A Cross-Correlation based Spectral Kurtosis RFI Detector

Accurate flagging of Radio Frequency Interference (RFI) is necessary to recover instrumental efficiency and avoid false astronomical detections. Spectral Kurtosis ($\widehat{SK}$) is a popular operator in RFI flagging for radio astronomy due to its detection sensitivity to non-Gaussian emissions and its competitive computational cost. Most $\widehat{SK}$ detection pipelines are applied to single antennas or autocorrelations products. This paper investigates the application of the SK to antennas cross-correlations, and demonstrates an improved detection performance compared to the auto-correlation-based approaches.