Harvard-Smithsonian Center for Astrophysics

Sloan Digital Sky Survey at the CFA Data Release 6

GETSPECID - get spectra from archive

The getspecid program, located in /data/oiropt/bin, expects a starbase table as its standard input and decodes the specObjID column to determine the plate number, date (as MJD) and fiber ID. If the specObjID is zero for a given row, that row is ignored. Getspecid then locates the appropriate multi-spectrum FITS file in the Spectro directory hierarchy and writes that fiber's spectrum to a FITS file in the user's area. There are a number of options that may alter this default behavior, including using the command line arguments instead of standard input.

  Usage: getspecid [options] [specID_1 specID_2 ...]

     If non-option args are present, they are taken as specObjIDs
     to be decoded. The -t, -l, -c, and -a flags are then ignored.

  Options:
        -t      - Take specObjID's from STDIN as starbase table (default)
        -l      - Take specObjID's as simple list from STDIN, one per line
        -d .    - set output directory path
        -f <id> - use fiberID=<id>
        -m <mj> - use mjd=<mj>
        -n      - Print source file path+name, fiberid only
        -s      - Print source file name, fiberid only
        -p <pl> - use plate=<pl>
        -c specObjID    - Column name to use from starbase table
        -o <name>       - Output file name instead of default.
        -a /data/astrocat/SDSS-dr6/Spectro      - Path to SDSS data
        -v      - verbose output
        -x      - Print this message and exit.

   Output is file in directory "dirpath", which is "." by default.
   File name is of form: spPlate-<plate>-<mjd>-<fiberID>.fits

Example:

To gather candidate Hypervelocity Stars for his project, Warren Brown (see Brown et al.) searched the SDSS catalog for late B-type objects. An approximate replication of the search would be the following script ( see explanatory notes):

#! /data/oiropt/bin/ksh
echo 18.25 |
fldtotable psfMag_g |
search  /data/astrocat/SDSS-dr6/PhotoPrim.db -C1 psfMag_g 1.25 |
row 'type == 6 &&
     and(flags, 0x03000810U) == 0x10U && and(flags2,0x80441044U) == 0' |
row 'psfMag_g_r > -0.38 && psfMag_g_r < -0.28'    | 
row ' (2.67 * psfMag_g_r) + 1.30 < psfMag_u_g && 
      psfMag_u_g < (2.67 * psfMag_g_r) + 2.0'     |
row ' -0.5 < psfMag_r_i && psfMag_r_i < 0.0   &&
      17.0 < psfMag_g   && psfMag_g < 19.5'       |
column -a l b                                     |
compute '[ l, b ] = eqgal(ra/15.,dec); if(b > 180.) b = (b % 180)-180' |
row 'abs(b) > -l/5. + 50 || abs(b) < l/5. - 50.'

Note that the multiple row statements are really for clarity and could be combined into a single command. Running this script (call it hvsearch) as:

hvsearch >hvstars.db

generates hvstars.db with 563 rows. That table can be piped straight into getspecid to pull out all the objects with observed spectra:

getspecid -v <hvstars.db
 Extract fiber 304 to ./spPlate-0339-51692-304.fits
 Extract fiber 261 to ./spPlate-0332-52367-261.fits
 Extract fiber 175 to ./spPlate-0372-52173-175.fits
   ...
 Extract fiber 373 to ./spPlate-0543-52017-373.fits
 Extract fiber 358 to ./spPlate-0439-51877-358.fits
 Extract fiber 533 to ./spPlate-0497-51989-533.fits

for a total of 39 spectra, of which only 2 do not appear to be late B-type stars.

A test run of this script took about 3 hours 52 minutes on the tdc computer, of which all but 7 minutes was the initial search with the condition 17.0 < psfMag_g > 19.5, which passed 6.8 million rows on to the pipeline of row statements.

Some notes on the hvsearch script:

[getobjid] [getspecid] [footserver] [sdssquery]