Goal: Make a PSF-Matched Warp/CoaddTempExp by warping first and PSF-matching second on ci_hsc data.
Why?
Warping after PSF-Matching will undo the matching if the warping is not a circularly-symmetric rotation or scaling. SDSS example: fields with Dec~60deg warped to an inappropriate CEA projection which skews images. We’d never do this, but it demos how PSF-matching first(right) fails.
Challenges:
Psf-Matching code has hard-coded assumptions about PSF dimensions and kernel sizes.
Obvious naive solution is padding the Science PSF dimensions, which hasn’t been giving good solutions.
Warping a PSF gives PSF dimensions that vary across an image.
Warping a PSF gives PSF dimensions that vary from visit to visit.
Step 1: Establish baseline performance for HSC.
Currently, ModelPSFMatch will only operate on a KernelPsf. To get the true baseline performance without editing ip_diffim requires running processCcd with PcaPsf rather than the obs_subaru default, PsfexPsf.
makeCoaddTempExp Configs:
Basic
doApplyUberCal=False
doPsfMatch=True
modelPsf.size=39
Check variety of Model PSF sigma:
modelPsf.defaultFwhm=7 to modelPsf.defaultFwhm=15
Goal: Make a PSF-Matched Warp/CoaddTempExp by warping first and PSF-matching second on ci_hsc data.
Step 1: Establish baseline performance for HSC.
# Number of rows/columns in the convolution kernel; should be odd-valued.
# Modified by kernelSizeFwhmScaling if scaleByFwhm = true
config.warpAndPsfMatch.psfMatch.kernel['AL'].kernelSize=35
# Sigma in pixels of Gaussians (FWHM = 2.35 sigma). Must in number equal alardNGauss
config.warpAndPsfMatch.psfMatch.kernel['AL'].alardSigGauss=[0.7, 1.5, 3.0]
#!/bin/bash
ks=(11 13 15 17 19 21 23 25 27 29 31 33 35)
for i in "${ks[@]}"
do
echo "config.warpAndPsfMatch.psfMatch.kernel['AL'].kernelSize=$i