LineStacker: A spectral line stacking tool for interferometric data

TL;DR

LineStacker is an open-source tool that efficiently stacks spectral lines in interferometric data, demonstrating high accuracy and robustness on simulated ALMA and VLA observations, with useful statistical features for analysis.

Contribution

It introduces a new spectral line stacking tool, LineStacker, with integrated statistical tools, optimized for interferometric data analysis and tested on realistic simulated datasets.

Findings

Successfully retrieves input parameters with over 90% accuracy.

Robust against complex simulated data mimicking real observations.

Identifies limitations related to redshift uncertainties affecting SNR improvement.

Abstract

LineStacker is a new open access and open source tool for stacking of spectral lines in interferometric data. LineStacker is an ensemble of CASA tasks, and can stack both 3D cubes or already extracted spectra. The algorithm is tested on increasingly complex simulated data sets, mimicking Atacama Large Millimeter/submillimeter Array and Karl G. Jansky Very Large Array observations of [CII] and CO(3-2) emission lines, from $z\sim7$ and $z\sim4$ galaxies respectively. We find that the algorithm is very robust, successfully retrieving the input parameters of the stacked lines in all cases with an accuracy $\gtrsim90$\%. However, we distinguish some specific situations showcasing the intrinsic limitations of the method. Mainly that high uncertainties on the redshifts ($\Delta z > 0.01$) can lead to poor signal to noise ratio improvement, due to lines being stacked on shifted central frequencies. Additionally we give an extensive description of the embedded statistical tools included in LineStacker: mainly bootstrapping, rebinning and subsampling. Velocity rebinning {is applied on the data before stacking and} proves necessary when studying line profiles, in order to avoid artificial spectral features in the stack. Subsampling is useful to sort the stacked sources, allowing to find a subsample maximizing the searched parameters, while bootstrapping allows to detect inhomogeneities in the stacked sample. LineStacker is a useful tool for extracting the most from spectral observations of various types.