Identifying backsplash galaxies using machine learning

Roan Haggar; Elizaveta Sazonova; Cameron R. Morgan; Alexander Knebe; Rhys Jordan; Weiguang Cui; Frazer R. Pearce; James E. Taylor

arXiv · 2607.26136

Identifying backsplash galaxies using machine learning

Roan Haggar·Elizaveta Sazonova·Cameron R. Morgan·Alexander Knebe·Rhys Jordan·Weiguang Cui·Frazer R. Pearce·James E. Taylor

View original source ↗Read source document ↗Export metadata ↓

Abstract

The galaxy population in the outskirts of a cluster contains members that have been pre-processed in groups and filaments, as well as backsplash galaxies -- those that have recently passed through the cluster's center. However, disentangling these two pathways is challenging observationally. In this work, we present a machine-learning-powered model, trained on simulations of galaxy clusters from The Three Hundred suite of simulations, which can identify individual backsplash galaxies in astronomical observations. This model can build samples of backsplash galaxies with a purity and completeness of up to ~70%, and galaxies on their first infall with a purity and completeness of over 80%. It can be tuned to optimise either of these two metrics, and can be used with any combination of a set of observable quantities. We have also applied this model to galaxies with asymmetric HI distributions in the Virgo Cluster, and have demonstrated that these galaxies are all likely approaching the cluster for the first time. This supports the idea that cold gas is removed from these galaxies soon after entering a cluster, and demonstrates how this classifier can provide a better understanding of which properties of galaxies are caused by a previous passage through a cluster. We have made this model publicly available in the form of a web app, with a link in the Conclusions of this paper.

Explore related subjects

astro-ph.GA

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Roan Haggar, Elizaveta Sazonova, Cameron R. Morgan, Alexander Knebe, Rhys Jordan, Weiguang Cui, Frazer R. Pearce, James E. Taylor. 2026-07-28. Identifying backsplash galaxies using machine learning. https://arxiv.org/abs/2607.26136

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Gas Phase Distribution in the Neutral ISM: A Comparison between Observation and Numerical Simulation

The neutral hydrogen (Hi) 21-cm line serves as a powerful tracer of the neutral interstellar medium (ISM). Thermal stability analysis suggests that the neutral ISM is bistable in nature, consisting of the cold neutral medium (CNM) embedded within the warm neutral medium (WNM), both in approximate thermal pressure equilibrium. When turbulence is incorporated into the numerical simulations, a third thermally unstable medium (UNM) emerges between the CNM and the WNM. Although observational studies support the existence of this intermediate phase, a clear empirical correlation between the fraction of the UNM gas and the strength of the turbulence remains elusive. In this study, we investigate the various phases of neutral ISM using Hi 21-cm emission-absorption spectra from the publicly available GWA and LAB surveys. We compare our results with several numerical simulations, including TIGRESS-NCR and TIGRESS-CLASSIC, and find that our results are more closely consistent with TIGRESS-NCR than with TIGRESS-CLASSIC. From our observational modeling, we find that 19.8% of the gas reside in the CNM phase, 32.5% in the UNM phase, and 47.8% in the WNM phase, assuming phase boundaries defined by spin temperature: T_s < 250 K for the CNM, 250 K < T_s < 4000 K for the UNM, and T s > 4000 K for the WNM. We further expect that deep, sensitive absorption studies with the Square Kilometre Array (SKA) or the Next Generation Very Large Array (ngVLA), capable of robustly detecting WNM clouds in absorption will place more tighter observational constraints on the fraction of the gas in three different phases of the neutral ISM.

astro-ph.GA↗

arXiv2026

Spatially Resolved Nebular-Stellar Reddening with JWST/NIRISS

An accurate determination of the dust attenuation within galaxies is essential to derive key physical properties such as the star formation rate (SFR). We present an analysis using the JWST/NIRISS data from the GLASS-JWST ERS programme to investigate and characterise the stellar and nebular reddening of galaxies at $1.0<z<2.4$, down to the sub-kpc scale. We use a multiregion fitting method to extract high-quality H$α$ and H$β$ emission line maps for 99 individual galaxies across a stellar mass range $7.0<\log_{10}(M_*/\mathrm{M}_{\odot})<10.5$. We find no evidence for ratios of the Balmer decrement (H$α$/H$β$) below the intrinsic limit for Case B recombination, beyond the expected variation from observational uncertainties. We reproduce the local correlation between the Balmer decrement and total stellar mass, and find no measurable difference when splitting the sample by redshift, with negligible attenuation below $\log_{10}(M_*/\mathrm{M}_{\odot})\lesssim8.5$. Similarly, the best-fit relation between the nebular and continuum reddening follows the same relation as in local starburst galaxies, $E(B-V)_{\mathrm{SED}} = (0.46\pm0.02)E(B-V)_{\mathrm{neb}}$, together indicating no significant evolution in the dust geometry within galaxies out to $z\lesssim2.4$. We derive best-fit linear relations between the differential nebular-stellar reddening and the SED-derived star formation rate (SFR) and stellar mass, finding statistically significant relations for both quantities. We use our spatially-resolved measurements to derive an empirical calibration between the resolved differential reddening, and the SFR surface density. These will enable crucial dust attenuation corrections for spatially-resolved science at higher redshifts where the Balmer lines are inaccessible, such as with future Roman grism observations.

astro-ph.GA↗

arXiv2026

A First Measurement of Circumgalactic Dust Reddening from Only 4.6 deg$^2$ of the Rubin Observatory's Data Preview 1

We present the first measurement of circumgalactic dust reddening from the Vera C. Rubin Observatory, using only 4.6 deg$^2$ of ComCam Data Preview 1 --- roughly $0.03\%$ of the final LSST footprint. Using photometric redshifts, we stack background-galaxy colors around foreground-galaxy positions and detect a chromatic reddening profile from $r_\perp \simeq 10$ kpc to $1$ Mpc. Interpreting average $E(g-z)$ with a Milky Way extinction curve, we find $A_V = (1.3 \pm 0.4) \times 10^{-1} (r_\perp / 20 \mathrm{kpc})^{-1.8 \pm 0.4}$ within $120$ kpc. The amplitude and radial dependence agree with earlier Sloan Digital Sky Survey (SDSS), KiDS, and Dark Energy Survey (DES) results despite the $\sim1000\times$ smaller survey area and a foreground sample extending 3--6 mag fainter and 1--2 dex lower in stellar mass. The innermost 10--15 kpc bin reaches $A_V \simeq 0.3$ mag, comparable to high-latitude extinction through the Milky Way disk near the solar circle; the steep power-law slope implies a dust distribution that does not simply trace the halo-gas profile. Splitting by rest-frame $g-r$ shows stronger extinction around red foreground galaxies (rest-frame $g-r > 0.5$), although the blue subsample is too noisy to establish a significant color dependence. This red sample, with median halo mass $5 \times 10^{11} M_\odot$, shows substantially more reddening within 50 kpc than previously measured around more massive LRGs and implies a dust-to-stellar-mass ratio of $\sim 2\%$, nearly saturating the dust budget allowed by stellar metal yields. These pathfinder data demonstrate LSST's promise for high-precision galaxy--dust measurements across galaxy mass, environment, and redshift.

astro-ph.GA↗