Tilt current meter was chosen since it appears to simplest type of current meter from manufacturing and deployment perspective.  An example of commercially available tilt current meter can be seen here: https://lowellinstruments.com/products/tcm-1-tilt-current-meter/. Even though there is more than one manufacturer of tilt current meters they are not particularly cheap. TCM-1 is quoted at $1-$1.5k. Other types of current meters didn't seem to be any cheaper.

Other prospective type of current meter considered was a drag force sphere + load cell described in this work: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0083240. Unfortunately this approach will only allow to measure absolute current speed. Another benefit of tilt current meter is that whole meter can be fully enclosed. There is no need to make any special waterproof ports for wires, shafts and so on. Even one cap with some o-ring or gasket should be sufficient.

I plan to use PVC pipes and plumbing components to make an enclosing. PVC is relatively cheap, doesn't require any special tools to work with, strong enough to sustain planned depth and it's also not susceptible to corrosion underwater.

Now "the brain" of the meter. In the long run I would like to receive live data stream. But that is much more work. For now my goal is to be able to deploy the meter for a month (or at least two weeks). After that I will retrieve it, download the data and redeploy it again during following dive. 

Onset HOBO Waterproof Acceleration Data Logger for G Forces, Tilt, Orientation, Activity and Motion is a very attractive option for following reasons:

- Has software to process data

- Waterproof down to 100ft (no need to deal with sealing enclosure)

- Reasonable price

However, it's limited memory (can keep only 64k measurements for all 3 axes combined - about 6-7 hours or measurements at 1Hz) is a show stopper. And there is really no way to extend that memory.

After googling a little bit I've found what appears to be a perfect candidate: https://www.adafruit.com/product/3497

Pros:

- Very affordable

- Built-in 3 axis +=1g accelerometer

- On-board voltage regulator allowing voltage range 3.6 to 16V

- MicroSD card slot - should allow to have enough space for prolonged continuous logging.

- Python as programming language

Cons:

- Need to write firmware for this controller to perform data logging

- Unknown power consumption

- Need to write software to process data log (since I'm a software engineer it's not a big issue, I'm much more comfortable with it than with soldering)

I've already ordered the board so hopefully can tinker with it late next week.

So far I see following challenges ahead:

- Making a waterproof PVC enclosing (out 2 or 3'' PVC pipes) with at least one cap that can be re-opened

- Figuring out controller power needs and arranging sufficient power supply for at least 2 weeks of logging

- Calibrating built-in accelerometer

- Finding a way to deploy the meter (on a rope or some elastic hose)

- Calibrating the meter to find match between inclination angles and current speeds

- Making a base for the meter (hopefully one-two cinder blocks will do)

- Finding a way to redeploy the meter without ruining calibration

- Writing some software to process datalog

Next action items:

- Try to make a waterproof PVC enclosing

- Get some hands on with controller board.