I was tinkering about with my reflection tests, and I noticed something strange. I’ve been convinced lately that what I’m primarily seeing is the autocorrelation function happening in the data. The initial chirp is producing some of the weird effects I see but what threw me for a loop is that as I decreased power the noise started to drop in altitude.
Now, if I turned down the power then it would seem to me that the noise should happen all over not just at a decreasing distance. If the initial chirp is noisy then I would think that so should the entire correlation not just the tail end.
It is with this finding that I feel renewed in interest. I also switched to a hyperbolic chirp, using scipy.chirp, and it too must have an impact on the performance.
According to Wikipedia, https://en.wikipedia.org/wiki/Chirp, the hyperbolic chirp shows maximum response after being distorted by the doppler effect. This would mean that despite the effects of motion on the air or reflecting medium that I would still have the highest response from the correlation out of all the chirp types. I’m not sure how much doppler plays a role considering by my calculations I shouldn’t be able to detect it.
From my calculations, I expect wind and movement to be under 1000-mph and that won’t produce even 1-hertz of doppler shift.
I was wrong. I forgot doppler shift was a function of the frequency in the sense that a higher frequency undergoes more of a doppler shift. With this in mind, and above 5-ghz, a 100-mph movement would cause about 748-hertz of shift which would be detectable. I used https://www.omnicalculator.com/physics/doppler-effect to calculate the shift.
Test 1
--freq 5060e6 --samps-mul 1 --factor 2 --samps-rep 1 --tx-gain 30 --interval 50*2.5
In the doppler shift, blue means a negative frequency shift and red mean a positive frequency shift. It is difficult to see but it starts out for a few samples with no shift then it gradually increases in a positive frequency shift and then it gets noisy after about 147km.
The reflectivity doesn’t show much but there also isn’t a correction made for power level and distance. Perhaps, the information is hidden in the data and I’ve yet to coax it out?
The phase delta/difference shows a very smooth and steady rolling of the phase up to about 100km where it starts to increase and finally it abruptly seems to stop around 150km and then turn into noise afterwards.
Here is the exact phase at any given time. You can see the offset is different per scan, but each is rolling out of the correlation at the same rate and finally hits that abrupt stop around 150km and then it gets noisy.
I was doing some reading about the different layers. Now, most of this is about high frequency which is much lower than the roughly 5ghz I am operating around. The problem is there isn’t much data about what energy at the frequency I am working on will do at these different layers. I did find it interesting that the E layer just so happens to be around 150km.
| Frequency | Altitude |
| 400mhz | 109km |
| 800mhz | 101km |
| 2000mhz | 125km |
| 3000mhz | 131km |
| 4000mhz | 107km |
| 4500mhz | 130km |
| 4800mhz | 120km |
| 5000mhz | 150km |
| 5500mhz | 127km |
What is also interesting is the altitude of this barrier seems to be frequency dependent. If I hold all other parameters and only change the frequency it affects the height/time of this boundary. The boundary height is not smooth though as it seems to go up and down as one progresses higher in frequency.
Now, I am getting 130km instead of 150km at 5000mhz. It is difficult to tell if the altitude is a manifestation of something local or not. In other instances, I got the exact same altitude for all frequencies leaving me to think it is a local manifestation of some kind.