Introduction:

Artificial intelligence has advanced global weather forecasting, outperforming traditional numerical models inboth accuracy and computational efficiency. Nevertheless, extending predictions beyond subseasonal timescalesrequires the development of deep learning (DL)–based ocean-atmosphere coupled models that can realisticallysimulate complex oceanic responses to atmospheric forcing. This study presents KIST-Ocean, a DL- based globalthree-dimensional ocean general circulation model. Comprehensive evaluations demonstrate the model’s robustocean simulation skill and efficiency. Moreover, it reproduces ocean responses, such as Kelvin and Rossby wavepropagation, and vertical motions induced by wind stress curl, demonstrating its ability to represent key atmo-spherically forced ocean dynamics underlying climate phenomena, including the El Niño–Southern Oscillation.These findings reinforce confidence in DL-based global weather and climate models by demonstrating their ca-pacity to capture essential ocean- atmosphere relationships. Building on this foundation, the present study pavesthe way for extending DL-based modeling frameworks toward integrated Earth system simulations, thereby offer-ing substantial potential for advancing long-range climate prediction capabilities.

pdf : Kim et al.(2026)