I think it is very difficult to predict bands observed due to change of salinity.

Halogenides are mostly inactive in the mid-IR, while sulfates (S=O) and carbonates (C=O) give high and distinct signals. These signals, however, may interfere with the signals coming from the parent rock minerals. A good source for getting an impression of IR spectra of minerals is http://rruff.info/

Hence, there are too many uncertainties for predicting salinity in general with mid-IR, I would not recommend to do so.

Thank you very much Thomas Fischer for your answer and for this website.