Ocean Acoustic Tomography with Moving Node Based on Tikhonov Regularized Kalman Filter

Abstract

Large scale and long period observation of ocean state can be a really tough task using direct measurement. Ocean acoustic tomography (OAT) is an impactful method to monitor large scale ocean state (sound speed, current etc.). This paper studies a sequential tracking approach for the tomography problem from travel-time data in deep water. Tomographic networks combined both fixed mooring receivers and a moving source are considered, which provide a sufficient information to estimate the three-dimensional (3D) sound speed field (SSF). Empirical orthogonal function (EOF) coefficients are used to parameterize the SSF, and a first order autoregressive (AR(1)) model is formed to describe the evolution of sound speed changes in short interval. So that linear state-space model can be established combining ray travel time inversion which means the inversion could be solved under data assimilation framework. Furthermore, in order to overcome the illness of observation matrix in the tomography tracking problem, Tikhonov regularization is integrated into the basic Kalman filter, which induced a Tikhonov regularized Kalman filter (TRKF). Simulations are conducted to test the feasibility of proposed tracking approach in deep water.