Study Sheds Light on Estimation of Oceanic Barrier Layer Structures
The oceanic barrier layer is situated between the base of the ocean's density mixed layer and the top of the isothermal layer, with its thickness fluctuations directly influencing the ocean's vertical mixing process. Changes in the barrier layer impact the transport of heat and salinity within the ocean, subsequently affecting regional weather and climate patterns.
Due to the limitations of observational data, acquiring high-quality empirical data of the oceanic barrier layer remains challenging. Hence, employing high-resolution satellite remote sensing data in tandem with observations to estimate key oceanic structures has become a pivotal topic in physical oceanography.
Recently, the research team led by Prof. YIN Baoshu from the Institute of Oceanology of the Chinese Academy of Sciences (IOCAS) has achieved significant advancements in the estimation of oceanic barrier layer structures in cooperation with scholars from the University of California, Los Angeles (UCLA).
The study was published in Environmental Research Communications on Sep. 25.
Researchers employed advanced meta-learning techniques, successfully integrating Convolutional Neural Networks (CNN), Gated Recurrent Units (GRU), and Artificial Neural Networks (ANN) into a novel multi-model ensemble approach. This has markedly enhanced the accuracy of the estimation of the oceanic barrier layer's structure.
By utilizing key sea surface parameters such as Sea Surface Temperature (SST), Sea Surface Salinity (SSS), and Sea Surface Wind speed (SSW), the research team was able to accurately reconstruct the structure of the oceanic barrier layer depth, with significantly reduced root mean square errors in the southeast Arabian Sea, Bay of Bengal, and eastern equatorial Indian Ocean.
"This achievement not only overcomes the limitations of traditional observational techniques and numerical models but also highlights the immense potential and broad application prospects of machine learning, especially meta-learning, in marine research," said QI Jifeng, first author of the study.
"Furthermore, this study offers invaluable support and contribution towards a deeper understanding of ocean dynamics, promoting research into oceanic environmental changes, and addressing global climate change, holding significant academic value and vast practical application prospects," added Prof. YIN.
This research was jointly funded by the National Key Research and Development Program of China and the National Natural Science Foundation of China.
Flowchart of the Meta-learning-based model for estimating the barrier layer thickness in the tropical Indian Ocean
Jifeng Qi, Tangdong Qu, and Baoshu Yin. (2023). Meta-learning-based Estimation of the Barrier Layer Thickness in the Tropical Indian Ocean. Environ. Res. Commun.
(Text by QI Jifeng)
Media Contact:
ZHANG Yiyi
Institute of Oceanology
E-mail: zhangyiyi@qdio.ac.cn
(Editor: ZHANG Yiyi)
