Verification Datasets Meeting:
Attendees: David, Hsin-Fang, Colin, Angelo
Regrets: Simon, MWV, Frossie
Colin:
- tried out intra-chip cross-talk correction for DECam, handed off to Simon
- want to add a xtalk type and be able to put them into the calibration registry
can use them for certain range of dates - need more code to get all of this to work for DECam
- linearity correction, Russell working on this, using lookup table since it should be faster than
using a polynomial (and don’t have to deal with different functional forms), will live in ip_isr - brief discussion of doing lightcurves with difference images
Hsin-Fang:
- ran multi-band driver on DECam test data
- jointCal working on DECam, John P. working on it, fixing some build issues
Angelo:
- working on tech document on SQUASH QA database schema, http://sqr-008.lsst.io/en/latest/
- looking at criteria for selecting data for image depth criteria
David:
- https://confluence.lsstcorp.org/display/SQRE/COSMOS+DECam+data+reduction
- made summary plots of the COSMOS rerun with SDSS as reference catalog
- lots of failures, still trying to figure out where the majority are coming from,
need to change qa script to gather more information on failed files - photometric scatter, between 0.01 and 0.05 mag depending on the band,
okay, not great, maybe add color-terms for photometric zero point - astrometric scatter terrible, 50-100 mas at the best
- CMD looks reasonable, but some issue probably from visits that had problems
- would like to try using the new version of processCcd
- also would like to use extendedness to separate stars/galaxies
- H-F: some failures due to few reference stars
- H-F: try new processCcd, spits out more information than old one
- Colin: sometimes many reference objects are thrown out during outlier rejection
- Angelo: extendedness only in coadds
- Jeff Calin working on HSC image, trying to get good CMD for dwarf galaxy with unresolved light
Michael: (gave update later in person)
- validation working on various datasets, cfht, hsc, decam
- baseline versions of output repos exist
- validate_drp, configuration to set pass or fail criteria
- ci_hsc been running every 8 hours on Jenkins