AUTHOR=Zhao Qianlong , Peng Shiqiu , Wang Jingzhen , Li Shaotian , Hou Zhengyu , Zhong Guoqiang 

TITLE=Applications of deep learning in physical oceanography: a comprehensive review

JOURNAL=Frontiers in Marine Science

VOLUME=Volume 11 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/marine-science/articles/10.3389/fmars.2024.1396322

DOI=10.3389/fmars.2024.1396322

ISSN=2296-7745

ABSTRACT=Deep learning, as a new data-driven technology, has attracted widespread attention from various disciplines with the rapid development of the Internet of Things (IoT) big data, machine learning algorithms and computational hardware in recent years. It has been shown to achieve comparable or even more accurate results than traditional methods in a more flexible manner in existing applications in various fields. As an important scientific field of oceanography, in the field of physical oceanography, the abundance of ocean surface data and high dynamic complexity pave the way for extensive applications of deep learning. Moreover, a considerable amount of research has already been conducted to innovate traditional approaches in topics such as ocean circulation, ocean dynamics, ocean climate, ocean remote sensing, and ocean geophysics, leading oceanographic studies into the "AI ocean era". Within this research, we categorize numerous research topics in physical oceanography (such as sea surface temperature, sea surface salinity, mesoscale eddies, and El Niño-Southern Oscillation) into four aspects: surface elements, subsurface elements, typical ocean phenomena, and typical weather and climate phenomena. We review cutting-edge applications of deep learning primarily over the past three years. We provide comprehensive insights into the development of deep learning, and from the perspective of three application scenarios, namely spatial data, temporal data, and data generation, three corresponding matching deep learning model types are introduced-convolutional neural networks (CNNs), recurrent neural networks (RNNs), and generative adversarial networks (GANs)-and their principal application tasks. Furthermore, this paper discusses the current bottlenecks and future innovative prospects of deep learning in oceanography. Through summarizing and analyzing existing research, our aim is to delve into the potential and challenges of deep learning in physical oceanography, providing reference and inspiration for future researchers in oceanographic studies.