ISMGCC: Finding Gas Structures in Molecular Interstellar Medium Using Gaussian Decomposition and Graph Theory

TL;DR

This paper introduces ISMGCC, a novel method combining Gaussian decomposition and graph theory to accurately identify molecular gas structures in the interstellar medium, overcoming previous over-linking issues.

Contribution

The paper presents ISMGCC, a new approach that effectively distinguishes molecular gas structures in crowded regions without over-linking, using Gaussian decomposition and graph theory.

Findings

Identified 300 molecular gas structures containing 92% of total flux.

Structures mostly show single-peaked line profiles in over 93% of pixels.

Method effectively distinguishes structures without global data clipping.

Abstract

Molecular line emissions are commonly used to trace the distribution and properties of molecular Interstellar Medium (ISM). However, the emissions are heavily blended on the Galactic disk toward the inner Galaxy because of the relatively large line widths and the velocity overlaps of spiral arms. Structure identification methods based on voxel connectivity in PPV data cubes often produce unrealistically large structures, which is the ``over-linking'' problem. Therefore, identifying molecular cloud structures in these directions is not trivial. We propose a new method based on Gaussian decomposition and graph theory to solve the over-linking problem, named ISMGCC (InterStellar Medium Gaussian Component Clustering). Using the MWISP ${}^{13}\mathrm{CO}~(1-0)$ data in the range of $13.5^{\circ} \leq l \leq 14.5^{\circ}, |b| \leq 0.5^{\circ}$, and $-100\leq V_{\mathrm{LSR}} \leq +200~\mathrm{km~s^{-1}}$, our method identified three hundred molecular gas structures with at least 16 pixels. These structures contain $92\%$ of the total flux in the raw data cube and show single-peaked line profiles on more than $93\%$ of their pixels. The ISMGCC method could distinguish gas structures in crowded regions and retain most of the flux without global data clipping or assumptions on the structure geometry, meanwhile, allowing multiple Gaussian components for complicated line profiles.