You can do this without a submersible sensor and that makes things easier. Measure air pressure in a plastic tube. Fill the tube with air (,small pump, controlled by processor) and measure the pressure. The only thing that gets wet is the tube. Put a weight on the tube when you lower it.  

Not exactly what you are looking for but Blue Robotics Sensors might work
https://bluerobotics.com/product-category/sensors-sonars-cameras/sensors/

If you pot the sensor in epoxy the sensor and run some long i2c (with booster) wires down, it might work.

Another idea would be to put two wires down the hole. Preferably in costant distance from each other.

Then you can measure the resistance between the wires. The resistance should scale with the waterlevel. You should only turn on the current flow when you are measureing to not degenerate the wires. Preferably use AC.

If you want to make sure get wire which is covered with plastic or just insulated wire and then measure the AC resistance.

BTW: make sure to post your solution. I would like to see what you came up with.

What about you put a pipe down the bore hole from the surface to the bottom of the hole. The bottom end of the pipe is open. On the top end you place a air preasure sensor in the pipe and close the end air tide.

The water in the pipe will rise to the same hight as the water in the bore hole and will compress or stretch the air in the pipe.

From the preasure of the air you can calculate the volume by which the air is compressed and together with the mechanical dimentions of the pipe you can calculate the hight of the water level.

But i guess you would have to calibrate the system from time to time. On the other hand no electronics below water and everthing is easy accessable. 

The way this is done is that you have a pump/compressor that trickles air down the pipe, so that it slowly bubbles out the bottom. The pressure in the air, is the same as the water pressure at the bottom. This is the best method for coastal sea water measurement, and in quite a few other scenarios. The air bubbling at the end of the pipe also reduces bio-fouling I think.. 

Robust, and cheap to replace when it all goes wrong.. I use 6mm irrigation dropper tube.

This submersible sensor has a 30m range for $109.25 US plus shipping. If pump is at 35 m down and normal is 20 m down then 15 is the normal depth above the pump. This would read up to twice that or  up to 5 m below the surface. You can also get a 4-20 mA to I2C interface board.

https://www.ebay.com/itm/304189719946

Another possibility, if you have enough wire...

Seal the end of a 2 conductor wire pair, drop it down the shaft. Attach the other ends to a 555 oscillator. The wire pair works as the capacitor, while the water depth will change the capacitance. You'll just need to know the initial depth to have a reference frequency.

Seems to work well in a test I just cobbled together. A few inches of twisted pair, with the end covered with hot-melt glue, dunked into a glass of water showed a 30kHz difference. About 2KHz / cm.

Hmmm, a solar powered depth monitor built into the wellhead cap...

If you only need course measurements, you could also mount a number of them along the wall and possibly still be much cheaper. The cheap ones CAN be submerged, right? They just only give a binary output?

It's not obvious that there are cheap ones that can be submerged and give a binary output.  There are *non-contact* sensors that can be mounted on the outside of a container, but I don't have access to the outside of the container; it's buried 35m down.  The well is only four inches in diameter so I don't have a lot of room to play with, either.

Turn on the pump, measure the power needed to pump water. The more power you need, the lower the depth. (Assuming things are operating normally.)

You are not the first person to suggest this but I'm not convinced.  Doesn't it depend on the type of motor the pump has?  As far as I can tell, it's a 1.5kW pump that always draws 1.5kW - if the water level is high, you get 10kL per hour of water out, if it's low you get more like 3kL per hour.

I first heard of the technique from a hydro-something engineering student at UC-Boulder. Unfortuntely, he hadn't memorized the necessary equations. So, I started with the well pump horsepower equation. It looks like a major variable is head. There's also the pressure tank's pressure, wire losses, motor and pump efficiencies. My current project involves a well pump monitor, as soon as I have an etched board, I'll have more data. 

If that doesn't give me at least a relative depth, a laser and a bunch of floating corner reflectors.

It will depend on the efficiency of your electric motor.  However, once calculated/calibrated it will provide extremely reliable information.

That's a pretty significant difference, sounds like you could do this with just a flow meter then.  

I know adafruit has some for like ten or twenty dollars that I've used, or I'm sure theres also some cheap ones on amazon or ebay or something.  Would have to be calibrated in the field but that'd probably only take three or four data points and interpolate from there.  

For greater accuracy maybe measure pump power as well, in case the wattage does change a little with temperature or something. 

Try an Ultrasonic Sensor looking down the well at the surface of the water?

Unfortunately this isn't really workable.  The "well" is a borehole with an HDPE case installed - internal diameter is four inches.  That already contains a 32mm MDPE pipe, the pump power cable and two backup ropes for lifting the pump out.  I'm also not at all sure just how straight the borehole is.  So I think there would be too much in the way for this to work.

With none of that installed, a simple echo sounder would work; click your fingers above the top of the well and you hear a very sharp echo back a fraction of a second later.  But it doesn't really work once all the other stuff is installed.