The Emission Line Properties of Star-forming Galaxies in the WEAVE-LOFAR Survey

Abstract

The third operational rehearsal of WEAVE-LOFAR has produced 7273 model optical spectra of radio-selected star-forming galaxies. It is the most realistic simulation of what we expect WEAVE's performance to be. The purpose of this research is to build an algorithm which can reliably extract emission line properties from the model spectra in preparation for the bulk arrival of the WEAVE-LOFAR data by the end of 2020. Spectral fitting software already exists; however, the algorithm described in this thesis has been designed to specifically model simulated WEAVE spectra. The algorithm uses Markov chain Monte Carlo (MCMC) techniques to determine posterior distributions of the emission line parameters, and is designed to accurately model low signal-to-noise (SNR) spectra of radio sources. The data preparation stages and the process of modelling the continuum, dust attenuation and the broad and narrow emission lines are discussed in detail in this thesis. The fundamental aim is to gain an insight into what we expect the real WEAVE data to resemble and understand everything we can prior to its arrival. This thesis presents evidence that reliable emission line diagnostics can be retrieved, even from low SNR spectra, by comparing their flux measurements to the input spectra. These emission line measurements are used to classify the targets into star-forming galaxies or AGN, with an accuracy of 89%. Hα fluxes are dust corrected using the Balmer decrement, from which star formation rates (SFR) are derived and compared to those derived from the SKA simulated skies LOFAR 150 MHz measurements. The purpose of this comparison is to understand how representative the radio derived measurements are of the SFR and whether a strong case can be made to use non-thermal radio continuum as an alternative SFR indicator. Radio surveys are very sensitive, impervious to dust and can survey the sky at fast speeds, therefore the use of this diagnostic will be highly beneficial to star formation research. During the process of this analysis, any complications or limitations presented by the model spectra are thoroughly investigated and reviewed. A potential issue with the flux calibration is highlighted which appears to distort the shape of the continuum, offsets the blue and red arm spectra and results in a 25% systematic offset from the input fluxes. Furthermore, the presence of skyline residuals, particularly at longer wavelengths, showed evidence of poor sky subtraction. Despite these issues, the research in this thesis provides a proof of principle that MCMC-based codes can be used in bulk to classify and recover physical information from WEAVE spectra.