We present an experimental study in which we compare two different pump–probe setups to generate and detect high-frequency laser-induced ultrasound for the detection of gratings buried underneath optically opaque metal layers. One system is built around a high-fluence, low-repetition-rate femtosecond laser (1 kHz) and the other around a low-fluence, high-repetition-rate femtosecond laser (5.1 MHz). We find that the signal diffracted by the acoustic replica of the grating as a function of pump–probe time delay is very different for the two setups used. We attribute this difference to the presence of a constant background field due to optical scattering by interface roughness. In the low-fluence setup, the optical field diffracted by the acoustic replica is significantly weaker than the background optical field, with which it can destructively or constructively interfere. For the right phase difference between the optical fields, this can lead to a significant “amplification” of the weak field diffracted off the grating-shaped acoustic waves. For the high-fluence system, the situation is reversed because the field diffracted off the acoustic-wave-induced grating is significantly larger than the background optical field. Our measurements show that optical scattering by interface roughness must be taken into account to properly explain experiments on laser-induced ultrasound performed with high-repetition-rate laser systems and can be used to enhance signal strength.