OH megamasers (OHMs) are rare, luminous masers found in (ultra-)luminous infrared galaxies ([U]LIRGs). The dominant OH masing line is at 1667 MHz and can spoof the 1420 MHz neutral hydrogen (HI) line in untargeted HI surveys. This ambiguity creates a potential source of “contamination” in HI surveys, particularly for next-generation surveys that will be reaching groundbreaking sensitivities and redshifts.
We present methods for distinguishing OHMs from HI sources in untargeted HI surveys without optical spectroscopic redshifts, using a k-Nearest Neighbors (kNN) algorithm, the observed line frequency, and near- to mid-IR photometry. We also present some initial results using these methods with the Arecibo Legacy Fast ALFA (ALFALFA) HI survey and preliminary Apertif HI survey data. We also discuss the benefits and limitations of using machine learning in large, untargeted surveys.
This work has been supported by the National Science Foundation through grant AST-1814648.