search

esheldon / y1shearcat-bugs

Submit bug reports here
0 stars 0 forks source link

griz vs irz metacal fluxes #7

Open hoyleb opened 7 years ago

hoyleb commented 7 years ago

This may be an artefact of how metcal measures fluxes [but I'm speculating] Some rows in metacal v3/v4 riz have measured fluxes, whereas in griz they have -999's

This effects ~120k objects in the WL sample.

For example, an object in question sits are row number 11882 and has COADD_OBJECTS_ID=3007838074

This causes the photo-z based on griz metacal to have NaNs in new skool BPZ, and an arbitrary value of 0.15 in original BPZ. Both BPZ versions use the HIZ templates.

In case it helps, in the Goldv103 catalogue the above object has a MAG_PSF_G value of 99.0 All other g-band magnitudes values are measured and reasonable.

esheldon commented 7 years ago

Thanks for the report @hoyleb. Can you please post a plot here of the distribution of these objects on the sky?

hoyleb commented 7 years ago

ra/dec plot here: [des u/p] http://129.187.203.185/~hoyleb/DES/ra_dec_nans.png

FYI the file of objects with NaNs in metacal griz, sits here: [des u/p] http://129.187.203.185/~hoyleb/DES/coadd_objects_id_nans_metacal.fluxes.fits.bz2

esheldon commented 7 years ago

These are cases where none of the g band images were considered "good".

What happened was after removing all images with bad pixels set, the remaining images were in the PSF blacklist.

There are more blacklisted g PSF solutions than r,i,z so I think this could make sense.