Deep Learning–based Pseudocontinuum Fitting of C <scp>iv</scp> Emission-line Regions for High-quality Absorption-line Measurement

Abstract Intrinsic associated absorption lines (AALs) in quasar spectra play a crucial role in providing insights into the chemical and physical properties of quasar outflows, host galaxies, and their surrounding interstellar medium. Affected by the complexity of the overlap of absorption lines with emission profiles, there are still many challenges to effectively measure the AALs of massive quasar data, which significantly limits the benefits of spectroscopic surveys. In this study, we first constructed a reliable sample of 3000 spectra containing only narrow absorption lines, 3000 spectra with broad absorption lines, and 3000 spectra with no detected absorption lines, based on the spectroscopic data from the Sloan Digital Sky Survey. We then implemented a deep learning model capable of automatically extracting pseudocontinuum profiles of C iv emission lines. The model proposed was validated by comparing its generated pseudocontinuum with those from the methods presented by W.-R. Huang et al., A. M. Quider et al., and G. Zhu &amp; B. Ménard, as well as by measuring and comparing the C iv AAL parameters based on different approaches, thereby demonstrating the model’s reliability. This dramatically simplifies massive data processing and improves the convenience of scientific studies related to quasars.