Using ultrasonic sensors to measure fluid level is one of the suggested uses by the sensor manufacturers, so it’s pretty common. I’m curious to know if you’ve had any issues so far with condensation, spiders making homes, finding adequately calm water with something above it to mount on, keeping track of sensor locations, etc.
Condensation can be countered by purchasing the more expensive sensors which can self-clean, but that comes with increased energy requirements (~68 mA avg current, must be on for ~40 sec at a time). This might mean you need a larger battery, depending upon measurement frequency, or a small solar panel and a rechargeable battery (which is more expensive and would require mounting stuff on top of whatever you mount the sensor under).
Spiders seem happy to spin their webs around ultrasonic sensors; the web doesn’t generally affect the reading, but anything caught in it and suspended under the sensor will. That means a maintenance visit to correct the issue, which is sub-optimal for sensors suspended half-way across a river.
An alternative method of measuring the depth of water is a pressure transducer, but those are relatively pricey, require installation in the river beneath low tide, and have other requirements (such as filtered access to air for pressure comparisons). The ones I’ve used also state that they require re-calibration at the factory yearly, though that isn’t going to happen.
Somewhat related, the problem with using ultrasonic sensors for bin level measurement isn’t that hooking up a sensor is difficult, it’s protecting it against damage/theft. It needs to be slim enough to have the greatest range above the bin, avoid damage when emptying the bin, be difficult to remove from the bin, and be semi-resistant to intentional damage (chewing gum, someone with a lighter trying to burn it, etc.) You can encase it in metal, but that makes it larger and more difficult to get decent radio range.
I’m quite interested in finding a way to solve these issues in a very low-maintenance way, to allow for mass installation of sensors. The problems seem quite easy to solve at face value, but it doesn’t appear to be quite so easy.