WALLABY Pilot Survey: Hydra Cluster Galaxies UV and HI morphometrics

TL;DR

This study uses galaxy morphometrics and machine learning on WALLABY HI data and GALEX UV images to identify extended disks and ram-pressure stripping in Hydra cluster galaxies, demonstrating effective classification methods.

Contribution

It introduces a combined morphometric and machine learning approach to analyze HI and UV galaxy morphologies in the Hydra cluster, highlighting the effectiveness of shallow UV data and machine learning in identifying specific galaxy features.

Findings

Extended HI and UV disks can be identified via concentration metrics.

Shallow GALEX data suffices to detect XUV disks when combined with WALLABY HI data.

Machine learning classifiers achieve up to 80% precision and recall in identifying ram-pressure stripping.

Abstract

Galaxy morphology in atomic hydrogen (HI) and in the ultra-violet (UV) are closely linked. This has motivated their combined use to quantify morphology over the full H i disk for both H i and UV imaging. We apply galaxy morphometrics: Concentration, Asymmetry, Gini, M20 and Multimode-Intensity-Deviation statistics to the first moment-0 maps of the WALLABY survey of galaxies in the Hydra cluster center. Taking advantage of this new HI survey, we apply the same morphometrics over the full HI extent on archival GALEX FUV and NUV data to explore how well HI truncated, extended ultraviolet disk (XUV) and other morphological phenomena can be captured using pipeline WALLABY data products. Extended HI and UV disks can be identified relatively straightforward from their respective concentration. Combined with WALLABY HI, even the shallowest GALEX data is sufficient to identify XUV disks. Our second goal is to isolate galaxies undergoing ram-pressure stripping in the H i morphometric space. We employ four different machine learning techniques, a decision tree, a k-nearest neighbour, a support-vector machine, and a random forest. Up to 80% precision and recall are possible with the Random Forest giving the most robust results.