Deriving Stellar Population Parameters of Globular Clusters in Late-Type Galaxies through SED Fitting in Multi-Band Surveys
Type: Poster
Session: Posters (Wednesday & Thursday)
Author: Pedro Ribeiro Floriano
Abstract: The understanding of the structure and properties of globular cluster (GC) systems allow us to unveil many aspects of its host galaxy formation and evolution. Therefore, it is of uttermost importance for one to be able to systematically identify GC candidates in and around galaxies. Moreover, the Rubin LSST big data era requires innovative and automatic pipelines to classify sources based only on their photometric information. Hence, the goal of this project is to produce catalogs of extragalactic GC candidates using an improved methodology, such that it allows for further automatization and implementation within the Rubin pipelines. Subsequently, we also aim to determine ages and metallicities of the selected sources in order to reveal the evolutionary history of the host galaxies. To achieve that, we currently have a three stages pathway to apply over test cases. First, we assemble and prepare a photometric catalog of a galaxy and its surroundings via SourceExtractor along with a custom Python semi-automatic script designed to reduce the contamination from non-GC objects within the catalogs. Second, these catalogs undergo the procedure to select potential GC candidates. The data is transposed into a latent space trained to distinguish between GC and other sources, allowing a more thorough selection of objects which occupy the same locus as confirmed GCs, hence producing a catalog of candidates. We have been assessing the problem of creating a training set composed of spectroscopically confirmed GCs data by means of a collaborative task-force in the Rubin star clusters working group. The data gathered by this task-force ought to be compiled and made available within a Globular Clusters Science Portal hosted by LIneA, facilitating further synergies with Rubin. Finally, the stellar population (SP) properties of the selected candidates are derived using Code Investigating GALaxy Emission (CIGALE) spectral energy distribution (SED) fitting. To better represent the SP of GCs in this process, we make use of a tailored combination of spectral libraries. As test cases, we focus on a sample of 10 nearby galaxies with distances up to 5 Mpc, with M81 and NGC 2403 with available multi-band data in ugriz as the main targets for our initial analyses, making use of 12 bands of photometric data as well as images. Preliminary results indicate that our catalog preparation stage is indeed able to reduce the number of contaminants, and it presents itself as indispensable when analyzing the disk region of late-type galaxies. Beyond that, in regards to selecting candidates, we explore the use of Principal Component Analysis (PCA) and Uniform Manifold Approximation and Projection (UMAP) for dimensionality reduction. We are able to show that these methods cluster confirmed GCs in different manners. UMAP reveals structures not present in PCA projections, thus producing different catalogs of candidates which still are to be compared in the space of SEDs. Moreover, concerning SED fitting, we have combined the X-Shooter Spectral Library with BC03 and now expect to implement them into CIGALE. As for perspectives, we want to further develop and analyze the intricate details of all the stages mentioned, ultimately aiming to automate the selection of extragalactic GC candidates in LSST and the subsequent analysis of their properties.