Analysis of fluctuating characteristics of temperature field in the shallow water on multiple temperature and depth sensor

摘要

The temperature in the sea is an oceanographic variable that indirectly determines many of the peculiarities of sound transmission in the medium. It varies with depth, the seasons, geographic location, and also time at a fixed location. In order to obtain the temporal and spatial statistical characteristics of temperature, the temperature field in the summer of 2013 was observed continuously at certain shallow water locations in the North China Sea, using vertically distributed multiple temperature and depth sensors. The Self-Organizing Map (SOM) is an unsupervised neural network based on competitive learning, and can solve the problem when the center of clustering is unknown. The SOM's basic theory and algorithm are studied in this paper. A simulation example is given to prove the feasibility of SOM in oceanographic data cluster analysis. At the same time, reliable temperature samples of different depths and time are processed by SOM. Based on the clustering results over a long period or over a large area, multivariate linear regression analysis is carried out, and a mathematic model is established which describes the relationship between depth and temperature in shallow water. Finally, the change of temperature structure and its influence on sonar detection is simulated on the basis of the Fast Field Program (FFP) at some particular frequencies, and this is then discussed in detail. These results are of particular importance for hydrographic statistic characteristics investigation and sonar performance prediction in shallow water.