Thanks for the paper. Well, it certainly appears that the team was able to use the JSN-SR04T to measure distance under water between 20cm and 45cm. But there is much in the paper that is not clear.
For example, note the paper claims, without substantiating evidence, that the sensor works out to 500m. Maybe they meant to say 500mm. Edited to add: But that doesn't make any sense either, because the blind distance in water should be about 20 cm x 4.3 = 86 cm = 860 mm. Maybe they meant 500 cm, the approximate max in air... who knows...
It is not clear whether it is a JSN-SR04T or JSN-SR04T-2.0.
Judging from the photos (why is there no schematic?), it looks like they're using the standard trigger and echo pins, so I suspect the use of the word "analog" is a Malaysian->English translation artifact.
They report success measuring an underwater distance of 20 cm, but that's the approximate dead zone for the sensor in air, so it is not clear whether the difference in speed of sound in air vs water (about 4 to 5x faster in water) was accounted for (why didn't they include their code?) and exactly how the tests were conducted.
The 2nd link I posted mentions that different JSN versions have different "detection angles" (why didn't they characterize their sensor); that may affect suitability for your purpose. Smaller is probably better.
Edited to add the bold above. Also, you may want to try the relatively inexpensive DS1603L if its limited range is acceptable. It's intended to be attached to the bottom of a tank, but it also works as a depth sensor according to this: How to read serial data from Non-contact Ultrasonic liquid level sensor - #22 by wvmarle <-- also has a link to a library for the device.