ippToPsps for SAS V9

This is heather's wiki of the changes, ingredients, and processing of sas_v9.

ipp changes

• best qf_perfect, best sg_sep reported for dvo averages
• does the right thing for nx=0 - mags are null
• nigel's "goofy mags" due to dvo having duplicates - should be fixed both in dvo and in ipptopsps
• Better modeling and subtraction of flux from very bright stars (subtract a smoothed radial profile) -- reduces number of false detections
• do not allow negative fluxes for real detected objects in linear flux fit - improves measurements of nearby sources
• use a mask to prevent detections on 2nd pass within 1" of detections from 1st pass
• Changes to psphotStack object matching- reduces spurious sources
  • Require object be detected in 2 bands before creating sources for forced photometry (matched sources)...
  • ... Except if only detection is in y band perform forced photometry in all bands.
  • Drop matched sources for which aperture flux is not finite (This usually happens when the pixels in the matched band are totally masked)
• Insure that NAN pixels in image and variance images are masked. This avoids several crashes that prevented some static sky analysis from completing for some skycells.

• Better error handling in Kron radius measurements. This reduces spurious large values.
• When psf matching fails for an input continue with the psphotStack analysis skipping the radial aperture measurements for the matched images in that band.
• Perform psphotStack even if stacks are available for less than five filters in a given skycell.
• Fix N_FRAMES measurement which was getting lost in the skycell calibration process.
• In DVO correct an error in zero point application to the object level flux measurments

dvo

The new dvo is called SAS.20120905, located /data/ipp005.0/gpc1/catdirs/SAS/20120901 (yes, different dates, that is a typo from when Heather queued) Ingredients for the dvo:

• cam stage label = 'SAS_v8', 620 entries
• skycal stage label = 'SAS_v9', this is rerunning psphotstack (ie, staticsky and skycal stage) on the SAS_v8 pixels
• dvo type is PS1_V4 with modifications (ie, the PS1_V4 database from before, SAS.20120801, is not compatible)

Commands used to 'spiff' the dvo (trunk from 2012-09-05):

 relphot -averages -reset -update -boundary-tree ~ipp/RINGS.V3.tree.fits -D CAMERA gpc1 -D CATDIR /data/ipp005.0/gpc1/catdirs/SAS/20120901
 relastro -update-objects -reset -update -D CAMERA gpc1 -D CATDIR /data/ipp005.0/gpc1/catdirs/SAS/20120901

Some more fixes were added, at which point I was instructed to update Ohana/src/relphot and to rerun the command (from 2012-09-14)

 relphot -averages -reset -update -boundary-tree ~ipp/RINGS.V3.tree.fits -D CAMERA gpc1 -D CATDIR /data/ipp005.0/gpc1/catdirs/SAS/20120901

Heather looked at the dvo: it looks reasonable.

ippToPsps

Lots of changes to ippToPsps this time around:

Object Tables

• [x]flags have been added
• extra has been added - this is now the best qf_perfect
• sgsept has been added
• raErr has been added
• decErr has been added
• x20pct/x80pct now correctly defaults to -999 for NULL
• xStackKronMag now has correct mags (not offset by 17.8)

Stack Detection Full

• kronrad has been added, with value of MOMENTS_R1*2.5
• nigel's "goofy mags" due to dvo having duplicates - should be fixed both in dvo and in ipptopsps
• infoFlags is now FLAGS2 << 32 | FLAGS to be consistent with other tables

problems

• default config had changed. It's actually not used, but it complained since some parameters were missing. Ugh.
• ipptopsps needs to know about the changes to the dvo to properly grab it. This was sorted out by checking out the trunk (ipptrunk-20120913), compiling it, and telling ipptopsps to use that psconfig instead of the default ipp tag.