Chlorophyll-a Analysis Based on Hilbert Huang Transformation and Convolutional Neural Network

The ocean and climate system are non-linear and non-stationary system for the earth. It is inadvisable to consider the linear and stable changing process, because it cannot greatly reflect the inherent non-linear characteristics of the system itself. On the other hand, there is big data of the system. Traditional methods cannot characterize them well. Therefore, this paper focuses on using Hilbert-Huang Transformation (HHT) and Convolutional Neural Network (CNN) to analyze the chlorophyll-a content of ocean and climate by satellite remote sensing observations, and discusses the nonlinear characteristics and seasonal changing trend in ocean and climate data. The trend which changes characteristics of chlorophyll-a in the Pacific Ocean is analyzed. The results show that the spatial mode of chlorophyll-a mainly depends on the distribution of phytoplankton, which is determined by the degree of light and nutrient supply. Chlorophyll-a presents a significant upward trend in the western Pacific warm pool area. In the equatorial Pacific divergence zone, chlorophyll-a shows a significant downward trend. However, the relationship between sea surface temperature and the changing trend of chlorophyll-a still needs to be further confirmed, and the long-term effects of global warming caused by human activities on the global ocean chlorophyll-a need to be explored in depth.