This is an extremely strong pulse detected at a scale of 12.8 microseconds wide. The plot shows the pulse's frequency as a function of time. (The frequency is offset so that 0 Hz on the plot actually means 1420 x 10^6 Hz.) The pulse has positive slope in this picture, but its dispersion is negative -- meaning that the pulse slopes from low frequency to high frequency. As such, it must have come from a terrestrial source (human made) such as a radar. Another sign of its strength is that it has produced a smaller positive dispersed component, by overwhelming our electronics and causing them to behave badly.
Here is the dedispersed version of the pulse. Essentially, the picture has been sheared until the strong component of the pulse lines up vertically.
This plot is obtained by summing the columns of the dedispersed plot above. You can see a big spike where the signal is located. For a smaller signal, this spike would not be so obvious.
Here's the statistical test -- the incomplete gamma function. A value of 30 or higher on the vertical axis means the pulse is significant. Only one such pulse is expected to appear randomly per workunit. A value much in excess of 30 means that the pulse is certainly not caused by chance fluctuations.