Bottom lock depends on a usable echo coming back from the seabed, and soft mud is the worst case: it absorbs energy instead of reflecting it, and the return that does come back is weak and variable.
What is actually happening:
- Low backscatter. Fluid mud and fine silt return very little energy at the grazing angles a DVL uses, so the correlation peak falls into the noise.
- Volume rather than surface return. In very soft beds the echo comes from a layer, not a boundary, so the measured range wanders and the velocity estimate drifts.
- Gating and altitude limits. Once the altitude falls outside the configured window, the unit stops reporting valid bottom track even though the hardware is fine.
Practical countermeasures: widen the bottom-track search window, lower the tracking threshold and verify against a known speed over ground from GNSS; reduce speed in very shallow, very soft areas; and log altitude and correlation separately so you can tell a soft-bed problem from a mounting or interference problem.
Related: DVL product overview and field reports on navigation trials.