Source depth estimation using modal decomposition and time-frequency representations

Source localization in shallow water environment is a crucial issue in underwater acoustics. Many methods for the source depth estimation use modal decomposition. Most of them use a multi-sensor array (vertical or horizontal). We propose a method based on time-frequency representations and matched mode processing. It needs thus only one sensor. Modes amplitudes are estimated on time-frequency representation starting for a theoretical model of waveguide. Depth estimation is done comparing estimated modes amplitudes with predicted ones extracted from simulations at different depths. The robustness against noise is discussed. Validation on a real dataset is given.