Post-correlation filtering techniques for off-axis source and RFI removal

TL;DR

This paper presents new filtering techniques for radio interferometry data that effectively reduce RFI and off-axis source contamination, improving data quality with faster processing compared to traditional methods.

Contribution

Introduction of novel fringe frequency filters, including a projected fringe low-pass filter, for efficient off-axis source and RFI removal in radio interferometric data.

Findings

Low-pass filter reduces noise and sidelobes effectively.

Filters are several times faster than existing source separation methods.

Initial tests show promising results, but further validation is needed.

Abstract

Techniques to improve the data quality of interferometric radio observations are considered. Fundaments of fringe frequencies in the uv-plane are discussed and filters are used to attenuate radio-frequency interference (RFI) and off-axis sources. Several new applications of filters are introduced and tested. A low-pass filter in time and frequency direction on single baseline data is successfully used to lower the noise in the area of interest and to remove sidelobes coming from unmodelled off-axis sources and RFI. Related side effects of data integration, averaging and gridding are analysed, and shown to be able to cause ghosts and an increase in noise, especially when using long baselines or interferometric elements that have a large field of view. A novel projected fringe low-pass filter is shown to be potentially useful for first order source separation. Initial tests show that the filters can be several factors faster compared to common source separation techniques such as peeling and a variant of peeling that is currently being tested on LOFAR observations called "demixed peeling". Further testing is required to support the performance of the filters.