GBT 450 MHz band RFI survey (Oct.03)

The 400-450 MHz band was observed while the GBT was doing 360-degree rotations in azimuth.
Summary of Data
Date Band Project Code Backend(s) Types of Data
2003 October 19 400-450 MHz TRFI450_OCT19 ACS(50MHz) Az rotation
All ACS spectra done with 50 MHz bandwidth, 1-second noise switching, 2 second integrations, 1080 arcmin/min az rate (i.e, one 360 deg rotation in 20 minutes).

Commentary on October 19 data.

The spectrometer was used in 50 MHz 9-level mode, with high spectral resolution (32768 channels, i.e. 1.5 kHz channel spacing). Scan numbers 15 and 16 were observed at night (2:45-3:30 AM local time), and scan number 18 in the day (11:20 to 11:40 AM).

Plot 15A shows the uncalibrated data for scan 15 averaged over all directions. Major RFI spikes are seen at 409.8, 422.8, and 432-434 MHz.
Plot 16A shows the uncalibrated data for scan 16 averaged over all directions.
Plot 18A shows the uncalibrated data for scan 18 averaged over all directions. Major RFI spikes are seen at 443-446 MHz, 409 MHz, and 432 MHz.

Doing median filter subtraction on all 599 integrations of 32786-channel spectra proved to be too time-consuming. The spectra were averaged to reduce the number of channels to 4096, increasing the channel spacing to 12.2 kHz. The data were calibrated and converted to units of Janskys (10**-26 Watts/sq.meter/Hz). A median-filtered baseline was subtracted to flatten the bandpass. A 7-point median filter was used on scans 15 and 16, 5-point on 18.

Plot 18B shows the calibrated data for scan 18 and integration number 599 after baseline flattening.

RFI spikes were identified as any data point exceeding 3.0 times the rms in the baseline-flattened spectra.

Plot 18C shows the same data as in plot 18B, with the identified RFI spikes superimposed in green.

Plot 18D shows the typical run of rms with frequency, for integration number 184. The rms was calculated over a 21-channel (256 kHz) window centered on each point in the spectrum. Outliers above 2.6 times the rms were thrown out and the rms re-calculated, so that the rms was not contaminated by RFI spikes. This procedure was sometimes overwhelmed by very large and wide RFI signals. In these cases, the rfi was so large that it was identified anyway.

Note that the typical rms ranges from 0.2 to 0.6 Jy across the spectrum. This may be a combination of the noise predicted by the radiometer equation (0.12 Jy) and confusion (0.5 Jy). The drop in rms at the band edges is probably bogus. In fact, the change in rms across the band is probably bogus because we used a single value for Tcal and deltacal for the whole band.

Download ascii files

Each of these files, when "gunzip"ed and un-tarred, becomes a directory with 599(or 594) ascii files, each file contains 4 spectra taken at a particular azimuth. The file has a header indicating the azimuth and time. The data have been calibrated and are in units of Janskys, i.e., 10**-26 watts per sq. meter per Hz. The columns in the table are 1) frequency, 2) X polarization, 3) Y polarization, and columns 4) and 5) is the same data with a median subtracted.

Scan 15, 594 2-second integrations. Sweep from 0 up to 360 azimuth. Taken Oct 19, 02:46-03:06 AM (EDT) (34 MB)
Scan 16, 594 2-second integrations. Sweep from 360 to 0 azimuth. Taken Oct 19, 03:07-03:27 AM (EDT) (34 MB)
Scan 18, 599 2-second integrations. Sweep from 360 to 0 azimuth. Taken Oct 19, 11:20-11:40 AM (EDT) (34 MB)