mechie said:
Air displacement could be the best method ?
As dingo suggests this keeps the sensor dry and should be maintenance free?
If the idea is taken one step further ...
A displacement chamber is placed close to the river bed, open at the bottom to allow water in, has a flexible pipe which leads up to the surface. any pressure inside the vessel will be seen by the sensor which can now be above the water.
Things to consider here are ...
Air is compressible - as the river level increases so too will the water level inside the vessel, this will give a small range error. If the vessel has a surface area which is massive compared to the volume of the flexible pipe then this error will be minimised - or calculated and allowed for? (chemical engineering books show calculations for this sort of thing).
The fluid will always require some form of purging. It might not need it that often. An air reservoir at the top will reduce its need, but hardly eliminate it. One, thermal variations will cause the air to expand, bubble out, and contract [bleep] water up the hose. Also, some air does dissolve in water. Adding air to the reservoir isn't a problem but dissolving it away is.
A small air pump would do it. A microcontroller could do a great job by taking an initial pressure reading, then run the pump and see if the pressure is rising. If it doesn't, it's safe to assume the pipe is now empty of water and it's just bubbling air out the bottom. It could even record the history and come to its own conclusions about how often a purge is needed to minimize power consumption while staying within specified tolerances.