Anish Roshi & Frank Ghigo (Jan 18, 2002)
This note describes Radio Frequency Interference (RFI) measurements in the 510--690 MHz frequency band of the GBT to determine which frequency ranges are relatively free of RFI. The survey was done with the PF1 600 MHz receiver system of the GBT. The time variability of the RFI was also examined by repeating observations over a 9 hour period. Frequency range 654 to 690 MHz has relatively less RFI. The lower frequency range (510 -- 568 MHz) of the PF1 600 MHz band contains considerable, closely spaced (in frequency) RFI. Much of the `strong' RFI was present almost all the time during the survey. The feed arm servo does not produce much `strong' RFI in this band. Although the RFI situation at the PF1 600 MHz band is bad, some frequency ranges within this band can be used for astronomical observations. Particulary: (a) continuum observations in spectroscopic mode could use 20 to 35 % of a 40 MHz frequency band within the range; (b) 6 out of 22 galactic recombination lines are located at frequency ranges relatively free of RFI; (c) observations of any red shifted spectral line whose frequency falls in the RFI free range could be attempted with good spectral resolution. Any sensitive blind spectroscopic survey can be made carefully with good spectral resolution. Good spectral resolution is required especially because many strong interference are carriers of TV transmission and therefore can be associated with weak, narrow band RFI near these frequency, which may appear like ``RFI free'' region in the spectra given in this report.
A data set for monitoring RFI in the 510 to 690 MHz was obtained on 2002 Jan 15-16 using the PF1 600 MHz receiver of the GBT in dual polarization mode. The backend used was the spectral processor in 1024 channel, 40 MHz bandwidth mode. To scan the full frequency band of the 600 MHz system using the spectral processor, the center frequencies were set to 530, 565, 600, 635 and 670 MHz for successive data acquisitions. The reference frame was set to topocentric for these tests. Spectra from two orthogonal linear polarizations were collected using 2 FFT banks of the spectral processor in the above described mode. A data scan (ie spectrum with a center frequency) consists of four 30 secs integrations thus forming a total integration of 120 secs. The data was collected at an interval of about 1.5 hr starting at Jan 15 23:00 hrs and ending at Jan 16 8:00 hrs. The antenna was not kept at the same position during the survey but was tracking different sources in between data acquistion. The feed arm servo was ``on'' for most of the observations.
Fig. 1 shows the 120 secs averaged raw spectrum (top) from one of the linear polarizations and band-shape corrected spectrum (bottom) at the time (in UT) indicated on the title. A running five point median filtered raw spectrum was used as an estimate of the band-shape. The raw spectrum is useful to identify any broad RFI feature which will not be present in the band-shape corrected spectrum. The yellow line in the bottom figure can be used as a guide to identify the RFI components picked up by an automatic RFI detection glish routine. The glish routine computes an RMS from the band shape corrected spectrum after eliminating all `strong' RFI. All spectral values above this RMS are considered as RFI. The frequencies of the RFI components thus picked up by this routine are written into the file rfi600jan15.freq .
Fig. 2 shows 14 mts averaged band-shape corrected spectra from one linear polarization. 300 KHz (50 KHz below the rest frequency of hydrogen and 250 KHz above it) spectral windows near the rest frequencies of hydrogen recombination line transitions are marked in blue. All the data taken for the RFI survey is averaged to get these plots. Note that frequency range 654 to 690 MHz has relatively less RFI. The lower frequency range (510 -- 568 MHz) of the PF1 600 MHz band contains considerable, closely spaced (in frequency) RFI. Also 20 to 35 % of a 40 MHz frequency range, especially at the higher frequency end of the 600 MHz band, could be used for continuum observations. The frequencies and relative amplitudes of the RFI components picked up by the automatic RFI detection glish routine are written into the file avrg600jan15.freq . As in Fig.1, the yellow line can be used as a guide to identify the RFI components picked up by the glish routine.
Fig. 3 shows 120 sec integrated band-shape corrected spectra from one linear polarization when the feed arm servo system is turned ``on'' (light blue) and ``off'' (yellow). No `strong' RFI from feed arm servo is observed.
Fig. 4 shows the gray scale display of the 7 spectra on each 40 MHz band taken over 9 hrs. Much of the `strong' RFI is present through out the observations. Some variability in the strength of the RFI could be due to the change in position of the GBT when these spectra were taken.
Frequency range 654 to 690 MHz has relatively less RFI. The lower frequency range (510 -- 568 MHz) of the PF1 600 MHz band contains considerable, closely spaced (in frequency) RFI. Much of the `strong' RFI was present almost all the time during the survey. The feed arm servo does not produce much `strong' RFI in this band. Although the RFI situation at the PF1 600 MHz band is bad, some frequency ranges within this band can be used for astronomical observations. Particularly: (a) continuum observations in spectroscopic mode could use 20 to 35 % of a 40 MHz frequency band within the range; (b) 6 out of 22 galactic recombination lines are located at frequency ranges relatively free of RFI; (c) observations of any red shifted spectral line whose frequency falls in the RFI free range could be attempted with good spectral resolution. Any sensitive blind spectroscopic survey can be made carefully with good spectral resolution. Good spectral resolution is required especially because many strong interference are carriers of TV transmission and therefore can be associated with weak, narrow band RFI near these frequency, which may appear like ``RFI free'' region in the spectra given in this report.