Spectroscopic QUasar Extractor and redshift (z) EstimatorSQUEzE II: Universality of the results

TL;DR

This study evaluates the robustness of the SQUEzE software for quasar spectral classification and redshift estimation across various observational conditions, confirming its stability and applicability for future large-scale surveys.

Contribution

The paper demonstrates that SQUEzE maintains high performance under different noise levels, spectral resolutions, and wavelength coverages, validating its universality for upcoming surveys.

Findings

Performance remains stable with noise dispersion 4 times higher than standard.

Spectral pixel width of 25Å does not affect results; 100Å reduces performance by ~2%.

Blue spectrum up to 7000Å suffices for accurate classification.

Abstract

In this paper we study the universality of the results of SQUEzE, a software package to classify quasar spectra and estimate their redshifts. The code is presented in \cite{Perez-Rafols+2019}. We test the results against changes on signal-to-noise, spectral resolution, wavelength coverage, and quasar brightness. We find that SQUEzE levels of performance (quantified with purity and completeness) are stable to spectra that have a noise dispersion 4 times that of our standard test sample, BOSS. We also find that the performance remains unchanged if pixels of width 25\AA are considered, and decreases by $\sim2\%$ for pixels of width 100\AA. We see no effect when analyzing subsets of different quasar brightness, and we establish that the blue part (up to 7000\AA) of the spectra is sufficient for the classification. Finally, we compare our suite of tests with samples of spectra expected from WEAVE-QSO and DESI, and narrow-band imaging from J-PAS. We conclude that SQUEzE will perform similarly when confronted with the demands of these future surveys as when applied to current BOSS (and eBOSS) data. {\it Keywords: cosmology: observations - quasar: emission lines - quasar: absorption lines}