Thirty-three years ago, I built a prototype soil moisture sensor for Greg Arnold for what later turned out to be "Design of a dielectric based soil moisture sensor", G. J. Arnold, M. L. Stone, R. L. Elliott. I found I had to increase the distance of the sensing circuit elements from the soil, or the dielectric constant between dry soil and wet soil was too great for any sensor to function..
If I were to try to do it today with your tools, I would start with a dipole on a piece of circuit board or plastic fed by 50-ohm stripline covered in Kaplan tape. The sensor must be waterproof at all times. Then, cover both sides with thicker and thicker layers of low-loss, low-dielectric plastic until the best results are found. If the method shows any promise, a full wave loop will have an 80 to 150-ohm impedance, which might help if SWR becomes problematic.
Good Luck, Gordon