If you aren't doing it for a vibration analysis but a clutch load anlysis, wouldn't you be better off treating it as a damping factor, since it's velocity dependent?
In either case, it's probably going to be easier to measure than it is to calculate. The energy absorbed by the water when the propeller is accelerated will vary not only due to the propeller characteristics (pitch, diameter), but the rate of acceleration, inflow velocity, and depth of propeller.
If you are fortunate enough to have propeller performance characteristic (torque coefficient(KQ)/thrust coefficient(KT)/advance coefficient(J)) curves, you could get a good feel for non-transient loads and effective power(thrust) for a given input power. An input/output energy balance would tell you how much energy is being absorbed in the water without producing work and might give you a feel for the energy losses you're dealing with. However, I don't know of any body of data explicitly valid for transient loads such as a clutch engage cycle.