The goal of FIRST is to survey, at 1.4 GHz, the 10,000 at the north Galactic cap scheduled for inclusion in the Sloan Digital Sky Survey. The source detection threshold is 1 mJy. With a positional accuracy better than and sources per square degree confusion is not a problem. By October 1995, of the survey had been completed and a catalog containing 138,665 sources with and was released (White et al.\ 1997, hereafter WBGH). Approximately 30% of these sources are within of another source and so, for the purposes of the correlation function analysis, are considered part of double-lobed or multi-component systems.
The details of the data analysis, map production and catalog generation are given in Becker, White and Helfand (1995, hereafter BWH) and in WBGH. In brief, images are produced at each point in a grid designed to provide an efficient and uniform tiling of the sky. These images are truncated at an off-axis angle of , weighted and summed to yield a set of coadded maps containing pixels. The rms noise at each point in the coadded maps is calculated from the weighted sum of the measured noise values in each of the images that contribute to that point. Sources with flux densities greater than 5 times the rms noise and a peak flux density 1.0 mJy are included in the preliminary catalog which is then screened for spurious sources using the algorithms discussed in BWH. Of the 138,665 sources, 4,813 were flagged as possible sidelobes. Our analysis indicates that less than 10% of the flagged sources are real sources, while less than 1% of unflagged sources are spurious (see below).
BWH estimate the catalog to be 95% complete to 2 mJy and 80% complete to 1 mJy. This estimate was inferred from comparison with a deep VLA survey containing only 49 sources, but one obtains similar predictions by taking the sources detected in the Westerbork-Leiden-Berkeley Deep Survey and accounting for the the effects of the CLEAN bias, resolution effects, and the adopted peak flux density threshold. It appears that only the most extended objects are missing from the catalog. Completeness studies are discussed further in BWH and in WBGH.
A coverage map has been calculated for the survey, providing effective sensitivity estimates with resolution. A region is considered to be covered if the sensitivity at all points in the area is greater than 0.75 times that at the center of the pointing. This map has been used to establish the large scale uniformity of the survey. On scales smaller than the coadded image size (), BWH claim the variation in noise level is less than 15%, with the sensitivity pattern repeated from one field to the next.