Optimal merging of point sources extracted from Spitzer Space Telescope data in multiple infrared passbands versus simple general source association

For collating point-source flux measurements derived from multiple infrared passbands of Spitzer-Space-Telescope data – e.g., channels 1-4 of the Infrared Array Camera (IRAC) and channels 1-3 of the Multiband Imaging Pho- tometer for Spitzer (MIPS) – it is best to use the 'bandmerge' software developed at the Spitzer Science Center rather than the relatively simple method of general source association (GSA). The former method uses both source positions and positional uncertainties to form a chi-squared statistic that can be thresholded for optimal matching, while the latter method finds nearest neighbors across bands that fall within a user-specified radius of the primary source. Our assertion is supported by our study of completeness (C) vs. reliability (R) for the two methods, which involved MIPS-24/IRAC-1 matches in the SWIRE Chandra Deep Field South. Both methods can achieve C = 98%, but with R = 92.7% for GSA vs. R = 97.4% for bandmerge. With almost a factor of three lower in unreliability (1 − R), bandmerge is the clear winner of this comparison.