INTRODUCTION

Passive deblending refers to the simultaneous fitting of parameters to two or more or more neighboring sources which are close enough that their point spread responses overlap, but are far enough apart to produce separate peaks on the coadded image. In PROPHOT, two sources are regarded as blended if their separation is less than or equal to a distance BLENDIST, the "standard value" of which is the distance between the 1% points of the PSF.

In the original version of PROPHOT, the grouping sequence began with the strongest source on the list, and then all NBLEND sources within a radial distance BLENDIST were reduced together. The parameters for all NBLEND sources were then regarded as final, and hence none of the sources in the blend was revisited at any later time. A serious disadvantage of that procedure was that it made no allowance for the possibility that the outer sources in the blend could themselves have close neighbors which were not included in the fit.

The passive deblending procedure in PROPHOT has now been modified so that after each fit to a blend, only the parameters of the central source are retained; the secondary sources are revisited later. This overcomes the problem discussed above.


TESTING PROCEDURE

For each set of blended sources, the total aperture flux of the blend was compared with the sum of the fitted fluxes. Three versions of the algorithm were tested:

(1) Old version; BLENDIST = 0.5 x the standard value

(2) New version; BLENDIST = 0.5 x the standard value

(3) New version; BLENDIST = 1.0 x the standard value

The maximum acceptable value of reduced chi squared was set at 3.0. Sources which failed to produce a value below this were presumed to be extended or complex, and excluded.


DATA

The data used for this test consisted of a single scan of relatively high source density (990726s, scan 35), at J-band.


RESULTS

The results are presented in Figure 1, which shows the magnitude difference (fitted - aperture) as a function of the total magnitude of the blend.

It is apparent that the new procedure is a great improvement over the old, and that there is a significant benefit to having BLENDIST at its preferred value of 1.0 rather than the "economy" value of 0.5.