# GaussPy+: A fully automated Gaussian decomposition package for emission line spectra

## TL;DR

GaussPy+ is an automated Gaussian decomposition tool designed for large emission line datasets, improving performance in noisy conditions and incorporating spatial coherence for detailed interstellar medium analysis.

## Contribution

It introduces significant enhancements to the existing GaussPy algorithm, enabling fully automated, accurate decomposition of large, noisy spectral datasets with spatial refitting capabilities.

## Key findings

- Effective in handling complex emission spectra
- Performs well at low to moderate signal-to-noise ratios
- Validated on synthetic data and Galactic Ring Survey

## Abstract

Our understanding of the dynamics of the interstellar medium is informed by the study of the detailed velocity structure of emission line observations. One approach to study the velocity structure is to decompose the spectra into individual velocity components; this leads to a description of the dataset that is significantly reduced in complexity. However, this decomposition requires full automation lest it becomes prohibitive for large datasets, such as Galactic plane surveys. We developed GaussPy+, a fully automated Gaussian decomposition package that can be applied to emission line datasets, especially large surveys of HI and isotopologues of CO. We built our package upon the existing GaussPy algorithm and significantly improved its performance for noisy data. New functionalities of GaussPy+ include: i) automated preparatory steps, such as an accurate noise estimation, which can also be used as standalone applications; ii) an improved fitting routine; iii) an automated spatial refitting routine that can add spatial coherence to the decomposition results by refitting spectra based on neighbouring fit solutions. We thoroughly tested the performance of GaussPy+ on synthetic spectra and a test field from the Galactic Ring Survey. We found that GaussPy+ can deal with cases of complex emission and even low to moderate signal-to-noise values.