Identifying Transient Hosts in LSST's Deep Drilling Fields with Galaxy Catalogues

TL;DR

The study addresses transient host identification in LSST's Deep Drilling Fields by leveraging field-specific deep galaxy catalogues. It compares two morphology-based host-matching metrics, Directional Light Radius ($d_{ m DLR}$) and A-Value ($d_A$), using Sherlock within Lasair and augments this with an XGBoost-based confidence model to flag reliable matches; DES transients in XMM-LSS and ECDFS serve as testing ground. Results show similar overall performance for DLR and A-Value, with DLR offering greater resilience to contaminants like diffraction spikes, and the integrated ML confidence score reducing manual inspection while enabling scalable real-time processing (roughly 0.8% runtime overhead). The work demonstrates practical integration of deep-field catalogues into LSST alert brokers and argues for hybrid approaches that merge morphology, redshift, and ML confidence to optimize transient-host associations at scale.

Abstract

The upcoming Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST) will enable astronomers to discover rare and distant astrophysical transients. Host-galaxy association is crucial for selecting the most scientifically interesting transients for follow-up. LSST Deep Drilling Field observations will detect distant transients occurring in galaxies below the detection limits of most all-sky catalogues. Here we investigate the use of pre-existing smaller-scale, field-specific catalogues for host identification in the Deep Drilling Fields (DDFs) and a ranking of their usefulness. We have compiled a database of 70 deep catalogues that overlap with the Rubin DDFs and constructed thin catalogues to be homogenised and combined for transient-host matching. A systematic ranking of their utility is discussed and applied based on the inclusion of information such as spectroscopic redshifts and morphological information. Utilising this data against a Dark Energy Survey (DES) sample of supernovae with pre-identified hosts in the XMM-LSS and ECDFS fields, we evaluate different methods for transient-host association in terms of both accuracy and processing speed. We also apply light data-cleaning techniques to identify and remove contaminants within our associations, such as diffraction spikes and blended galaxies where the correct host cannot be determined with confidence. We use a lightweight machine learning approach in the form of extreme gradient boosting to generate confidence scores in our contaminant selections and associated metrics. Finally, we discuss the computational expense of implementation within the LSST transient alert brokers, which will require efficient, fast-paced processing to handle the large stream of survey data.

Figures (15)

• Figure 1: Pointings for the DDF mini-survey. Although the scale saturates at 1200 visits, the DDF pointings, clearly visible in yellow, will each receive 10,000-20,000 visits Yoachim2025.
• Figure 2: Comparison of the DLR and A-Value methods. In the A-Value method, a less detailed morphology creates 'larger' galaxies from which to perform cross-matching.
• Figure 3: Example field coverage plots for a DDF catalogue in four of the fields. Lestrade performs a conesearch in the region surrounding the central pointing for each catalogue to identify subsets of data for each field. Top left: the CFHQSIR survey Pipien2018 in XMM-LSS. Top right: Revised SWIRE Photometric Redshift Catalogue in ELAIS S1 Rowan-Robison2013. Bottom left: The COSMOS2020 Catalogue in COSMOS Weaver2022. Bottom right: The Herschel Multi-tiered Extragalactic Survey catalogue Hermes2017 in ECDFS.
• Figure 4: Example redshift coverage plots for a DDF catalogue in four of the fields. Top left: Spectroscopic redshifts in the XMM-SERVS survey Chen2018 in XMM-LSS. Top right: Spectroscopic redshifts in the Revised SWIRE Photometric Redshift Catalogue in ELAIS S1 Rowan-Robison2013. Bottom left: Photometric redshifts in the COSMOS2025 Catalogue in COSMOS shuntov2025. Bottom right: Photometric redshifts in the ACS-GC catalogue Griffith2012 in ECDFS.
• Figure 5: Primary catalogue coverage for XMM-LSS.