After coming to the conclusion that it cannot be said that the low-DM dispersed transients seen in HawkRAO filterbank archive data are of cosmic origin, a search for dispersed transients with DMs between 100 and 1000 was conducted.
After completing the search run (which took about 12 hours to process the 1350 filterbank format files), just 3 candidates were found.
NOTE: The HawkRAO fast search algorithm expects the trace of time delay versus frequency across the narrow 2.4 MHz bandwidth of the HawkRAO observations to be very close to a straight line - which it is - even at DM=1000 as shown below.
The dispersion delay vs frequency over 2.4 MHz at an observation frequency of 436 MHz is linear (follows y=mx+b) within an R-squared = 0.99999' even at DM=1000.
Figure 1.
This basic fact was forgotten when first viewing the plots of the 3 high-DM transient events.
What initially made me think that these must be of cosmic origin was the curve in the traces (as is commonly seen in FRB dynamic spectra in published papers) - as well as a pulse profile showing an exponential scattering tail.
What I missed was that while typical published FRB dynamic spectra plots show a distinct curve in the trace, these plots cover much wider frequencies (10's of MHz) where the slope df/dt changes significantly. For the narrow bandwidth of 2.4 MHz of the HawkRAO observations there should not be a discernible bend in the curve.
The three highly-'dispersed' transients are shown below showing a significant curve in the trace.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Figure 6.
Figure 7.
A general equation which gives the relative delay between two frequencies is as follows...
...where the delay is in milliseconds and frequencies f1 and f2 are in MHz and in ascending order.
The exponent of -2 applied to frequencies f1 and f2 is called the 'dispersion index' (α). This value has been determined theoretically and confirmed by observation. Any cosmically-dispersed transient should exhibit this value of dispersion index. The 'Dispersion delay vs Frequency' plot shown above (top-most image) was drawn by evaluating the above Tdispersion equation using -2 as the dispersion index.
The following plot show the results when the delays seen in the highly-'dispersed' traces above are plotted against the delays calculated using the Tdispersion equation with a dispersion index of -2. The start and end delays have been adjusted to coincide for direct comparison.
The orange trace is the calculated dispersion delay and the blue trace is the measured delays.
It is clear that the measured delays (blue trace) do not follow a dispersion index of -2.
This is strong evidence that the highly-'dispersed' transients are not of cosmic origin.
Figure 8.
The apparent presence of a scatter tail in the de-dispersed pulse profiles (appearing in both PSRPLOT and HawkRAO code plotting) at first glance seems to be a strong indication of cosmic origin.
What terrestrial RFI source could possibly mimic scattering ?
The answer lies in the fact that the dispersion delay traces of the three transients have a different dispersion index to the expected value of -2. If an assumption is made that the dispersion index is -2, it can be seen that a fit to that index (which is almost linear over the 2.4 MHz bandwidth) in the de-dispersion process, over the range of DMs which matches the dispersion delay gradient in any selected channel, will display an apparent scatter tail.
This was investigated by 'de-dispersing' one of the transients using an equation fit to the observed delay Tdelay (secs) vs channel (c) - where...
Tdelay = (4.8147e-05 * c^2) + (4.991e-3 * c) - 5.932e-6
The results are shown below.
The left-hand profile plot below (Figure 9.) is the result of de-dispersion assuming the dispersion index = -2.
The right-hand plot (Figure 10.) is the 'de-dispersion' result using the above sweep delay equation fit.
Note: ignore the 'Best DM' value in the right-hand plot - it's an arbitrary value.
As predicted, the scatter tail seen in the left-hand plot is missing from the right-hand plot - which now has a symmetrical profile - showing that the 'scatter tail' seen in the left-hand profile plot (as well as in Figures 5 and 7 above) is an artifact caused by applying a dispersion index of -2 to a transient not following that characteristic.
Figure 9. 'De-dispersed' Profile Using Dispersion Index = -2
Figure 10. 'De-dispersed' Profile Using Fitted Sweep Equation
These negative results and also the negative results from the low-DM transients seen in HawkRAO archival filterbank are discussed in Summary.