Characterizing transient radio‐frequency interference

Transient radio‐frequency interference (RFI) events can be generated as unintended by‐products of the normal operation of devices such as mechanical relays or electric motors. In the radio astronomy reserve in South Africa, several new radio telescopes such as MeerKAT, the Square Kilometre Array, and the Hydrogen Epoch of Reionization Array are planned or under construction. Some will begin observations before others are complete. The associated construction equipment and infrastructure could include transient RFI sources with the potential to degrade astronomical observations. In this paper we present an analysis of RFI events in the time domain, for which we have recorded a data set of nine common sources of transient RFI. We statistically characterize these RFI signals and investigate the effectiveness of analyzing them using components analysis methods such as principal components analysis (PCA) and kernel PCA. Several interesting insights are presented on the statistical properties of this type of RFI in the time domain. The procedure we propose for analyzing transient RFI is expected to be useful for real‐time analysis and identification of RFI events. Key Points Thorough characterization of a new data set of time domain recordings of transient RFI events The RFI events are non‐Gaussian and are best characterized using nonparametric methods Preliminary classification analyses with nonlinear components analysis techniques show promise