106 New Emission-line Galaxies and 29 New Galactic H ii Regions are Identified with Spectra in the Unknown Data Set of LAMOST DR7

Abstract

Abstract This work is to retrieve emission-line spectra from the “Unknown” data set in LAMOST DR7 V1.2, most of which are low signal-to-noise ratios spectra. In the work, we perform emission line search and redshift calculations on the Unknown data set to get possible emission line galaxy spectra. Taking the galaxy spectra released by LAMOST as templates, the Product Quantization (PQ) based approximate nearest neighbor (ANN) search is used to retrieve the nearest neighbors of each spectrum. We keep the spectra for which the calculated redshift and the published redshift of the template meet the threshold, and 16,188 spectra with emission lines are obtained from the LAMOST DR7 Unknown data set. After visual inspection of spectra 10,266 spectra are left, in which 5828 spectra are identified as emission-line galaxies, 1782 spectra show ionization nebula features, and other 2656 are not clearly classified. Among 5828 spectra, 5720 can be found in Strasbourg astronomical Data Center catalog, Sloan Digital Sky Survey catalog, or NASA/IPAC Extragalactic Database catalog. The 108 spectra (corresponding to 106 unique coordinates of R.A. and decl.) which have no record in these three databases are new discoveries; for the 1782 spectra showing ionization nebula features, most of them have radial velocities less than 150 kilometers per second. We check them with the latest version of WISE H ii catalog V2.0 (short for HIICat V2) and 985 out of the 1782 spectra belong to 72 H ii regions. Of these H ii regions, 43 were previously identified while the other 29 are newly identified in this work including 797 newly observed spectra. Besides, there are still 2656 spectra that cannot be clearly classified although they have obvious emission lines and with small redshift. Finally, 106 new emission-line galaxies and 29 new Galactic H ii regions are identified, and we conclude that the ANN method sped up by the PQ algorithm is efficient in solving the problem of pairing spectra with massive data set to figure out their classes. We present our result at the link http://paperdata.china-vo.org/LY_paper/Work2/pressWork2_last.zip.