[KIDS] KIDSCAT completeness

[Hugo Buddelmeijer]

Hi Francesco, Nancy,

This seems a robust methodology indeed. Do you have the 98% completeness 
numbers in text as well? Then Nancy can put them in the release table.

The part of the red line that is fitted with the solid cyan line seems a 
bit curved. As in, there seems to be a bit more 'red' on the lower-right 
side of the cyan line, except at the edges. It seems that a slightly 
shallower line would fit as well, leading to slightly brighter limits. 
I'll have to read up on the algorithm to fully understand it though.

Greetings,
Hugo



On 29/11/12 12:33, Francesco La Barbera wrote:
> Dear All,
>
> I apologize for leaving yesterday before the end of teleconf .....
> here it's an update about completeness and its implementation within
> KIDSCAT.
> The completeness algorithm is based on the idea of Garilli+99, A&A, 342,
> 408,
> where one determines the magnitude at  which objects start to be lost in
> the catalog because they are
> below the brightness threshold in the detection cell. You can see an
> application of this algorithm
> to SDSS and UKIDSS data in our SPIDER-I paper, La Barbera et al.
> 2010MNRAS.408.1313L (e.g. Fig.~22).
>
> Completeness is computed using the MAG_AUTO vs. DETECTION_MAG diagram,
> as illustrated, for one field,
> in the attached jpg
> file, COMPL_Sci-DBOXHOORN-OMEGACAM-------OCAM_g_SDSS---Coadd---Sci-56240.5315550-6.jpg.
> The bz2 archive also includes the same kind of plots (automatically
> produced when running KIDSCAT) for all the 52 images.
>
> Symbols in the plot(s) are:
> blue dots -> stars, according to KIDSCAT classification
> grey dots -> objects whose MAG_AUTO and DETECTION_MAG have low S/N (<5
> and 4, respectively). These are excluded
> in the computation of the limiting mag to make the computation of
> detection limit more stable (see below).
> black dots -> all remaining sources in the catalog (considering only
> objects with S-Ex and IMA flags=0)
>
> Notice that since we do not have the detection brightness for the
> sources, I'm using the 2pixels diam. mag from KIDSCAT
> as a proxy for it (hereafter, and in the plots, I refer to this as
> DETECTION_MAG). This approximation should be OK for most purposes.
> The distiibution of points in the diagram is approximately limited, by
> construction (S-Ex detects objects if they are above a given S/N
> in the detection cell), wrt DETECTION_MAG. I determine the detection
> limit (vertical dashed line) by binning the data wrt
> MAG_AUTO, and computing in each bin the 98-th percentile of the
> DETECTION_MAG distribution. This gives us the red, solid, curve
> in the diagram. A fixed part of the red curve, *below* the detection
> limit, is fitted by a least-squares line (cyan line).
> The intersection of the cyan line with the detection limit (red, dashed
> line), gives you the completeness magnitude of the catalog
> (horizontal dashed line). By construction, the catalog should be more
> than 98% complete below this completeness value.
>
> Notice that one might use a second-order polynomial fit, rather than a
> line, to model the solid, red, curve. After several
> experiments, I decided to keep the linear fit. It gives more stable
> results (in any case, the completeness estimates do not change
> by more than 0.1-0.2mags). Notice that for u-band, because of the low
> number of galaxies with good S/N, the completeness
> estimates are somewhat uncertain, but still reasonable, in my view.
>
> Of course, using external, deeper photometric catalogs, would help us to
> obtain independent estimates of completeness,
> to compare with our internal ones.
>
> As usual, any comment/suggestion is more than welcome !!
>
> Cheers,
> Francesco
>
>
>
> _______________________________________________
> KiDS mailing list
> KiDS at astro-wise.org
> http://listman.astro-wise.org/mailman/listinfo/kids