I suppose artificially increasing drag slightly by a known amount (tow something with a strain guage in the rope?) and noting the change in velocity might allow it to be calculated. Would have to be something small so that that the velocity change wasn't too great otherwise you can't assume that engine power and prop efficiency are the same at both speeds.
Power1 = drag x velocity1
Power2 = (drag + extra drag) x velocity2
Assume constant power so Power1 = Power 2
drag x velocity1 = (drag + extra drag) x velocity2
rearrange to give
Drag = Extra Drag x velocity2 / (velocity1-velocity2)
Main problem with this approach is that you want (velocity1-velocity2) to be large to minimise error in the calculation but yet small so that engine and prop power is constant or you can't equate them.
Some practical issue I suspect.