We have been exploring the possibility of removing Radio Frequency Interference from astronomical using a post-correlation technique. Details of the algorithm can be found in astro-ph0006222.ps. Note that fig8 is appended here.
Some images which show the success of the operation are given below.
The astronomical data is obtained from the central beam of the multibeam feed system. The reference copy of the interference was obtained with a separate feedhorn pointing roughly in the direction of the ground-based interferer. The IF's were digitised and recorded on tape; the processing took place with a software FX correlator.
The Pks-4096 image shows, in the upper panel, the raw data (in blue), and the cleaned data in green. The third spectrum is the correction spectrum, computed according to the algorithm, which is subtracted from the raw data. The lower panel shows the raw and cleaned spectra after a bandpass correction.
Glonass Interference (Narrabri).
A variant of the previous technique can be applied to the problem of the navigation satellite signals – in this case the copy of the interference spectrum is obtained from the a priori published code for the satellite. The Raw/Cleaned image shows an observation made with the compact array of the ATNF, at a frequency of 1610 MHz. An astronomical spectral line (OH, at 1612 MHz) can be seen to the right of the glonass interference.
A summary of these techniques was presented at a recent Jodrell bank meeting.
Work in ProgressA recent 12 hour synthesis observation shows the promise of the technique. Raw data shows the typical data - very strong interference in the centre of the band. Cleaned spectrum shows the same visibility after automatic removal of the interference. The synthesised map based on the cleaned visibilities still shows traces of the interference, but the bulk has been removed, and astronomical information can be extracted; four sources with their dirty beam signature are clearly visible. The cleaned image shows the sources standing well clear of the background. The source at the lower left is PKS1259-44, PMN1302-4447, is 27 arcmin from the beam centre. The FWHP is 31 arcmin. This map was processed in the standard way - a phase calibrator was observed at regular intervals, and then used to provide the phase and gain corrections for the synthesis. The image was produced automatically - no hand editing was applied. The spectral cube is also of interest - the spectrum averaged over the central arcmin at the centre of the image shows traces of the interference; however, astronomical spectral lines would be detectable at the level of a few mJy/beam. It was not possible to reduce the un-processed data in the same way - the calibration process failed. Applying the gain/phase corrections determined from the "interference-free" calibration data produced an image which is probably the best obtainable under the circumstances - it is clearly of no use.
Some recent presentations on RFI mitigation
IVS - New Technologies in VLBI, nov. 2002
See the presentation power pnt given at the symposium, Gyeong-Ju, nov. 2002
A detailed report pdf and ps is also available
see also the presentation to the IUCAF (Bonn, march)
Last updated: 11-November-2002M. Kesteven (firstname.lastname@example.org)