Identifying potentially missed extended sources in the Fermi-LAT 4FGL Catalog using clustering analysis

Abstract

Context. Since its launch in 2008, the Fermi Large Area Telescope (LAT) has detected thousands of sources, many of which remain unassociated. Some may be extended sources represented in the catalog by multiple point-like entries. The reinterpretation of HESS J1813-178 as a single extended source motivates a systematic search for further missed extended sources. Aims. We search for clusters of unassociated Fermi-LAT sources and test whether single extended-source models describe them better than multiple catalog sources. Methods. We apply the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm to 4FGL sources with a linking scale of $ε= 0.3^\circ$ over the 5 GeV to 1 TeV energy range. Each cluster contains at least one unassociated source and up to one extended or point-like associated source from selected categories, including pulsars, pulsar wind nebulae, and supernova remnants. Using Fermipy, we compare extended- and multiple-source models and characterize each candidate spectrally and morphologically, focusing on the Galactic plane. Results. We identify 48 clusters containing 124 sources, each with at least one unassociated source. For all eight clusters passing our quality selection, an extended-source model is statistically preferred. At linking scales of $0.4^\circ$ and $0.5^\circ$, all eight retain their core sources. Cross-matches with the Second Fermi Galactic Extended Sources Catalog (2FGES) and the HESS Galactic Plane Survey (HGPS) show that five overlap known extended sources. The other three, including one with morphology dependent on the interstellar emission model, have no counterpart in either catalog and are new extended-source candidates. Conclusions. Spatial clustering combined with likelihood-based model comparison can uncover extended sources missed in the Fermi-LAT catalog and complements existing searches.