Originally Posted by Hyperveloce
I know that this kind of multipath may happen for airborne EM radar in a look-down configuration (versus the target) over a rough relief, but would it be possible for sound waves propagating in water and sonars (more curvy wave-rays than EM waves propagating in the air) over a rough seafloor?
Certainly surface/terrain reflections are possible, in addition to the various forms of refraction, reflection and ducting due to sound speed variations in the water, which have already been discussed. As I am sure is also true for EM waves, the type of reflecting surface and the angle of incidence can have a large effect on the strength of the reflected signal and the amount absorbed or scattered. The ocean surface is a particularly good reflector, but it also scatters or focuses due to the shape of the surface waves.
The 37kHz signal is weak, attenuated by the water, and more easily absorbed and scattered by soft or rough surfaces than are lower frequency signals. I would not expect terrain reflected paths to produce strong signals, though every rule has an exception.
As for localization, it all depends on what Emeraude was using for transducers. An omni-directional transducer will not provide any localization at all, other than knowing that the source is "near" the receiver. If they used a towed line array, they could compute a bearing (with or without left-right ambiguity, depending on various techniques), but no elevation. If they used a two dimensional array, such as a bow dome sonar, they could, in principle, compute bearing and elevation. Any bearing information is likely to be more reliable than elevation because there is usually less horizontal deflection of signals than vertical (because of the prominent horizontally layered sound speed variations). BEA's second report implies that several sensors on Emeraude were used, but it does not detail any of them (not surprised, the details are likely classified). The best localization often comes from crossed bearings when the source is received at different locations. There is no way to directly compute range, because the pinger is asynchronous.
If yes, the waveform detection would be correct but the AoA would be largely biased... and I wonder if the new Thales sonar signal processing can/do make use of the real bathymetry as it was measured during the 1st phase.
As I said before, I have no idea what sensors or signal processing have been used. However, I would bet that they are mainly trying to extract any sign of the signal, and are likely not bothering, yet, with any ray bending or possible reflections to refine the localization. IF (big if) they really are convinced that they detected a signal from a pinger, then they might now be going back to the environmental data to see if they can guess the path of the signal. I'm not really convinced that it would be possible to learn much from such an analysis, depending on how well they actually know Emeraude's position and how densely they sampled the sound speed profile.