Analysis of Three-dimensional Thermocline Based on Argo Data

The thermocline in the ocean is an important phenomenon in marine scientific research. Its distribution affects not only the delineation of vertical boundaries of water masses, but also the activities of underwater vehicles, the use of underwater acoustic instruments and the development of marine fisheries. Therefore, it is significant to determine the boundary of thermocline accurately. This paper adopts the K-means clustering method in machine learning to analyze the Argo data of the sea area (10.5 degrees similar to 25.5 degrees S, 55.5 degrees similar to 80.5 degrees S), and the distribution stereogram of this sea area data is obtained to determine the highly possible area of thermocline. Then the traditional thermocline determination method is combined with the information entropy method in machine learning to determine the thermocline and its boundary in this area. Finally, the occurrence area and boundary position of thermocline are obtained. In the future, this method can be applied to determine the thermocline in different areas, and then merge the regional thermoclines together to eventually realize the determination of the three-dimensional thermocline in the global sea area.