An objective and automatic cluster finder: An improvement of the matched-filter method

We describe an objective and automated method for detecting clusters of galaxies from optical imaging data. This method is a variant of the so-called "matched-filter" technique pioneered by Postman et al. (1996). Simultaneously using positions and apparent magnitudes of galaxies, this method can, not only find cluster candidates, but also estimate their redshifts and richnesses as byproducts of detection. We examine errors in the estimation of cluster's position, redshift, and richness with a number of Monte Carlo simulations. No systematic discrepancies between the true and estimated values are seen for either redshift or richness. For clusters with z = 0.2 and with richness similar to that of the Coma cluster, typical errors in the estimation of position, redshift, and richness are evaluated as Delta theta similar to 10 " (one third of the projected core radius), Delta z similar to 0.02, and Delta N/N similar to 12%, respectively. Spurious detection rate of the method is about less than 10% of those of conventional ones which use only surface density of galaxies. A cluster survey in the North Galactic Pole region is executed to verify the performance characteristics of the method with real data. Despite poor quality of the data, two known real clusters are successfully detected. No unknown cluster with low or medium redshift (z less than or equal to 0.3) is detected. We expect these methods based on "matched-filter" technique to be essential tools for compiling large and homogeneous optically-selected cluster catalogs.