DUSTSWARM: A Calibrated ESA Swarm Star-Tracker Dataset for Space Situational Awareness

DUSTSWARM is a processed and annotated dataset developed to investigate the opportunistic use of spacecraft star trackers for space situational awareness. It is derived from star-tracker imagery acquired by the European Space Agency’s Swarm A, Swarm B, and Swarm C satellites and released through the Star Tracker for Space Debris initiative, STR4SD. The source collection comprises 2,504 star-tracker images acquired between 2013 and 2018. The DUSTSWARM release provides derived astrometric, photometric, annotation, and orbital-correlation products generated from these images. The dataset includes plate-solving outputs for approximately 109,510 matched star positions, World Coordinate System solutions, Gaia Data Release 3 cross-matches, astrometric residual measurements, colour-corrected photometric calibration products, and celestial sky-coverage information. All 2,504 images were visually inspected for elongated features inconsistent with point-source stellar detections. This process identified 26 resident space object candidates: six in Swarm A imagery, fourteen in Swarm B imagery, and six in Swarm C imagery. The candidate streaks were manually labelled using oriented bounding boxes that record their image-plane position, dimensions, and rotation angle. Orbital-correlation products were generated using Two-Line Element data and SGP4 orbit propagation. For each annotated candidate, the observed streak was transformed from pixel coordinates to celestial coordinates using the corresponding WCS solution. Propagated catalogue objects were filtered according to field-of-view, range, and angular-motion constraints. A likelihood-based scan over possible time offsets was then used to account for uncertainty between the image filename timestamp and the actual observation time. The archive includes diagnostic products such as observed-versus-predicted image overlays, sky-plane trajectory comparisons, angular-separation plots, right-ascension and declination time series, and time-offset likelihood curves. The dataset is intended to support research in: opportunistic space-based space situational awareness; resident space object detection in star-tracker imagery; astrometric and photometric calibration of wide-field spaceborne sensors; automated streak-detection and annotation methods; catalogue correlation and object-identification methods; and assessment of existing spacecraft instruments as supplementary SSA sensors. Dataset contents Depending on the selected Zenodo release, the archive may contain: Swarm image metadata and observation timestamps; Astrometry.net plate-solving outputs; FITS WCS files and star-correspondence files; Gaia DR3 cross-match tables; astrometric residual and image-distortion results; photometric calibration tables and fitted calibration parameters; manually annotated RSO-candidate images; oriented bounding-box annotation files; TLE subsets used for orbital correlation; SGP4 orbital-correlation results and diagnostic figures; celestial coverage and simultaneous-observation products; and Python notebooks used to generate the released products.