In discussions with colleagues I've learned that the sensitivity of a recorder can vary over the duration of a deployment.
I'm not sure why or how this happens, but how can you evaluate if it has occurred? (without going back to the field +altering equipment design).
The only way to know if calibration has changes is to calibrate the full system before and after deployment. This can be done for a single frequency, e.g. 2 kHz and has not to be done for the whole frequency range.
Edit: this can be done even in air: have a time with same amplitude be played to system before and after deployment keeping same geometry.
Edit2: this applies also to hydrophones, even if sensitivity is not as high as in water, but for the OP (changes in sensitivity) only the difference matter.
Edit3: Obviously, using a single tone only implies that changes in sensitivity are not spectral but effect whole spectrum in the same way, Otherwise multiple tones with different frequencies are required.
Answer is for hydrophones as I think A.Simonis has more hydrophones than microphones (but probably applies to microphones as well)
The normal approach to this is to have a known precision sound source with a suitable coupler to the hydrophone (e.g. a pistonphone https://www.bksv.com/en/transducers/acoustic/calibrators/pistonphone) and then play a tone into the hydrophone before deployment and after deployment (pre- and post-calibration).
Note that to get good results the hydrophone needs to be at "deployment temperature" for both calibrations, so dip into the water for some time prior to pre-calibration (In my experience, the sensitivity drift for short deployments have always been smaller than 1 dB, so not very significant, compared to other parameters).
With heavy biofouling, post-calibration becomes tricky, as a coupler device might interfere with the bio-fouling layer and not give a correct calibration (in this case you might need a more elaborate pre- and post-calibration scheme involving calibrated transducers in a controlled water body).
Edit: this answer tackles the topic of what could be done if the sensitivity change needed to be quantified, based on reading the question state at that point!
Not an aquatic person here, but this is a gener problem for sound/vibrations in any medium I'm guessing. Also, full disclosure, I have no direct experience with doing this.
Would it be possible to have a regular playback every few days/weeks/etc. ? Here of course two issues arise: 1) is this going to hinder the primary objective of the experiment (how invasive is the playback?) and 2) how much variation do you expect in the playback itself (internal noise, weather related etc.)?
If 2) is expected to be relatively stable, then perhaps a regular ping of some sort could work - but it shouldn't conflict with 1) in the first place...
I guess that most of the variation is typically expected from the 'mechanical' parts of the sensor (membranes, cavities, etc). If the concern is that the sensitivity variation is coming at the ADC stage - a fixed electronic signal needs to be fed into the recorder part of the chain occasionally...
If sensor sensitivity is stable, you'd expect constant RMS/peak levels of the playback.
These solutions are definitely not a 'real-time' one of course, but rather post-hoc ones. The real-time solution, if one exists would be pretty neat!!!
