In my experience, for underwater RF, you use very low frequencies only, so your data rate is low. But this is not my current field of interest. I did run across this paper, which might point you in the right direction.

Thank you Albert!

Classification of channel depend on obstacles; line of sight or multi-path, so the noise is AWGN or Rayleigh for example, and it must take the effect of water on wavelength of propagation.

Maybe we have to test in the field.