The Arecibo Legacy Fast ALFA Survey: IV. Strategies for Signal Identification and Survey Catalog Reliability

Abstract

We present a signal extraction utility written for the purposes of the Arecibo Legacy Fast ALFA survey (ALFALFA). This survey, when completed, will have covered 7000 square degrees of the high galactic latitude sky and should detect over 20,000 extragalactic objects. It is the most sensitive blind HI survey to date. The large size of the survey justifies in itself the need for an automated way of identifying signals in the data set. The matched-filtering signal extractor proposed is based on convolutions in the Fourier domain of templates of varying widths with each spectrum. The chosen templates are built from a simple combination of Hermite functions to mimic the shape of typical galactic HI profiles of varying widths. The main advantages of this matched-filtering approach are a sensitivity to the total flux of the signals (and not only to peak flux), robustness against instabilities and short computing times. The details of the algorithm are given here, as well as results of simulations that assess the reliability and completeness of the process.