With these radar parameters defined and fine-tuned by using information from all the scans, a simple time-blanking method was used to extract the line profile. First, for each scan, the time of arrival of the first pulse was calculated as follows: we found an 8-pulse sequence in each scan near the time when the radar was pointing near the GBT (since each scan was 180 sec., there were 15 sweeps of the radar in each scan). Once an 8-pulse sequence was identified for each scan, using the nominal pulse frequency we could calculate the expected arrival times of all the pulses in a scan.
A single pulse width is 2 sec, but due to multiple reflections from nearby sources, pulses can arrive as late as 150 sec after the expected arrival time. Thus, we skipped 270 sec of data around a pulse: 70 sec of data were skipped before the expected arrival time, and 200 sec of data was skipped after the expected arrival time. Since we skipped 270 sec of data for approximately every 2929 sec of data, we lose 9.2% of the data.
There are very few transient pulses which have delays longer than 300 sec, which are most probably due to aircrafts near GBT. But since such reflections are very rare, we ignored those in our analysis.
We integrated the ON and OFF scans by computing 1024-point Fast Fourier Transform (FFT) of the blanked real data samples, which produced 512-point power spectra. The two ON and two OFF channels were added together to yield the final ON and OFF spectra. These are plotted in figure 3(a). The sharp peak at 3.2 MHz is due to internal RFI in the receiver which was a harmonic of 5 MHz: 1260.69 - (3.2 - 2.5) = 1289.99 MHz. For comparison, the unblanked integrations are shown in figure 3(b). As can be seen from these two figures, time blanking does result in a significant amount of RFI excision.
[After blanking] [Before blanking]