Frontiers in ocean non-equilibrium dynamics and challenges in operational oceanography application

Ocean circulation exhibits complex multiscale dynamics, among which non-equilibrium processes, marked by transient behavior, strong perturbations, and pronounced nonlinearity, are of particular importance for maritime environmental support. However, limited observational capability has long constrained the study of subsurface non-equilibrium dynamics, leaving numerical models insufficiently validated and positioning this issue at the forefront of modern physical oceanography. Artificial intelligence (AI) offers a promising pathway to overcome this bottleneck. Built on this concept, the “iEddy” AI large ocean model can efficiently reconstruct high-resolution, high-accuracy three-dimensional temperature, salinity, and velocity from sea-surface remote sensing data, demonstrating strong capability in capturing non-equilibrium ocean currents and providing a new technological paradigm for marine environmental support. Nevertheless, observational gaps and the lack of a complete data-model-application loop remain major challenges. Developing an integrated “observation-assimilation-reconstruction-forecast-application” intelligent ocean environmental support system that enables a fully operational, automatically iterating workflow will be key to achieving transformative improvements in marine environmental support capabilities.