Getting Down to Business: Telling Real Pulsars from RFI

Now that you understand just what a prepfold plot tells us, you are ready to begin looking for real pulsars. As you will see, sometimes it can be easy to identify RFI, and sometimes it can be a little trickier. Of course, experience will help you to make better judgments about what is a real pulsar and what isn't, but there are also some characteristics of RFI and pulsar signals that you can use.

Using the Time Domain Plot and Pulse Profile

Figure 10 &mdash A prepfold plot of some broadband RFI that shows up at a variety of phases. This RFI is also transient. We can tell that this is probably RFI just by looking at the pulse profile, which is very broad. Most real pulsars will have a pulse profile that is fairly narrow. Another clue that this is RFI is that the DM curve peaks at zero.

It can always be useful to think back to our analogy between pulsars and lighthouses. Just as a lighthouse is only "on" for a small fraction of their rotation, so too are pulsars. This is equivalent to saying that the pulsar should only appear at one phase, and that phase should be the same throughout the observation in most cases (we will see an important exception shortly). In other words, the pulse profile should be a narrow peak, not a broad bump. Figure 10 shows an example of RFI that is spread out in phase, and consequently doesn't have a narrow pulse profile.

However, it is possible for a pulsar's signal to drift in phase. This will happen when the pulsar is highly accelerated, possibly due to a nearby companion star. We call these binary systems, and you can see this in Figure 11 (which is also an eclipsing system; see below). The time domain plot for binary pulsars will exhibit this wave-like structure. The signal still shows up as a narrow curve, and it is smooth without any sudden jumps from one phase to another. Contrast this with Figure 12, which is actually RFI. At first glance it seems to have a wave-like structure, but the signal abruptly jumps in phase. This is something a real pulsar would never do.

Figure 11 &mdash A prepfold plot of a binary pulsar. You can see the effects of the pulsars orbit as a wave-like structure in the time domain plot.

Figure 12 &mdash A prepfold plot of some RFI that drifts in phase. Notice the phase jump at 8000 seconds. There are other indications that this is RFI. The signal is narrowband, the DM curve is flat, and the Reduced $\chi^2$ vs. Period curved doesn't peak at one period. Also, notice that the period is almost exactly $3.\overline{3}$, corresponding to a spin frequency of almost exactly 300 Hz. Nature doesn't usually produce "convenient" numbers like this.

Broadband vs. Narrowband Signals

When you tune your car radio to your favorite radio station, you usually have to get the frequency right to within 0.1 MHz. For example, 103.1 MHz may play country, while 103.3 MHz plays classic rock. The fact that a given radio station only broadcasts its signal in a very narrow range of frequencies is no accident -- it is done purposely to avoid overlapping with other stations, and to avoid wasting energy by dumping it into frequencies that people don't have to listen to. These are examples of narrowband signals, and they are very common amongst RFI. Pulsars, on the other hand, don't have a way of restricting their emission to a very narrow range of observing frequency (remember, this is different than spin frequency). As a result, pulsars beam radiation at a wide range of frequencies, and this is known as broadband emission. Can you think of how this might be helpful to us when looking at a prepfold plot?

If you said that pulsars and RFI would look different in the Frequency/Sub-band plot, you'd be right. Since pulsar emission is broadband, we will see power at all the observing frequencies. We may see more power at some frequencies and less at others, but there should still be a clear increase in signal strength at all of our observing frequencies. On the other hand, narrowband emission from RFI will only show up in a very narrow range of frequencies and often only in one sub-band. Going back to Figure 12, you will see an example of narrowband RFI. If you see that the emission is narrowband, then you can be sure that the signal is in fact from RFI, and not from a real pulsar.

Transient Emission

Sometimes, the source of RFI that is masquerading as a real pulsar is only turned on for some small fraction of our observation. This may occur if the source of RFI (like a communications satellite) is only active for a brief period of time. If this is the case, then you won't see a continuous line in the Time Series. Instead, the signal will appear broken, sometimes being on and sometimes being off. We call this transient emission. If you see this, you should be cautious about labeling the detection as a real pulsar. But there are important exceptions to this rule!.

Figure 13 &mdash A prepfold plot of an eclipsing pulsar. This is actually the same pulsar as in Fig. 11, but the effects of the binary on the phase have been removed and a longer observing time has been displayed so that you can see the pulsar ``turning off'' as it is eclipsed. Notice that the pulsar seems to fade away rather than turn off abruptly. This is because of a wind from the pulsar's companion star that acts like a cloud, slowly causing the pulsar to diminish in brightness before disappearing completely.

Pulsar emission can actually appear to turn off during an observation. This can occur in binary systems. If the orientation between the Earth and the pulsar is just right, the pulsar's companion star will sometimes pass in front of the pulsar, stopping the signal from reaching us. This is known as an eclipsing system, because we say that the pulsar is being eclipsed by its companion. The same principle applies when the moon blocks the light from the Sun, resulting in a solar eclipse. Eclipsing systems will have transient emission. Sometimes, you may see the pulsar signal disappear and reappear in the same observation. Sometimes, the pulsar may only appear part of the way through the observation. Figure 13 shows an example of an eclipsing pulsar. Other pulsars, called nulliing pulsars, will actually "turn off". This phenomenon is not perfectly understood, but it too is an exception to our rule of thumb about transient emission.

So if you see transient emission, it could be a sign of RFI. But it could also be a sign of an eclipsing system or nulling pulsar. To tell them apart, you will need to examine other parts of the plot, such as the Frequency/Sub-bands and the Dispersion Measure. Experience will also help. At any rate, eclipsing systems are relatively rare, so if you do see transient emission, then you should be suspicious. More often than not, it will be RFI.

The Dispersion Measure

One of the most useful ways of identifying RFI is to look at the DM curve. Recall that any real pulsar signal must travel through space to reach us, and in doing so will certainly encounter some electrons. On the other hand, a signal originating near the Earth will not encounter any electrons and will have a DM very nearly zero. So if the DM curve peaks at zero, then you almost certainly have found RFI, not a real pulsar. It is important to note that some pulsars can have a fairly low DM (under $10~\mathrm{pc~cm^{-3}}$)³, so be sure to examine the plot closely to make sure you know where the DM curve is truly peaking.

You must also be aware that some sources of RFI can appear to have a non-zero DM (see Figures 10 and 12). So seeing a signal that has a non-zero DM does not necessarily mean that the signal is from a real pulsar. As always, you should carefully examine other parts of the prepfold plot before making a judgement.

Some Other Things to Consider

In some cases, you will see a signal that has a "convenient" spin period or frequency. By this, I mean that the period or frequency seems very close to a nice, round number. Some simple examples are a period of almost exactly 10 milliseconds (corresponding to a frequency of 100 Hz), or a frequency of 300 Hz (corresponding to a period of $3.\overline{3}$ milliseconds). It would be very strange if nature produced these types of signals, and in fact seeing these suspicious periods or frequencies is an almost sure sign of RFI. So be sure to check these numbers carefully -- it is an easy and quick way to identify potential RFI.