The steps taken to calibrate the photometry are as follows.
- KiDS-ESO-DR2 photometric calibration starts with individual zero points per CCD based on SA field observations. The calibration deploys a fixed aperture (6.3 arcsec diameter) not corrected for flux losses, and uses SDSS DR8 PSF magnitudes of stars in the SA fields as reference. Magnitudes are expressed in AB in the instrumental photometric system (i.e., no color corrections between OmegaCAM and SDSS photometric system applied).
Next, the photometry in the g, r, and i filters is homogenized across CCDs and dithers for each filter in each tile independently. For u-band this homogenization is not applied because the relatively small source density often provides insufficient information to tie adjacent CCDs together. This global photometric solution is derived and applied in three steps:
- From the overlapping sources across dithered exposures, zero point differences between the dithers (e.g. due to varying atmospheric extinction) are derived.
- Zero point differences across CCDs are calculated using all CCD overlaps between the dithered exposures. Steps 1 and 2 both apply a minimization algorithm (see Maddox et al. 1990, MNRAS, 246, 433).
- The zero point offsets are applied to all CCDs w.r.t. an average zero point valid for the night, derived from the nightly SA field observations. If no SA field observations are available for the night, default values are used instead.
The photometric calibration provided in KiDS-ESO-DR2 is in AB magnitudes in the instrumental system. Color-terms have been calculated with respect to the SDSS photometric system.
Aperture-corrected magnitudes taken from the multi-band catalogue, were matched to SDSS (DR8) PSF magnitudes of point-like sources. For each filter, the median offset to SDSS is first subtracted, rejecting tiles where this offset exceeds 0.1 mag in any of the g, r and i bands. The fit is done on all points from the remaining tiles. The figure below shows the photometric comparison and the following equations give the resulting color terms:
uKiDS - uSDSS = (-0.051 ± 0.002)(uSDSS - gSDSS),
gKiDS - gSDSS = (-0.052 ± 0.002)(gSDSS - rSDSS),
rKiDS - rSDSS = (-0.033 ± 0.002) (gSDSS - rSDSS),
iKiDS - iSDSS = (+0.012 ± 0.002)(rSDSS - iSDSS).
The matter of photometric quality can be split up in two:
- the uniformity of the photometry within each tile
- the quality of the absolute photometric calibration per tile/filter
Unfortunately, the distribution of tiles included in KiDS-ESO-DR2 is not contiguous, with many isolated tiles. Because of this, a complete homogenization of the photometric calibration for the entire data set was not possible. This should improve greatly in future releases.
Both the internal photometric homogeneity within a coadd and the quality of the absolute photometric scale is assessed by comparing the KiDS-ESO-DR2 photometry to SDSS DR8 (Aihara et al., 2011, ApJS, 193, 29). For this, PSF magnitudes were extracted from SDSS DR8 and compared to the aperture-corrected magnitudes in the KiDS-ESO-DR2 multi-band catalogue. Only stars with photometric errors both in KiDS and in SDSS smaller than 0.02 mag in g, r, and i or 0.03 in u were used. This comparison can be done only for all tiles in the KiDS-North field, but since KiDS-South was calibrated in the same way as KiDS-North, we expect the conclusions to hold for all data.
The quality of the absolute photometric scale is illustrated in the left panels in the figure below, where the distributions of photometric offsets between KiDS and SDSS are shown for the overlapping tiles in the KiDS-North field. A systematic offset of ~0.02 mag is present in all filters. This could be due to the fact that nightly zero points are determined using a fixed aperture on stars in the SA field without aperture correction. The scatter and occasional outliers are due to non-photometric conditions (during either KiDS or SA field observations) and, particularly in case of the u-band, use of default zero points for nights without good SA field observations. The quality of this absolute photometry will improve greatly in future releases with significant contiguous area, which will allow a global calibration for the entire survey. All zero point offsets determined from this comparison with SDSS are available in the source catalog table.
The figure below shows the offsets between KiDS and SDSS DR8 magnitudes for one tile (KIDS_129.0_-0.5), which is a representative example. Generally, the photometry in a filter is uniform within one tile to within a few percent. The right column of the figure above shows the distribution of the Median Absolute Deviation (MAD) of the stellar photometry between KiDS and SDSS for the tiles in KiDS-North, demonstrating the photometric stability within survey tiles. The relatively poor photometry in u-band is due to the lack of photometric homogenization within a tile.