Near-Earth Object Orbit Linking with the Large Synoptic Survey Telescope

Abstract

We have conducted a detailed simulation of LSST's ability to link near-Earth and main belt asteroid detections into orbits. The key elements of the study were a high-fidelity detection model and the presence of false detections in the form of both statistical noise and difference image artifacts. We employed the Moving Object Processing System (MOPS) to generate tracklets, tracks and orbits with a realistic detection density for one month of the LSST survey. The main goals of the study were to understand whether a) the linking of Near-Earth Objects (NEOs) into orbits can succeed in a realistic survey, b) the number of false tracks and orbits will be manageable, and c) the accuracy of linked orbits would be sufficient for automated processing of discoveries and attributions. We found that the overall density of asteroids was more than 5000 per LSST field near opposition on the ecliptic, plus up to 3000 false detections per field in good seeing. We achieved 93.6% NEO linking efficiency for H<22 on tracks composed of tracklets from at least three distinct nights within a 12-day interval. The derived NEO catalog was comprised of 96% correct linkages. Less than 0.1% of orbits included false detections, and the remainder of false linkages stemmed from main belt confusion, which was an artifact of the short time span of the simulation. The MOPS linking efficiency can be improved by refined attribution of detections to known objects and by improved tuning of the internal kd-tree linking algorithms.