With the development of this year's Super El Niño, extreme weather events are occurring frequently around the world, such as the heatwave in Western Europe in June that saw temperatures exceed 40°C. As climate change increases the frequency and intensity of extreme weather events-including heatwaves, torrential rains, droughts, and typhoons-the importance of more accurate climate prediction technologies is growing. In particular, the ocean covers approximately 70% of the Earth's surface and serves as a key factor in determining seasonal and long-term climate variations-including El Ni?o and La Ni?a-by storing and circulating vast amounts of heat and carbon while continuously interacting with the atmosphere. However, existing ocean prediction models require massive supercomputing resources and lengthy computation times to solve complex physical equations, which has limited their ability to provide rapid forecasts and analyze various climate scenarios.
A research team led by Dr. Kang Daehyun at the Center for Climate and Carbon Cycle Research of the Korea Institute of Science and Technology (KIST; President Oh Sang-rok) announced that it has developed KIST-Ocean, an AI-based global ocean prediction model, to overcome these limitations. This achievement is the result of the KIST Climate and Environment Research Institute, which was established to predict and respond to climate change through science and technology. It is significant in that it has successfully integrated AI technology into the field of long-term ocean and climate forecasting-which requires vast amounts of observational data and physics-based analysis capabilities-thereby achieving both high accuracy and computational efficiency.
KIST-Ocean, developed by the research team, is an AI model that predicts future ocean conditions by learning from global 3D ocean data accumulated over the past several decades. Based on various physical parameters of the global ocean-such as sea surface temperature, ocean currents, and salinity-it forecasts the three-dimensional state of the ocean at five-day intervals and can generate changes down to a depth of 600 meters. In particular, by applying the latest AI technology, it can generate approximately 200 days' worth of global ocean forecast results in just a few seconds using a single GPU, significantly reducing computation time and costs compared to existing numerical models. This high computational efficiency is expected to be effectively utilized in research involving the iterative analysis of various climate scenarios or the performance of large-scale ensemble forecasts.
The research team conducted various experiments to verify how accurately KIST-Ocean replicates actual oceanic physical phenomena. In a virtual wind-generation experiment, ocean waves and upwelling and downwelling phenomena resulting from atmospheric changes were observed to align with existing ocean physics theories, demonstrating that AI can effectively reflect the complex physical interactions between the atmosphere and the ocean-going beyond simply learning from historical data. Furthermore, using the 2015 Super El Niño-a representative climate phenomenon-as a case study, the model successfully reproduced key developmental processes, such as the rise in sea surface temperatures in the equatorial Pacific and changes in the internal heat distribution of the ocean, thereby demonstrating the predictive accuracy and reliability of AI-based ocean models.
The research team expects KIST-Ocean to serve as a foundational technology that expands the scope of AI applications beyond existing short-term weather forecasts to include seasonal and annual climate forecasts. It is expected not only to lay the groundwork for the development of AI-based Earth system models that integrate the atmosphere, ocean, and land surface, but also to lower the barriers to entry for ocean and climate research and accelerate various studies on climate change adaptation, thanks to its speed and cost-effectiveness.
Dr. Kang Daehyun of KIST stated, "Through KIST-Ocean, we have demonstrated that AI predictions not only exhibit outstanding efficiency and accuracy but can also realistically reproduce the complex physical relationships between the atmosphere and the ocean," adding, "By actively utilizing these AI models, we will be able to significantly enhance our capacity to respond to future climate crises." Responding to climate disasters is a national public research area that requires a long-term accumulation of research and continuous investment. KIST plans to further refine this technology and develop it into a proactive forecasting tool that contributes to ensuring public safety and minimizing socio-economic damage in the era of the climate crisis.
###
KIST was established in 1966 as the first government-funded research institute in Korea. KIST now strives to solve national and social challenges and secure growth engines through leading and innovative research. For more information, please visit KIST’s website at https://kist.re.kr/eng/index.do
This study was conducted with support from the Ministry of Science and ICT (Minister Bae Kyung-hoon) as part of KIST's major projects, including the Project for the Advancement of Extreme-Performance Computing (NRF-2022M3K3A1094114). The findings of this study were published in the latest issue of the international academic journal *Science Advances* (IF 13.9, 8.2% in the JCR field).
Science Advances
Data-driven global ocean model resolving atmospherically forced ocean dynamics
12-Jun-2026