The high water has some energy E = mgh. In the basic school, we were told that hydropower plants extract the energy of "falling water". Now, that banned principle says that if your water falls upon still turbines, you will loose energy. The water will turn into a foam, vapor and heat up until it slows down to the speed of rotating turbine. At this point, half of the "falling energy" is lost. This means that speed of turbines must match the speed of the water stream.
Now, the waterfall has the highest speed at the basement and, in order to capture all the energy, you need to stop the water completely because water escaping at speed v, carries energy $E = mv^2/2$, where mass m grows linearly with time (your losses accumulate over time). These losses are proportional to water density m and speed v. Which means we loose more energy with each time unit if outbound stream speed v is higher. The only way to achieve 100% efficiency is to stop the water completely, to v=0. However, this implies no flow and no electricity produced.
All these considerations visited me when I did some switching power supplies. Switching power supply achieves 100% efficiency by operating in one of 2 modes: full power, when water input is 100% open to accelerate the turbine and itself, and zero power, when fresh water input is closed and previously accelerated water altogether with massive turbine slow down by the energy consumer (assume a spring that we charge against its force -- it will slow down the stream with turbine). In electronics, the role of moving water and rotor is played by constant inductor, whose inductance is constant, unlike the water, which would arrive all the time and accumulate at v whenever you open the switch. I therefore would stop the water completely in the lowland, before releasing it there, achieving 100% extraction. But, it seems that in real hydro plants and wind mills, turbine rotates constantly, without any switching.
So, what theoretical efficiencies can wind and hydro turbines achieve?