The power of AI is illuminating the mysteries of ocean currents, specifically the Indonesian Throughflow. This current, a unique connector between the Pacific and Indian Oceans, plays a vital role in global ocean circulation and temperature regulation. However, its complexity arises from the diverse ocean processes and deep basins surrounding Indonesia, making accurate measurements a challenging task.
Wang and colleagues have developed an innovative approach, combining AI modeling with observing system simulation experiments. Their method utilizes sea surface height measurements to predict the behavior of the Indonesian Throughflow and identify the most influential straits.
The researchers employed a deep learning model, utilizing two neural network types: a convolutional neural network (CNN) and a recurrent neural network (RNN). The CNN extracted trends from the data, while the RNN processed these trends and analyzed their temporal changes. This approach offered a more efficient and cost-effective alternative to traditional observing system simulations.
The results confirmed the effectiveness of their method, accurately predicting water transport trends and identifying the Maluku Strait as a key passage with a strong influence on the entire system. The combination of information from the Maluku and Halmahera Straits further enhanced the accuracy of system-wide condition predictions.
This research not only highlights the potential of AI in oceanography but also provides valuable insights for future monitoring efforts. It's an exciting development, but it also raises questions: How can we ensure the responsible and ethical use of AI in such sensitive environmental studies? And what other hidden ocean secrets might AI uncover in the future?
Let's discuss in the comments!
Source: Journal of Geophysical Research: Machine Learning and Computation
(https://agupubs.onlinelibrary.wiley.com/journal/29935210)
Citation: Sidik, S. M. (2026), AI sheds light on hard-to-study ocean currents, Eos, 107, https://doi.org/10.1029/2026EO260027. Published on 14 January 2026.
Text © 2026. AGU. CC BY-NC-ND 3.0 (https://creativecommons.org/licenses/by-nc-nd/3.0/us/)
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.
