The HST Calibration Pipeline
Before data are archived in MAST, it passes through the data calibration pipeline. The pipeline itself is implemented as workflows in HTCondor with the in-house Open Workflow Layer (OWL) managing the jobs that are submitted for processing. A typical workflow for a new science association is shown in Figure 1. The process names and descriptions are detailed in Table 1.
|dan_receipt||Science data receipt|
|INGEST||Archive raw exposure|
|2FITS||Convert raw data format to FITS|
|RF||Determine best reference files|
|INGEST_SCI||Archive calibrated science data|
|PVW||Create preview files|
|INGEST_PVW||Archive preview files|
|Table 1: New science calibration pipeline steps|
Record of Pipeline Steps Taken
The steps taken by the calibration pipeline are recorded in several of the FITS files delivered by MAST. As an example, take the ACS association J9CV58020 made from the exposures J9CV58TZQ and J9CV58U1Q. In the primary header for the raw data files (j9cv58tzq_raw.fits and j9cv58u1q_raw.fits), there is a section of keywords with the header "/ CALIBRATION SWITCHES: PERFORM, OMIT, COMPLETE". A series of keywords/values follows:
DQICORR = 'PERFORM ' / data quality initialization ATODCORR= 'OMIT ' / correct for A to D conversion errors BLEVCORR= 'PERFORM ' / subtract bias level computed from overscan img BIASCORR= 'PERFORM ' / Subtract bias image FLSHCORR= 'OMIT ' / post flash correction CRCORR = 'OMIT ' / combine observations to reject cosmic rays EXPSCORR= 'PERFORM ' / process individual observations after cr-reject SHADCORR= 'OMIT ' / apply shutter shading correction PCTECORR= 'PERFORM ' / cte correction DARKCORR= 'PERFORM ' / Subtract dark image FLATCORR= 'PERFORM ' / flat field data PHOTCORR= 'PERFORM ' / populate photometric header keywords RPTCORR = 'OMIT ' / add individual repeat observations DRIZCORR= 'PERFORM ' / drizzle processing
These are the steps in the calibration pipeline that need to be performed, are omitted, or have been completed (none have been completed in the raw file). The final associated products (j9cv58020_drz.fits and j9cv58020_drc.fits) have the following for the same series of keywords:
DQICORR = 'COMPLETE' / data quality initialization ATODCORR= 'OMIT ' / correct for A to D conversion errors BLEVCORR= 'COMPLETE' / subtract bias level computed from overscan img BIASCORR= 'COMPLETE' / Subtract bias image FLSHCORR= 'OMIT ' / post flash correction CRCORR = 'OMIT ' / combine observations to reject cosmic rays EXPSCORR= 'COMPLETE' / process individual observations after cr-reject SHADCORR= 'OMIT ' / apply shutter shading correction DARKCORR= 'COMPLETE' / Subtract dark image FLATCORR= 'COMPLETE' / flat field data PHOTCORR= 'COMPLETE' / populate photometric header keywords DRIZCORR= 'COMPLETE' / drizzle processing
Now the steps that had been marked as "PERFORM" are "COMPLETE." Note that the list of pipeline steps in the raw and drz/drc files are not identical because calibration steps for associations and their members may not be the same.
The full processing log for data files are found in the trl files (e.g., j9cv58020_trl.fits, j9cv58tzq_trl.fits, and j9cv58u1q_trl.fits). The logs are stored as a binary table in the first data extension of the FITS file. These log files are somewhat more difficult to read, as seen in the excerpt below:
('>>>>>>>>>>>>>>>>>>>> /ifs/archive/ops/hst/store/HSTDP-2015_3-160126//bin/exposure_times.py j9cv58020 <<<<<<<<<<<<<<<<<<<<') ('2016074184159-I-INFO-Start ------ Exposure Times Updater for j9cv58020 ------') ('2016074184159-I-INFO-exposure_times-UPDATE_EXPOSURE_TIMES is False. No update necessary.') ('2016074184159-I-INFO- End ------ Exposure Times Updater Nothing to do for j9cv58020 ------') ('FYI: exit( 0 )') ('2016074183916-I-INFO DP_open_newobs: -------------- Data Partitioning Started: j9cv58tzq ------------ (46927151842912)') ('2016074183916-I-INFO DP_open_newobs: Partitioning from POD file: lz_bdc5_067_0000102262_j9cv58tzq (46927151842912)') ('2016074183916-I-INFO Search osf is ????????- p???????????????????????.j9cv58tzq_______________________________-acs-???-????') ('(46927151842912)') ('2016074183916-I-INFO Search osf is ????????- ????????????????????????.j9cv58tzq_______________________________-acs-???-????')
They do however provide more detailed information, including timestamps in the form “YYYYDDDHHMMSS” where YYYY is the year, DDD is the day of year, HH is the hour, MM is the minute, and SS is the second.
The version of the pipeline software used to process the data is given in keywords in the primary header of the FITS files. The section of keywords with the header "/ DIAGNOSTIC KEYWORDS" contains the following keywords and values:
OPUS_VER= 'COMMON 2017_2 ' / data processing software system version CAL_VER = '3.4.1 (20-April-2017)' / CALSTIS code version PROCTIME= 5.793008144676E+04 / Pipeline processing time (MJD) CSYS_VER= 'hstdp-2017.2' / Calibration software system version id
"OPUS_VER" and "CSYS_VER" give the current Data Management System (DMS) build version, and "CAL_VER" gives the version of CALSTIS used to process this dataset (STIS association od0m070b0). The keyword “PROCTIME” gives the modified Julian date of the last processing (26 June 2017, 01:57:17 UT for this dataset). Information regarding the current DMS build can be found at https://archive.stsci.edu/hst/processing_status/. Note that these keywords are standard for the current DMS build, but may be different for older archival data.
Are you interested in more details or do you have additional questions? Please send an email to the MAST Helpdesk at email@example.com.