
   frames of 320 fiber spectra each, of stars, galaxies,
   QSOs, sky, scattered light, and serendipitous objects

2. Flatfields

3. Bias frames

4. Calibration arc frames

5. Absolute fluxes in 5 bands in 3" aperture (from
   photometric pipeline) and positions
5a. Enhanced requirement: 5 absolute fluxes per band,
   including 4 sets of exposures offset by 1.5" N, S, E,
   and W of the nominal target position (from photometric
   pipeline)

6. Enhanced requirement: Four sets of exposures with
   the telescope offset by 1'' N, S, E, and W of the
   nominal target position

Data will be accepted under the following observing
   conditions:

1. Seeing 2" FWHM or better

2. Sky brightness of nominal dark sky (APO) plus moon
        no brighter than V = 19th mag/sq arcsec

3. Variable transparency

4. Defects: Single column defects

SCIENCE GOALS

1. For all galaxy candidate spectra:
   a. Measure parameters and redshifts with total rms
        errors less than 30 km/s and completeness of 95%
        of all feasible.
   b. For all galaxy candidate spectra with detected
        flux greater than m(r')=19, a velocity dispersion
        with systematic errors less than 75 km/s rms
        and 50% completeness
   c. Enhanced goal: velocity dispersion error of 10%
        above 200 km/s
   d. Enhanced goal: 90% completeness

2. For all QSO candidate spectra:
   a. Identify QSO spectra as such with a completeness
        of 95% and an error rate of 5%.
   b. Measure parameters and redshifts of QSOs with
        systematic errors less than 100 kms/s rms and
        a completeness of 90% (enhanced goal 95%) of
        all feasible.

3. Enhanced goal: For all object spectra, obtain a
        spectrophotometric calibration with relative
        systematic errors less than 10% maximum.

3a. Enhanced goal: For all object spectra, obtain a
        spectrophotometric calibration with absolute
        systematic errors less than 10% maximum.

4. Enhanced goal: For all stars, identify spectral type and radial velocity

RETAINED OUTPUT

1. Corrected 2D frames
        Storage memory: 57GB
        (10^6 x 2 color x [2192x2068 pix] x 2 bytes/ 320 fbrs)

2. For all object spectra, a corrected rebinned red-blue
   combined 1-D spectrum with errors, photometry, wavelenght,
   dispersion relation, and continuum fit (DPS says no to continuum fit).
   No covariance arrays.
        Storage memory: 41GB
        (10^6 x 4096 pix x 10 bytes)

3. For all galaxy candidates:
   a. An absorption line redshift with r value, a peak
 likelihood estimate, z and r for second highest
 peak best fit template id; Enhanced goal: velocity dispersion
   b. Enhanced goal: For each absorption line, wavelength, FWHM, equivalent
        width, and errors.
        Storage memory: 0.5GB

4. For all galaxies and QSO candidates:
   a. For each emission line, wavelength, FWHM, equivalent
        width, and errors. (QSO's: This is enhanced goal only).
   b. A redshift and estimated error with total rms error
        &lt; 30 km/s for galaxies and &lt; 1000 km/s for QSOs,
        using both QSO and galaxy emission line lists

5. Enhanced goal: For all object spectra, a spectrophotometric
   calibration in the form of polynomial coefficients giving a
   function going from wavelength or pixel to sensitivity (TBD)
   (counts)/(ergs/sec/cm^2/Ang).

6. For each set of 4-point frames, extracted 1-D spectra.

TECHNICAL PERFORMANCE

1. Process all frames for one field in 3 hours elapsed time
   without failure.

COMPUTER MEMORY

1. Input data: 320MB scratch disk
        For 1 spectrograph, 1 color, 320 fibers, need 9 input
        data frames, each 2192x2068 x 2bytes = 9MB:
          2 He-Ne-Ar arc lamp calibrations
          1 bias (avg of many)
          1 flat field (avg of many)
          2 `nodding frames' or
             4 `diamond pointings'
          3 data frames
          -------------
          9 frames x 9MB = 81MB
          (these are read from the data tape off the mountain
          to spinning disk)
        IRAF runs on these frames, generating about 4 times
        this in temporary storage: 4x81 = 324MB

2. Analysis: 70-140 RAM memory
        2-D to 1-D IRAF extraction needs 8 x 2192x2068 floating
        points in memory at once = 145MB max. If pushed, we may
        be able to do this in 70MB RAM.

               REQUIREMENTS FOR SPECTROSCOPIC TEST DATA

Test data will be generated to simulate observations of 4
spectroscopic target fields.

Spectra shall be simulated for the following types of objects:
   a. Sky (10 fibers per field)
   b. F subdwarf (2 fibers per field)
   c. Galaxies and QSO's from Mamoru/Weinberg simulations.
   d. Stars (chosen from white dwarfs (low res), M star,
        K giants/dwarf, A, B F G stars, 10 to 100 per field)
 

INPUTS:

1. Mamoru/Weinberg catalog
   Enhanced goal: Extract the above file from a survey database
   which contains photometric pipeline output of a photometric
   simulation.

2. Filippenko spectrophotometric atlas of galaxies

3. QSO simulator.

OUTPUTS

An "exposure" is a set of 4 FITS format CCD frames, corresponding
   to blue &amp; red channels for 2 spectrographs, with 320 fibers per
   frame.

Each target field will have the following sets of exposures
   a. A flatfield exposure
   b. A bias exposure
   c. A comparison lamp exposure
   d. 3 exposures of the targeted field
   e. Documenting files
   f. Enhanced goal: 4 binned exposures based on a "diamond" raster
      pattern

A TCL interface and documentation to run the simulator.
 

SIMULATED EFFECTS
 
1. Sky continuum and prominent night sky emission lines,
   plus prominent telluric absorption features ( 7600 A band,
   6880 B band 8200 H2O band)

2. Optionally, photon noise and read noise.
   a. Spectra may have regions of unusually high noise
      due to low signal, telluric abs, or near edge of chip.

3. Spectra should have resolution and sampling similar to the
   expected spectrograph resolution over the full range of
   wavelength coverage (3900-9000A). 1.5A/pixel sampling, 3A
   resolution. This restriction will be relaxed somewhat for
   'spectra obtained from the literature' since little existing
   data has that large wavelength range coverage at that
   resolution.

4. Object types:

   a. Blank noise (broken fiber or bias/flat image)
   b. Galaxies
 spiral, elliptical, irregular, em-lines
 redshifts from -0.01 to 1.
   c. QSOs
        range of spectral indicies for continuum slope
        broad emission lines O VI, Ly-Alpha, CIV, OIV, NV, MgII, [CII,
        [CIII, Mg II, Hbeta, Halpha, siIV, etc
        narrow absorption features including ly-alpha forest,
 damped Ly alpha systems, Lyman-limit systems, C IV and Mg II doublets
        of varying depths.
        redshifts from 0 to 7. At high redshifts, deep Ly-alpha
        BAL spectra
        Continuum slope -2.5 to +0.5 f_nu
        absorption and Gunn-Peterson depression of continuum
        should be included.
        Include reddening from local QSO environment
 Redshifted wavelength range 3000 to 11000 A
   d. Stars
        Range of spectral types, O,B,A,G,K,M,KI,KV, etc
        Extended goal: White dwarf and peculiar stars, WD+red dwarf
        binaries redshifts +/- 1000 km/s

Templates for these types to be drawn from the existing literature
or may be generated 'from simple models' (black bodies, power laws
plus gaussian absorption and emission feature and noise.

5. Specific to 2-D simulations:
   a. Pincushion distortion up to 6 pixels top to bottom.
   b. Broadening of spectra with position (spectra are broader near
      corners of chip)
   c. Broken fibers
   d. Cosmic rays
 

6. Specific to 1-D simulations:
   a. Full length red and blue joined spectra to be simulated
   b. with or without sky lines.
   c. with or without residual cosmic rays.

               REQUIREMENTS FOR QSO TEST DATA

1. Simulated QSO spectra:
   a. The simulations consist of a QSO power law continuum
 of slope f(nu)=nu^(alpha) where alpha ranges from -2 to 2,
 centered at -0.7.
   b. The continuum is modified by a ly-alpha forest of
 depth which increases as a function of z
   c. The spectra are modified by strong broad emission lines of
 Ly-alpha, Ly-beta, OVI, NV, CIV, MgII, NeIII, CII],
 CIII], etc.  The lines have varying equivalent width
 from 10 to 1000 Angstroms, the ratios of the
 equivalent widths vary to a smaller extent.
   d. Redshifts z=0 to z=7.
   e. The spectra are modified by a set of QSO absorption lines  at
 intermediate redshift, such as CIV and MgII doublets.
   f. Some variable amount of internal and external reddening
 should be added to the QSO spectra, up to several mags
 of reddening for very high redshift objects.
   g. Resolution 3 Angstroms,  sample at 1.5 A/pixel over 3900-9000Angstroms.
 
2. Simulated QSO colors:
   a. The above spectra shall be used to generate QSO magnitudes in 5 bands
 Extrapolations shall be made for u and z filters which are
 mostly outside the 3900-9000A spectral coverage range.
   b. Check  simulated colors against existing QSO spectra and color catalogs
 to determine if they are realistic in the regions where QSOs are
 known to exist. Additionally, existing QSO spectra and colors from the
 literature are to be converted to the SDSS filter system
 to investigate scatter in colors and relative density
 of QSOs in different parts of the color-color-color cube.

 REQUIREMENTS FOR ASTROMETRIC PIPELINE

The astrometric pipeline shall:

1. Provide astrometric calibration for photometric (and Monitor Telescope)
   frames, on a frame-by-frame basis, with an accuracy (excluding errors
   due to differential refraction) such that
   a. For photometric frames, the rms error of the position of any point over
 any contiguous 3 degree field in the survey area is no worse than 260
 milliarcseconds (mas), and,
   b. For Monitor Telescope Secondary Field frames, the rms error of the
 position of any point in the frame is no worse than 2 arcseconds.

   Extremes:   all of survey area; airmass &lt;= 1.7; scan lengths &gt;= 20 minutes,
   seeing &lt;= 1.5 arcsec

   Enhance goal: For photometric frames, rms error of no worse than 80
 milliarcsec; no requirement on turnaround time

   INPUT DATA
   a. CCD star positions and magnitudes from astrometric, photometric,
 and Monitor Telescope ccds (frame-by-frame)
   b. Telescope pointing/timing information and ambient conditions
 for each frame
   c. Catalog star positions from astrometric reference frame catalog(s)

   OUTPUT DATA
   a. Translation coefficients (row/column to coordinates on the sky)
 for each frame

2. Provide capability to compute corrections due to astrometric errors
   introduced by aberration and atmospheric differential refraction, over
   a 3 degree field, for representative objects of given Survey colors, to
   an accuracy of 250 mas.

   Extremes: all of survey area; airmass &lt;= 1.7

   INPUT DATA
   a. Observing conditions (ambient temperature, barometric pressure,
 relative humidity)
 Telescope coordinates and length of observation
 Bandpass of observations
 Spectral energy distribution of objects (e.g., Survey colors)

   OUTPUT DATA
   a. Corrections for apparent position as function of object color and
       position in focal plane

3. Provide the capability to predict the effects telesopce optics has on
   astrometric accuracy as a function of position in the focal plane, color of
   objects, and observed bandpass.

4. Provide the capability to evaluate and monitor the rigidity and the overall
   astrometric capability and stability of imaging camera observations.

   Extremes: all of survey area; airmass &lt;= 1.7

   INPUT DATA
   a. High density (circa 40 stars per square degree) star catalog of
 calibration patches with accurate (circa 35mas) coordinates.
   b. CCD star positions and magnitudes from astrometric and photometric
 ccds (frame-by-frame) taken specifically for astrometric
 calibration purposes at a frequency TBD.
   c. Telescope pointing/timing information for each frame

   OUTPUT DATA
   a. Translation coefficients (row/column to coordinates on the sky)
 for each frame

5. Provide a suite of astrometric software routines for survey-wide use, with
   tcl interfaces (e.g. coordinate transformations, precession).

6. Maintain catalogs of astrometric reference frame objects in a format
   suitable for insertion into survey database. Provide capability to extract
   reference frame objects for survey bricks and strips.  Include the
   following catalogs:
   a. ACRS
   b. High density astrometric calibration patch star catalogs
   c. USNO twin astrograph catalog (when available)
   d. Hipparcos Input Catalog (and final Hipparcos catalog when available)
   e. Tycho Input Catalog (and final Tycho catalog when available)

 REQUIREMENTS FOR MONITOR TELESCOPE PIPELINE

The monitor telescope pipeline shall accept multiple nights of CCD frames
of primary and secondary photometric standard star fields and produce
calibrations sufficient to calibrate an arbitrary secondary standard star
with an rms error of 1% per filter and 1% in any color averaged over
1 field.  The pipeline shall also accept comparable data for primary and
secondary spectrophotometric standard stars and shall achieve an rms
accuracy of no less than 3% per filter.

INPUT

1. A set of primary standard stars with photometry in 5 colors.

2. Sets of primary standard star CCD frames, each field in 5 colors
   approximately 6 fields per hour.

3. Sets of secondary star CCD frames, each field in 5 colors,
   approximately 3 fields per hour.

4. Bias and flatfield frames

5. Sets of primary spectrophotometric star CCD frames, each field in
   12 colors, approximately 3 fields per hour

6. Sets of secondary spectrophotometric star CCD frames, each field in
   12 colors, approximately 2 fields per hour.
 

CONDITIONS

1. Pointing errors 15" rms
2. Primary standard star field: highest star density l=0, b=20,
   standard star no fainter than V=12.5
   no other star brighter than V=9, no other star brighter than V=18 closer
   than 10"
3. Photometric secondary star exposures 1 minute (2 in u'), no stars
   brighter than V=10, highest star density
   l=0, b=30.
4. Spectrophotometric secondary star exposures 1 min sec (2 in u'), no stars
   brighter than r=15, highest star density l=0, b=30.
   (TBD: This needs further thought)
5. FWHM &lt; 5"
6. Air mass &lt;= 2.2

OUTPUT

1. Three parameters per filter:
   a. Zero point offset, one per night
   b. Color term, one per month
   c. Extinction coefficient one per hour

2. Calibrated secondary star magnitudes and instrumental positions


   REQUIREMENTS FOR SURVEY DEFINITION/SURVEY PROGRESS

Survey Definition will meet the following requirements:

1. Maintain a description of the North imaging survey area.  The area
   will be defined by:
   a) Survey Pole = 12h 20m, 32.8d
   b) Survey limits satisfy A(V) &lt; 0.3, dec &gt; -22 degrees; dec &lt; 88 degrees.
   c) Scanline pattern defined with overlap of 1.0 arcmin; stripes overlap by
 3 arcmin.
   d) Survey longitude limits for each stripe to be used for target selection.

2. Construct a list of know objects.  Include:
   a) All stars with V &lt; 14
   b) All globular clusters
   c) All known galaxies with diam &gt; 1 arcmin or V &lt; 14, |b| &gt; 25,
        dec &gt; -22 deg.
   d) All planetary nebulae
   e) All reflection nebulae
   Parameters:
   a) ra, dec with errors no worse than 6 arcsec (absolute? rms?)
   b) V magnitude with unspecified error
   c) For galaxies, globular clusters, planetaries, and reflection nebulae,
 position angle in degrees; major and minor axis diameters in arcmin
   d) B-V for stars, if known.
   Required query capability
   a) Position (ra/dec and great circle coordinates for survey-defined
 great circles)
   b) Magnitude

3. Construct a list of primary astrometric standards:
   a) Density &gt; 2 star/deg^2 at all positions in the sky
   b) J2000 position errors &lt; 0.3 arcsec rms per coord.
   c) Covers sky in North and South survey region
   Parameters:
   a) Ra, dec, errors, proper motions, errors, epochs
   b) V mag
   c) B-V
   d) Identifier
   Required query capability
   a) RA and DEC
   b) Great circle coord for survey defined great circles
   c) Magnitude

4. Construct a database of high density astrometric standards:
   a) Density &gt; 20 star/deg^2 at all positions
   b) J2000 position errors &lt; 0.1 arcsec rms per coord.
   c) Covers 4 regions 2.5 deg square spaced along celestial equator
   Parameters:
   a) Ra, dec, errors, proper motions, errors, epochs
   b) V mag
   c) B-V
   d) Identifier
   Required query capability
   a) RA and DEC
   b) Great circle coord for survey defined great circles
   c) Magnitude

5. Construct a list of primary photometric standards:
   a) 50 standards spaced uniformly between dec = 0 to 60 deg and
        |b| &gt; 25 deg
   b) Distribution over stellar types O5 to M0(V) or M0(III)
   c) Must not saturate MT camera and have better than 1% statistics for
        exposure times 5-10 s (g' to z') or 5-20s (u')
   Parameters:
   a) Ra, dec, errors, good to ?? arcsec
   b) 5 standard colors
   c) Identifier
   d) Relative positions and r' mags for all stars r' &lt; 15 within 15' of
        standard to accuracy of 0.2 arcsec rms
   Required query capability:
   a) RA and DEC
   b) Airmass, give a date and TAI
   c) g'-r'

6. Construct a list of secondary photometric patches with the following
   properties:
   a) 22 arcminute square patches distributed on pairs of adjacent scanlines
      in each survey stripes with a mean density of one patch per hour
   b) No stars with v &lt; 10
   c) Enhanced goal: No patch with nearby stars, such that radius &gt; (2200
 arcminutes) * 10 ^(V/5).
   Required query capability:
   a) position (ra, dec and great circle coordinates)

7. Contruct a list of spectrophotometric calibration standards for primary
   spectrophotometric flux calibration with the following properties
   a) 10 F subdwarfs located with |b| &gt; 25 degrees
   b) Magnitude range: 8 &lt; V &lt; 10
   Parameters:
   a) ra, dec, and errors, with positions better than
      1 arcsecond.
    o spectrophotometric calibration with &lt; 5% peak error

8. Construct an a priori reddening map with the following properties:
   a) Coverage |b| &gt; 25 degrees
   b) Provides extinction in 5 colors
   c) Angular resolution of 5 degrees (TBD: Can we do better?)
   d) Maximum error at r' of 0.15 mag

Survey progress will provide a user interface to view and maintain the
following:

1. Status of completed photometric runs. The status information will be:
   a) In an arbitrary range of survey longitude per strip: data exists
 (including repeat observations); data accepted/rejected/no decision;
 data has quality factor Q.
   b) In an arbitrary range of survey longitude per stripe: data tiled.
   c) In an aribtrary tile, list of targets encoded by type

   INPUT: 2.5m imaging observing report; MT, astrometric, photo data processing
 reports; tiling report

2. Status of completed MT primary star and overlap patches.  The status
   information will be:
   a) For primary standards, number of times observed.
   b) For secondary patches, number of times observed; data accepted/rejected/
 no decision.
   c) For spectrophotometric primary standards, number of times observed;
 data accepted/rejected/no decision.

   INPUT: MT observing report; MT data processing report

3. Maintain list and status of spectroscopic tiles and plates
   a) For tiles, maintain field centers; list of targets; guide stars
   b) For plates, maintain id; hour angle and temperature ranges; number of
 times observed; data accepted/rejected/no decision

   INPUT: Tiling report; spectroscopic observing report.

   Required query capability:
   a) ra and dec;
   b) plate identifier
 

4. Provide ability to view one or more of the following items:
   a) Northern and/or southern survey boundaries.
   b) Reddening map
   c) Location of secondary photometric patches with completion status
   d) Photometric run status with completion status
   e) Spectroscopic status (by tile) with completion status


 REQUIREMENTS FOR TARGET SELECTION

Target selection will take objects identified in the photometric pipeline
and generate coordinate lists for plate drilling.

1. Final Object List: merge multiple detections of objects from
   individual runs and apply astrometric and photometric calibrations and
   reddening corrections.  The following operations will be provided:
   a) Match multiple detections of same object in arbitrary independent
 runs (TBD: This may be a QA function instead?)
   b) Flag which of multiple detections to be used for target selection
 based on: survey coordinates; quality factor Q of each run
   c) Apply astrometric, photometric calibrations.
   d) Apply reddening corrections

   INPUTS:
   a) A specified area of sky or set of runs
   b) A list of objects with instrumental parameters
   c) A set of current calibrations

   OUTPUTS: Final Object List
   a) Ra, Dec (J2000) with errors
   b) Petrosian magnitude in r', corrected for reddening
      correction applied and error
   c) PSF magnitude in r', with reddening correction applied and error
   d) Colors (u'-g', g'-r', r'-i', i'-z') with reddening correction applied
 and errors
   e) Surface brightness in r', with final calibration and reddening
      correction applied and error
   f) P parameter in r' band for star/galaxy separation
   g) Ad-hoc target designation flag, for serendipity (TBD: Additional
 parameters needed by serendipity?  Where does the flag get set?
 Whose job is it to look for coincidences with other catalogs?)
   h) BCM target designation flag (TBD: Who does the cluster finding and
 identification?  Where does the flag get set?)
   i) Reddening correction for each color
   j) r' logarithmic raidal profile

   Required query capability: All of above parameters

1a. Enhanced goal: (For serendipity)
   Created cross-identifications with external catalogs.  Merge parameters
   from those catalogs with final object list.

2. Target Selection: Create a list of target candidates with the following
   criteria:
   a) Galaxy: all objects with reddening corrected Petrosian magnitude in r'
      &lt; 18; SB(r') &lt; 22; star/galaxy shape parameter P &lt; some value.
   a') Big galaxies: All
   b) QSO: all objects with reddening corrected PSF magnitudes and colors
      satisfying criteria defined by the QSO working group
   c) High z QSO: all objects with reddening corrected PSF
      magnitudes and colors satisfying color criteria defined by the QSO
      working group
   d) Brightest cluster member: all objects with the BCM target selection
      flag set
   e) Blank Sky: As selected by Photo.
   f) Reddening Stars: shape and color selection criteria as defined by the
      stars working group
   g) Spectrophotometric Standards: all objects with PSF magnitude
      14 &lt; r &lt; 16; star/galaxy shape parameter P &gt; some value; colors
      satisfy TBD F subdwarf requirement defined by the stars working group,
      no stars with V&lt;9 within 22' radius.
   h) Guide Stars: have PSF magnitude in 14 &lt; r &lt; 15; star/galaxy
      shape parameter satisfies the "star" requirement defined by the stellar
      working group; colors TBD by USNO
   i) QA: selected with criteria with the algorithm for galaxy and QSO
      positions, but with the parameters set looser (TBD: What does this
      mean?)
   j) Serendipitous: all objects with the ad-hoc target selection flag set
   k) Stars: shape and color selection criteria TBD
   l) Bright stars: All cataloged stars with V &lt; 9

3. Tiling: Group target candidates into "tiles" of diameter 3.0 deg
   and 640 fibers optimized to minimize time-to-completion of survey.

   The following operations will be performed:

   a) Create Tilable Object List: Combine the lists of galaxy, QSO, High z QSO,
 and BCM targets and reconcile close pairs as follows.  Scan for pairs
 of targets closer than 50 arcseconds. Eliminate one of member
 by a random lottery, except that high Z QSOs will always be retained.
 The excluded target is put at the top of the "as fibers are
 available" list.
   b) Tile: Assign each tile to a position
      based on the local density of targets, assign the plate centers. Use an
      algorithm that optimizes the time to completion for the survey. Note
      that the efficiency of the algorithm may be affected by the shape of the
      region. We allow 10^-3 of the targets to be missed when their
      inclusion seriously affects the tiling efficiency. The central 3.2mm of
      the tile can not contain a target. Plate "jiggling" may be used
      to meet this requirement.  The plate diameter is 3 degrees.

   INPUTS:
 i) Tilable object list (Objects within 1 degree of the outer boundary
 of the photometric survey are not required to be selected)
    ii) Number of tiles to be used
 iii) Number of fibers reserved for tilable objects (561) (TBD: Is
        this OK?)

   OUTPUTS (tiling report):
 i) Ra, Dec (J2000) for each tile
    ii) List of objects per tile

   c) Assign reserved targets to tiles as follows:
 i) Blank sky: 20 (Uniform distribution)
 ii) Reddening stars: 5 (TBD OK?) (Uniform distribution)
 iii) Secondary spectrophotometric standards: 3
 iv) Tracking stars: 10 total (no fibers)
        v) Sky monitor fiber: 1 (special purpose fiber)
   d) Assign targets to free fibers: The "as fibers are available" targets
 are used with the following priority:
 i) Targets excluded by the minimum separation requirement
 ii) Remaining targets with the following percentages: stars 26%;
         serendipity 12%; QA 62%
   e) Light traps for bright stars: TBD: How will we deal with these at
 plugging time???????

4. Plate Design: Convert celestial coordinates to rectilinear coordinates
   for drilling with rms mapping error no worse than 20 marcsec rms.

   INPUT:
   a) Ra, Dec for plate target list
   b) Ra, Dec for plate center
   c) Temperature, Hour Angle, date
   d) Exposure time

   OUTPUT (Drilling plan):
   a) x, y position in mm
   b) Temperature range
   c) Hour angle range
   d) Date range

   Correct for the following effects:
   a) Aberration
   b) Atmospheric refraction
   c) Optical mapping from sky to focal plane
   d) Plate bending distortions;
   e) Temperature (TBD: where does this occur?)


 REQUIREMENT FOR QUALITY ANALYSIS

QA will provide a mechanism to monitor the scientific and technical performance
of all systems.

QA will perform the following activities:

1. Match multiple detections of same object in arbitrary independent
 runs (TBD: This may be a target selection function instead?);
 compute average and residuals in position, mag.

2. Match object detections against arbitrary input catalogs.  Each
   catalog provides the following information: Ra, Dec (J2000), error,
   magnitude in some band; compute average and residuals in position, mag.
 

3. Query all retained output for the following information:
   a) All retained instrumental quantities
   b) All reduced quantities
   c) All internal quantities
   d) For an arbitrary object, return observing conditions (TBD: what are
 needed? Moon phase is not kept anywhere)

4. Present graphical interface to the following quantites:
   a) For an arbitrary run, seeing, mean sky level, # detected stars,
 # detected galaxies per frame

5. For objects with spectra, comparison of flux derived from spectrum vs.
   magnitude from photometry
 
 

 
Submit comments and questions to the SDSS Helpdesk. Last updated 04/01/14