As @MisterBill2 said, If your capacitor is made of soil, then it's going to have a lot of leakage resistance. That will totally screw up any circuit based on resistance that is designed to measure the capacitance. The only hope you could have is to make the capacitance part of a resonant circuit with an inductor and measure its resonant frequency. (But if there is too much parallel resistance, the Q might be too low for it to oscillate)

 

As @MisterBill2 said, If your capacitor is made of soil, then it's going to have a lot of leakage resistance. That will totally screw up any circuit based on resistance that is designed to measure the capacitance. The only hope you could have is to make the capacitance part of a resonant circuit with an inductor and measure its resonant frequency. (But if there is too much parallel resistance, the Q might be too low for it to oscillate)

All of this is piled on top of the problems involved in fabricating VHF circuits that operate consistently and reliably. The usual bread boarding techniques that work well at audio frequencies just won't cut it in this domain.

 

A few years ago I created an interesting capacitance monitoring circuit for comparing capacitors. It was a quite stable and very simple circuit. The frequency source was one of those crystal oscillator modules, which are available for a very wide range of frequencies. The IC was a quad 2-input nor gate. One section was used as an inverter, with two signals out of phase fed to the inputs of another nor gate. Each of those signals had a matching series resistor, and then one had a reference capacitor to common, while the other input had the unknown capacitance to common. The result would be narrow pulses out, depending on the difference in the delay of the two signals, with the difference in delay being proportional to the difference in capacitance in the simple RC delay circuit. The final 2 sections were used as buffers in series to drive a low-pass filter to provide an analog output.
The useful frequency was limited by the IC speed, and the oscillator module. Not a big limitation.

 

I've seen that circuit too, and it works very well for real capacitors. As you say, it needs two matched resistors.
This will be a leaky capacitor, so the resistances will no longer be matched, and that will mess up the result.

 

To counteract the capacitor leakage a high value variable resistor could be tried across the reference capacitor. Since the resistance between the capacitor segments can be measured, that value could be set to match, and that would allow the circuit to see mostly the capacitive effect.
But now I am wondering how those who claim that the salinity has a big effect measure capacitance?

Certainly the salt content would effect conductivity, BUT it seems that if one is only testing in a small area there should be a way to compensate for that, maybe even a simple way.

 

Isn’t the answer to the leakage problem to insulate the electrodes with plastic film?
It will produce a compound capacitor equivalent to three capacitors in series, two consisting of plastic films of known thickness, and one of the soil, where we are detecting the change in dielectric constant due to the presence of water (ε=80) in soil (ε=4).
The capacitances will be rather small, so higher frequencies would be better, but I’d not go above the low MHz range.
Mr. Google found me this, which is interesting:
https://soilsensor.com/articles/dielectric-permittivity/