To be presented at The Vulcano Workshop: Beginning the SpaceGuard Survey (1995 September 18-22, Vulcano, Italy).
The Lowell Observatory Near-Earth-Object Search (LONEOS) is a system to survey asteroids and comets that has been under development for a little more than 2 years. Hardware consists of a 58-cm, f/1.91 Schmidt telescope; a CCD camera containing two Loral 2048 X 2048 chips (eventually, two 2048 X 4096 chips); a Silicon Graphics IRIS 4D/220GTX computer containing six processors; and other computers. The instantaneous field of view will be 10.1 square degrees.
To image the sky, the telescope will scan in declination at a rate up to 6~deg/min, corresponding to a data-acquisition rate of 1~Mb/s. At this fastest scan rate, the system will have the capability of making three scans/region, each lunation, over the entire accessible dark sky to a limiting magnitude that should exceed V_lim = 19.7 (50% detection rate in three consecutive scans at S/N ratio of 3.2), though that will probably not be the strategy adopted.
Our goal is to maximize the number of NEOs discovered that exceed 500~m to 1~km diameter. Model calculations indicate that, after ten years of full operation, about 60% of near-Earth asteroids and Jupiter-family comets larger than 1~km diameter could be discovered by LONEOS. Scans will be made on fixed regions of the sky, so images of fixed celestial sources (stars, galaxies, etc.) and repeating "cosmetic defects" (diffraction spikes, bleeding from saturated stars, etc.) will always occupy known pixels. By co-adding a number of scans, we will build high-S/N-ratio fixed-source maps, which will allow us to search for moving targets only in pixels thought to contain dark sky, thereby gaining a factor of 10^3 to 10^4 in speed over what we could achieve if we tried to detect and analyze all sources. Initially, detections will be made only on the basis of single-pixel data numbers exceeding a chosen threshold.
We will describe algorithms that maximize V_lim; i.e., minimize the S/N ratio and the false-positive detection rate. Because LONEOS images are close to being undersampled-untrailed images will typically occupy about 9, and no more than 16 pixels-the detection process is a delicate one. First, by examining unsaturated star images (about 10^5/scan), we determine the point-spread function (PSF, itself a function of zenith angle and off-axis distance in R.A.) to subpixel resolution. By moving the peak of the PSF into each subpixel of the pixel containing the peak signal, and then repixelizing, we develop a family of so-called PSF masks. After a sequence of three scans, putative moving-target detections are compared to the masks, and are accepted or rejected on the basis of chi-squared tests. The best-fit mask provides position and brightness estimates (the latter appropriately calibrated).
Approximately constant with time, our detection rate should amount to about 170~NEOs/month, of which about 50/month are larger than 1~km diameter. Over a five-year interval, we expect to discover an average of 60~NEOs/month, of which 20 exceed 1~km diameter. Thus, to discriminate the larger NEOs of interest and to secure orbits accurate enough for recovery at a subsequent apparition, we will need help in follow-up work.
Research supported, in part, by NASA grant NAGW-3397.