Was the doppler burst vector created from just taking the greatest velocity of all the possible bins (y) per pulse (x), or is it derived from radar specs that are out of our purview?
Hi,
The doppler burst vector is not necessarily the greatest velocity. If you'll check it carefully you'll see that sometimes the greatest value is not where the doppler burst vector is pointing, for example, the greatest value can be the surrounding vegetation and not the target.
Tsabar
(MAFAT Challenge Team)
If so, could you elaborate as to why we see such sudden jumps with the doppler burst (not only in the vegetation)?
Can this be explained simply as part of the internal noise of the radar?
And is the doppler burst isn't just the max velocity, how was it calculated?
Thanks.
Posted by: shaulsolomon @ Aug. 25, 2020, 7:04 a.m.Hi,
When the radar misses the detection of the “object doppler” (could happen for various reasons) it associates the last detected doppler value to the new time frame and therefore when detection resumes a “jump” in the value may happen.
Tsabar
(MAFAT Challenge Team)
I apologize for coming back to the same core point, but I am still unsure how this explains the sort of "fluctuations" that we see in many of the cases.
Taken from the CodaLab front page, but just cropped: https://drive.google.com/file/d/1sSIkthnVygMfxY8saUAk4Mod98_pGpVJ/view?usp=sharing
(From the Animal - Low SNR : https://raw.githubusercontent.com/idanbarak/test/master/Spectrograms.png)
According to how I understand your last point, I still don't see the justification for the doppler burst to shift rapidly and then to shift back shortly after (we can't blame it on the radar confusing it with the shrubbery, because as mentioned, there is only vegetative noise at the top and bottom of the spectogram).
I hope my question is understandable.
Thank you for taking the time to answer.
Hi,
In addition to misses, there are other factors that cause discontinuities in the marked dopplers. The marked dopplers are not body velocity estimations but are simply local power maxima in a specific search window. At the high signal to noise ratios, these are pretty good estimations to body velocity. In low signal to noise ratios however, these might include returns from other body parts (usually legs) returns from other objects like birds passing through or branches swaying. They also include electronic noise. In these cases said dopplers must be filtered to achieve a reasonable body velocity estimation.
Tsabar
(MAFAT Challenge Team)