| For a long time,ocean environment monitoring has been faced the contradiction between res-olution and working distance,time-space scale and equipment resources.Sound speed in the ocean can be measured using instruments,but such measurement is only capable of obtaining the data that corresponds to points or lines among the region instead of the whole interested area.Ocean param eter evolving models are capable of predicting environment parameters in a large scale,however with a relatively great error.This article primitively explores the relationship between the tide phe-nomenon and the variations of sound speed as well as the fusion of the tide phenomenon and the information in high-precision data obtained from the distributed measurement system.The idea of relating the tide phenomenon and measurement data leads to our proposed method for sound speed field estimation under the small-scale region.Experiment data demonstrate the validity of the proposed model and algorithm.The specific work is as follow:According to the evolution characteristic of the sound speed,an analysis is conducted on the most noticeable variation pattern of sound speed in experimental data and its periodical feature is explained from a tide aspect.Furthermore,the key tidal constituent that owns the dominant influence on the variation is identified after associating the tidal characteristic in the experiment area.As long as the key tidal constituent’s movements are confirmed,it’s feasible to establish a tide-based sound speed field model which can primitively describe the space and time-evolving feature of the sound speed in a small-scale region under the influence of the tide phenomenon.In order to deal with the model mentioned above,Empirical Orthogonal Function(EOF)is adopted to decompose the sound speed field into background field and perturbation field,each part is provided with an estimation method.Kriging Interpolation,a classical spatial interpolation method is introduced to obtain the background field under the assumption that background field is relatively uniform across the small-scale region;Maximum a posteriori estimation is also devel-oped to estimate the perturbation field.In the meantime,the line measurement as well as point measurement commonly used in the distributed measurement system is also considered.The proposed model and algorithm are evaluated with experimental data.Two measurement situations mentioned above are compared under different sensor distributions and different amount of data.The result shows the effectiveness of the proposed model and algorithm along with fol-lowing conclusions:positions inside the area covered by measurement system tend to get better results,and more amount of data can help to improve the precision of the estimates-just as we have expected;The proposed method is still capable of capturing the variation of the sound speed even under the circumstances of the sparse distribution and few measured data.Finally,this paper presents the tracking and inversion of the sound speed with multi-source data which consists of acoustic data and estimated sound speed field.By combining the prior field with the acoustic data using the proposed coupled state-space model and solving the correspondent model with Markov Chain Monte Carlo-Ensemble Kalman Filter(MCMC-EnKF),inversion result of sound speed profiles along the propagation path is improved.Simulations and experiments show that the joint inversion method is more accurate and more robust than the inversion method without prior field information. |
