Automatic Detection of the Number of Raypaths in a Shallow-Water Waveguide

Correct identification and tracking of stable raypaths are critical for shallow-water acoustic tomography. Separating raypaths using high-resolution methods has been presented to improve resolution ability based on the prior knowledge of the number of raypaths. It is clear that the precise knowledge of the number of raypaths largely determines the separation performance. Therefore, a noise-whitening exponential fitting test (NWEFT) using short-length samples is proposed in this paper to automatically detect the number of raypaths in a shallow-water waveguide. Two information-theoretic criteria are considered as comparative methods in terms of the capability of correct detection. Their performances are tested with simulation data and real data obtained from a small-scale experiment. The experimental results show that the NWEFT can provide satisfactory detection compared to the two classic information-theoretic criteria—the Akaike information criterion (AIC) and the minimum description length (MDL). MDL is asymptotically consistent while AIC overestimates even if analyzed asymptotically. Compared to these criteria, the proposed method is more suitable for short-length data.