There are 16115 rd_typ='111' sources with a rd_typ='111' neighbor within 10", or about 1.5% of all rd_typ='111' sources. There are 4000-6000 sources in the peak near 5" (~0.5%). A selection of sources (~60) in this peak have been examined, and all were found to be true multiples with two or more obvious components in the quicklook images. This enhancement near 5" is thought to be due to sources fainter than the (R2-R1) saturation limit represented by R1 measurements due to confusion biases and subtleties of the confusion artifact identification algorithms.

Nearest neighbor proximites to ANY other source in the catalog for R1 sources with rd_flg='111' have also been histogrammed using proximites from the all-sky catalog. There are 323223 (29%) rd_typ='111' sources with ANY catalog source within 10".

For comparison

Probable Mechanisms

Some possible mechanisms were considered and found unlikely to explain the peak in the histogram at 5":

There are two situations which cause sources fainter than the (R2-R1) saturation limit to represented by R1 measurements, and are more probable causes:

Both these effects cause the number of R1 sources in close proximity to brighter R1 sources to be enhanced. The first primarily affects sources within 6" of a brighter source, and the second affects sources at somewhat larger proximites to brighter sources.

To get some additional insight into these phenomena, the following plots were made:

This plot has two interesting features. The (R2-R1) saturation level is clearly seen at J~9.4, and below that are seen a number of low snr R1 sources. A selection of these have been examined, and found to be cases where the unsaturated (R2-R1) source was marked as a confusion artifact due to proximity to a brighter R1 source. This leaves the lower snr R1 measurement to represent the source in the catalog. There are several thousand of these low snr R1s in the catalog.

The following plots show the nearest neighbor proximities within the 3 band Read 1 subset vs J magnitude:

These last two plots also show interesting features. There are a scattering of low snr R1s due to the (R2-R1) measurements being marked as confusion artifacts, and there is also a clear enhancement of R1s at low proximities which may be due to low snr R1s being promoted due to the two mechanisms discussed above. The radius of enhanced density increases with source brightness (and thus saturated radius). The aperture flux confusion bias can be seen in the paucity of fainter sources at proximites <4". A scattering of sources with very low proximites (<3") can also be seen. These small separations are probably due to partially resolved sources as shown below. The symmetry in the enhanced source density at small proximities is due to pairs of bright and faint sources in close proximity. At proximities <6", the measured fluxes tend to converge due to overlap in the apertures.

Email traffic on this topic through 9/25/02 is here.

Discussion of Biases for Blended Sources

Since there is no deblending for R1 single frame extractions, resolution is limited and photometric and position biases will be present where sources are blended in the R1 data frames. These biases may contribute to the peaks seen in the R1 proximity histograms, but it seems unlikely at this point that these effects can completely explain these peaks. The position bias does explain source pairs with very small (<3") proximites, however.

R1 sources come from single frame extractions, and are not deblended. Blended sources can only be separated if there is a local maximum pixel for each. This means the resolution can be no better than ~3+". Since active deblending was turned off for (R2-R1), the resolution is similar in both cases. For sources separated by less than ~6" in the R1 processing, the psfs will overlap, and the extracted positions will be biased such that the position estimates will be drawn together. Since aperture photometry is used, the photometry will also tend tward the sum of the two components as the separation is decreased. At the point the two sources are no longer resolved, the positions are identical, and the photometry approaches the sum of the fluxes.

An example image of a source pair with prox=4.76" with all frame extractions overlayed:

This source pair was separated in all 6 r1 frames in all 3 bands. The 4" radius aperture photometry will still be blended to some degree, however.

An example image of a source pair with prox=2.87" with all frame extractions overlayed:

This source pair was separated in 3 frames in J and H, but only 1 frame in Ks. Frames in which the pair was not separated gave positions close to that of the brighter member of the pair. The true separation of the pair is probably 3.5-4".