Ok. After a pretty intensive day, I just couldn't stop fiddling around with the data. I started looking at our data at the closest approach. I sparsely looked at our low resolution data during the first three hours at 1 minute intervals to see if there was any difference with the ephemeris. Each spectrum is calculated in individual decoded pulses of length 1950 microseconds and 16 ms IPP. The length of the spectral window is 2 seconds. And indeed there seems to be a nearly 5 km difference, which can be used to refine the orbital elements of the asteroid. The power fluctuates a lot, which is due to speckle. The spectrum width due to speckle reduces as a function of range.
During the next few days there will be much more results to come. We picked up the asteroid on the radar at about 20:12 UT and continuously received echos until at least 4 UT.
Update: 18/2/2013: I clarified some of the radar processing parameters and added more explanation.