The Extragalactic Sky at Low Radio Frequencies: A Study of Peaked-Spectrum Sources

Abstract

We have entered a new era of radio astronomy where the low radio frequency sky is accessible to astronomers at a level of detail never before attainable. In particular, radio surveys are now being completed with incredibly wide fractional bandwidths, which can be exploited to study the unique radio spectra of gigahertz-peaked spectrum sources (GPS) and compact-steep spectrum (CSS) sources. GPS and CSS sources have been hypothesised to be an early stage of radio galaxy evolution due to their small scale morphologies. However, such an interpretation is contentious as there is evidence that these sources are not young but are confined to small spatial scales due to a high density circum-nuclear medium. One of the reasons that there has not been a resolution between these two competing hypotheses is because the absorption mechanism responsible for the turnover in their radio spectra remains ambiguous. In this thesis we produce high quality, low frequency spectra of GPS and CSS sources to test the dominant absorption mechanism in the population. We first study PKS B0008-421, which has the steepest known spectral slope below the turnover and the smallest known spectral width of any GPS source. We find that free-free absorption is responsible for the turnover in the spectrum. This thesis then describes the empirical modelling of the primary beam of the Murchison Widefield Array (MWA) and the production of the GaLactic and Extragalactic All-sky MWA (GLEAM) catalogue. The GLEAM survey represents the widest fractional bandwidth radio survey to date, with the derived extragalactic catalogue constituting an unprecedented database of high quality low frequency spectra. We then use the GLEAM catalogue to double the number of GPS and CSS sources known. We discuss the inconsistency of the standard radio galaxy evolutionary model with the presence of sources with high power and low intrinsic peak frequencies, suggesting that current evolutionary models are deficient.