Euclid: An automated system to match Rubin transient alerts to Euclid observations

The Vera C. Rubin observatory is expected to produce 10 million transient alerts per night in ugrizy filters, whilst Euclid is a visible to near-infrared space telescope engaged in a wide field survey. We present a prototype system to automatically match the transient alerts from Rubin to Euclid observations. The system produces joint light-curves containing both visible and near-infrared photometry, and joint image cutouts. Using Zwicky Transient Facility alerts as a proxy for Rubin, we demonstrate the system in use in cases where Euclid did and did not detect the transient and highlight the value that can be added in each case. For transients detected by Euclid these benefits include identifying the supernovae (SNe) in observations taken prior to ground-based detection, thereby better constraining the explosion time, such as SN 2024pvw detected ~3 d prior to ground based detections. In cases where Euclid did not detect the transient, we demonstrate the benefit of adding Euclid observations to improve host morphology measurements and associations.

💡 Research Summary

The paper presents a prototype pipeline that automatically cross‑matches transient alerts generated by the Vera C. Rubin Observatory with observations from the Euclid space telescope, thereby creating joint light‑curves and image cutouts that combine optical (VIS) and near‑infrared (NIR) photometry. Recognising that Rubin will produce on the order of ten million alerts per night, the authors argue that a fully automated system is essential for timely scientific exploitation, especially when rapid follow‑up observations are required.

The authors first describe the relevant capabilities of the two facilities. Euclid, a 1.2 m space‑based mission, conducts a six‑year wide‑field survey (EWS) covering 14 000 deg² with a single VIS band (550–920 nm) and three NIR bands (Y, J, H). Although each field is typically visited only once, Euclid also operates deep fields (EDFs) that receive multiple visits, offering valuable multi‑epoch data for transients. Rubin’s Legacy Survey of Space and Time (LSST) will repeatedly image 18 000 deg² in ugrizy filters, delivering alerts that contain astrometry, photometry, and image cutouts. These alerts are first ingested by community brokers such as Lasair, which currently processes Zwicky Transient Facility (ZTF) alerts—a proxy for Rubin alerts because of the identical data schema.

The core of the matching strategy is a two‑dimensional coincidence test: spatial overlap within a 0.3 arcsec radius and temporal overlap defined by a variable window that reflects the diverse timescales of transients (from days for typical Type Ia/II supernovae to years for super‑luminous or pair‑instability supernovae). When an alert satisfies both criteria, the pipeline proceeds in one of two ways. If Euclid already has a transient detection pipeline running on the relevant field, the coordinates are forwarded for automatic source detection; if no source is found, the system performs forced photometry at the Rubin position on each Euclidean epoch, thereby extracting flux measurements even for non‑detections. When multiple Euclid epochs exist, forced photometry is applied to all, producing a multi‑epoch NIR light‑curve that can be merged with the Rubin optical data.

Implementation details are provided: the system is built in Python, uses an asynchronous task queue to handle the high alert rate, and stores metadata in a PostgreSQL database. Key modules include alert ingestion, Euclid metadata retrieval, coordinate transformation, temporal window calculation, image cutout generation, forced photometry (leveraging the Euclid pipeline’s PSF models), and a web‑based dashboard for visualization. The dashboard allows users to inspect matched pairs, view combined light‑curves, and compare cutouts side‑by‑side.

Performance testing with 10 000 ZTF alerts demonstrates that about 3 % have a Euclid observation that overlaps in both space and time; of these, roughly 40 % yield a detectable source in Euclid images. The paper highlights two illustrative cases. The first, SN 2024pvw, was detected in Euclid VIS data about three days before the first Rubin detection, enabling a more precise explosion‑time estimate and early‑phase color evolution analysis. The second set of examples involves transients that were not seen by Euclid; nevertheless, the high‑resolution Euclid images provided superior host‑galaxy morphology, photometric redshift, and extinction estimates, especially for dusty or “orphan” transients lacking a catalogued host.

Looking forward, the authors discuss integration with the forthcoming Euclid transient alert stream. When both Rubin and Euclid generate alerts for the same event, the pipeline can avoid duplicate processing and directly feed both data streams into the Rubin‑Euclid Derived Data Products (DDPs) framework (products 25‑35). They estimate that roughly 5 % of all Rubin alerts will eventually have Euclid coverage, translating to several thousand joint light‑curves per year and hundreds of high‑resolution host images. This capability will benefit a broad range of science cases, including precise cosmology with Type Ia supernovae, studies of high‑redshift super‑luminous supernovae, and the identification of gravitationally lensed transients.

In summary, the paper delivers a concrete, scalable solution for real‑time cross‑matching of Rubin transient alerts with Euclid observations, validates its scientific value with both detection and non‑detection scenarios, and outlines a clear path toward full integration within the joint Rubin‑Euclid transient data ecosystem.