A target depth estimation method in the deep water direct arrival zone based on tucker decomposition denoising

Abstract To address the problem of large estimation errors in source depth caused by blurred Lloyd’s mirror interference (LME) fringes in the deep water direct arrival zone (DAZ) under low signal-to-noise ratio (SNR) conditions, this paper proposes an acoustic field interference fringe denoising method based on Tucker decomposition. By constructing a three-dimensional interference tensor of grazing angle-frequency-range, Tucker decomposition is used for low-rank approximation of the noisy tensor, retaining main signal components and suppressing noise. The high-SNR interference fringe structure is recovered after reconstruction. The proposed method is verified by a simulated dataset constructed based on the Kraken normal mode model. This method overcomes the limitation of insufficient information utilization in traditional two-dimensional interference patterns and improves the accuracy of target depth estimation under low SNR. However, the effectiveness of this method still requires further verification through sea trial data to ensure its practical application.