Inversion Of Time-Evolving Sound Speed Profiles With A Moving Sound Source

The sound speed profile (SSP) in water column can be directly measured by the instruments. However, only the speed data of local points and lines can be obtained and this way is unable to obtain the observation data of regional sea. By combing the propagation model and the observed data, the acoustic inversion technique provides a method which could estimate the interested environment parameters in a rapidly and efficiently way.According the evolution characteristics of SSPs, this paper investigates the tracking algorithms of SSPs based on the sequential filtering. In order to improve the variability of the samples in the ensemble Kalman filter (EnKF), the Markov Chain Monte Carlo (MCMC) method is introduced and an algorithm named MCMC-EnKF is proposed and applied to the SSP inversion. The results of numerical simulations and experimental data processing show that the MCMC-EnKF improves the performance of EnKF under the same computational cost, especially in the case of fast changing SSPs.In addition, in order to target the dilemma of observation range and the resolution in a fixed network, this paper puts forward an approach that utilizes a moving sound source to invert the SSPs. By employing the moving source, the local sampling points for acoustic pressure are increased and the accuracy and resolution of the estimated parameters are improved. As for the unknown quantities of inversion problem increased with the introduction of the moving source, we also explore an algorithm for large degrees of freedom of the inversion. The revised MCMC algorithm, i.e., the differential evolution adaptive metropolis (DREAM-zs) is employed to the SSP inversion. The advantage of this method for the large dimension inversion problem is tested by numerical simulations and experimental data processing.Using part of the sea trial data, the validity of the MCMC-EnKF and the DREAM-zs is further testified. The results of both algorithms are compared by that of EnKF. In addition, using the lake trial data, the SSP inversion method using a moving source is preliminarily verified.