Submersible devices are hard to build , the transducer itself has to get the sound into the water and thus requires the right sort of interface - a waterproofed cheapo sensor for example just won’t work
I have one of the “fish finders” linked in the OP (US$25 on eBay) and it works quite well. The sonar driver provides 200VAC at 200 kHz, so the batteries don’t last very long.
It is almost impossible for a hobbyist to make one that works as well, as (a) the design information is simply not available, (b) many of the parts are specialized and (c) the parts that are available, like the transducer, cost more than the finished product.
Acoustic properties of materials aren't that impossible to find, since you only need the
bulk/shear/elastic modului and density to determine the speed of sound and acoustic impedance.
All the maths for RF signals propagating down transmission lines directly translates to
acoustic or optical signal flow, this is basic physics of waves, not something obscure.
So a lot of the design information is available - something that might be hard to find
is the loss tangent v. frequency for softer materials like polymers, and its dodgy to
extrapolate from audible frequencies to ultrasound for these.
The basic bottom line is try to match the acoustic impedances of all the materials to
reduce reflection, and just like an anti-reflection coating for optics if you can use 1/4 wavelength
interface shims of intermediate impedance to join two unmatched materials.