Determining the Accuracy and Precision of Astrometric Positions and Covariances from the LSST Science Pipelines and the Vera C. Rubin Observatory Using LSST’s Data Preview 1

Abstract

This work extends SITCOMTN-159 as an investigation into the agreement between astrometric positions and their corresponding covariances as produced by the LSST Science Pipelines. Here we use the Data Previews as the earliest pre-cursor to the main LSST survey. These will serve as the initial benchmark of the accuracy and precision of the LSST astrometry, and guide users of the data on where the data are well aligned and what corrections may need to be made.

The analysis compares the pipeline-derived precisions, as described by the covariance reported in the Data Preview, to the variation in measured position of each dataset entry. The measured position variation is computed by comparing LSST positions to two proxies for the ground truth, both high-quality astrometric datasets: Gaia and the Hubble Source Catalog. Fitting these distributions, for a set of similar detections – all with the same pipeline covariance, brightness, all in regions of the sky with the same density of background detections and same LSST depth, and so on – we can build a set of empirical uncertainties to compare to those generated by the data reduction process. In the case of both the Source and Object tables in DP1 we see a consistent under-estimation of the precisions as compared with experimental position-measurement scatter. Some of the under-estimation of the precisions can be explained by an anomalous scaling of precisions with Object table coadd depth, suggesting systematic factors affecting the reported location of detections that isn’t captured by pure photon-noise statistics.