T1000 analysis for SG separation¶
The T1000 COSMOS reduction is a good candidate to be included as part of round 3 of the SG separation Challenge. The first step is to run the SLR correction to calibrate the colors of the field. To do that, a match of selected stars to 2MASS catalog and posterior execution of BigMACS (see Keith Bechtol's procedure [[descalibration:Stellar_Locus_Regression_for_SVA1_on_HEALPix_Grid|here]]) is performed.
Selected stars in DES (from Keith Bechtol):
| spread_model_[ri] | < 0.002
| spread_model_[gz] | < 0.005
magerr_psf_[ri] < 0.05
magerr_psf_[gz] < 0.1
mag_psf_i < 22
For the old COSMOS (deep) reduction, only 40 such matched stars were found, the constraint being the high number of matches to 2MASS that are saturated in these coadds.
However, the shallower coadd (T1000), produces only 5 matches!
I downloaded my own selection from 2MASS (just area selected) and did the matching. I obtain ~1200 matched objects for COSMOS and ~140 matched objects (not sure where the large differences wrt to the SLR matching come from).
The star selection described above keeps the number ratio of objects:
(old COSMOS)/T1000 =~ (stars old COSMOS)/(stars T1000) =~ 1/3
The ratio changes significantly after matching
(matched stars old COSMOS)/(matched stars T1000) =~ 1/8 (similar to what Keith finds in the SLR procedure)
Analyzing which matches are "lost" in the way, it turns out that the main culprit is the selection of stars in the DES side. The cut in MAGERR_PSF_[GRZ] (specially in Z) eliminates many of the stars at the bright end (mag_psf_i<~18) which are precisely those who match the 2MASS stars. This happens in T1000, but not so prevalently in the old COSMOS reduction. These stars have assigned MAGERR_PSF=99.
Plot of MAG_PSF_I distribution of star-like objects in T1000 before and after the MAGERR_PSF cuts. Notice that the bright magnitude tail (where 2MASS matching candidates lie), is suppressed.
Distribution of star-like objects in old COSMOS and T1000.
Distribution of star-like objects in old COSMOS and T1000 matched to 2MASS.