Tile dependence of PSF in the SN Field Coadds¶
Contact: William Wester <firstname.lastname@example.org>
In addition to tile dependence of S/G separation in the SN Fields, there is evidence of tile dependence of the psf model in the SN Field Coadds as demonstrated by the
In 100 x 100 bins in RA and dec, for each of the SN fields, plot the average in each of the bands for stars (CBDT algorithm of S/G challenge) with mag_psf < 22, the following:
δ mag = mag_psf – mag_auto
where the quantities come from the SVA1_COADD_SNx database.
For brighter stars, you expect mag_psf and mag_auto to agree very well average near δ mag = 0. Indeed, most of the area in the SN fields are green/white near 0. However,
you see unusual regions that seem to match title boundaries. Although the region in g-band for the E field is particularly unusual. It's fairly important in that the mag_psf
can be wrong by magnitudes exceeding the color scale of the plot (which is 0.2). In fact, in the upper right plot, mag_psf in the unusual region sometimes differs up to
1.5 or 2 magnitudes from mag_auto. I should point out that mag_auto appears to be consistent with the magnitudes obtained in the Single Epoch SVA1_FINALCUT catalog.