OceanLight achieves pointwise forecast accuracy and kinetic energy spectral fidelity exceeding both operational numerical analyses and state-of-the-art AI-based models, while surpassing all AI-based ocean models in geostrophic balance consistency.
Abstract
Reliable global ocean forecasting is critical for climate monitoring, marine navigation, and extreme event early warning. Physics-based ocean forecasting models impose prohibitive computational costs, while existing deep learning approaches predominantly rely on structured-grid architectures, incurring unnecessary computation on masked land cells and enforcing uniform resolution across dynamically heterogeneous ocean regions regardless of local flow complexity. Here we present OceanLight, an efficient global ocean forecasting framework innovatively combining geometry-adaptive unstructured mesh tokenization with a graph neural network (GNN) backbone. OceanLight achieves pointwise forecast accuracy and kinetic energy spectral fidelity exceeding both operational numerical analyses and state-of-the-art AI-based models, while surpassing all AI-based ocean models in geostrophic balance consistency. Furthermore, OceanLight demonstrates reliable mesoscale eddy representation, capturing coherent ocean structures beyond pointwise statistical optimization. These capabilities are delivered with a 62% reduction in GPU memory consumption and 70\% reduction in FLOPs relative to structured-grid baselines. Our unstructured mesh representation establishes a generalizable paradigm for scalable data-driven oceanography.
The first oceanic 4D sparse observation reconstruction dataset, named OceanVerse, is presented, providing a novel large-scale dataset that meets the MNAR (Missing Not at Random) condition, supporting more effective model comparison, generalization evaluation and potential advancement of scientific reconstruction architectures.
Bin Lu, Jingjing Shen, Ze Zhao et al.· Proceedings of the 32nd ACM...· 0 citations
OceanBench is a benchmark designed to evaluate and accelerate global short-range data-driven ocean forecasting, constructed from a curated dataset comprising first-guess trajectories, nowcasts, and atmospheric forcings from operational physical ocean models, typically unavailable in public datasets due to assimilation cycles.
Anass El, Quentin Gaudel, Juan Emmanuel Johnson et al.· Advances in Neural Informati...· 7 citations· ⚡2
OceanDepths is introduced, the first open, global, regridded AI-ready dataset that pairs satellite-derived sea surface temperature, sea surface salinity, and sea surface height products with co-located EN4 subsurface temperature and salinity profiles, complemented by matched GLORYS12 ocean reanalysis data to support comparisons or multi-stage learning.
Simon Donike, Ruben Cartuyvels, A. I. Ferola et al.· 0 citations
Ocean forecasting is crucial for both scientific research and societal benefits. Large artificial intelligence (AI)-based models have recently boosted forecasting efficiency and accuracy. However, it remains challenging to develop a comprehensive AI-driven ocean forecasting system capable of integrating cross-spatiotemporal and atmospheric forcing. This study introduces LangYa, a cross-spatiotemporal and atmospheric forcing ocean forecasting system featuring: (1) a large-language-model-based (LLM-based) time embedding to explicitly represent forecast lead times, (2) an asynchronous cross-iterative random sampling strategy to represent the impacts of atmospheric forcing on ocean processes, (3) an ocean self-attention module to enhance network stability and accelerate training convergence, and (4) an adaptive loss function to capture ocean dynamics in the thermocline, at depths ranging from tens of meters to about 300 m. LangYa is trained on 27 years of global ocean data from the Global Ocean Reanalysis and Simulation, version 12 (GLORYS12). Using reanalysis and observational data, compared to existing open-source AI-based forecasting systems and numerical models, LangYa enables a single model to produce forecasts with lead times of 1 to 7 d (1/12°, daily) and achieves 7 d RMSEs below 0.0736 m/s, 0.0701 m/s, 0.4376 ℃, and 0.1302 psu for global currents, temperature, and salinity respectively. These quantitative results indicate that LangYa provides clear advantages in forecast accuracy, lead-time robustness, and stability for global OSV forecasting, demonstrating its potential for real-time operational deployment.
Nan Yang, Chong Wang, Zimeng Zhao et al.· Science Bulletin· 0 citations
Regional high-resolution ocean environmental forecasting combines spatial numerical modeling with temporal prediction, and is essential for monitoring the ecological security of specific ocean regions. In recent years, deep learning methods are generally more computationally efficient than traditional numerical models and enable fast, accurate forecasting. However, as data resolution increases, the training and computational costs of existing approaches increase substantially. To address this issue, we introduce Slow-OCast, a transfer-learning based model designed for high-resolution ocean environmental forecasting. Specifically, Slow-OCast incorporates the slow-varying motion characteristics of the ocean and comprises two insightful modules. The Fluid Motion Separator that injects low-frequency background dynamics into the fine-tuning process of a foundation model, functioning as a "magnifier" to encode physical priors of ocean dynamics. The Hydrokinetic Energy Path Integrator that provides an implicit representation of flow-field evolution, serving as a "compass" to guide accurate change prediction. We evaluate Slow-OCast on two high-resolution Mediterranean datasets, and results demonstrate Slow-OCast consistently outperforms all baseline methods across forecasting tasks with different lead times.
Qixiu Li, Xiang Zhu, Xiaoyong Li et al.· Proceedings of the 32nd ACM...· 0 citations
Accurate regional forecasting of extreme precipitation remains difficult because the scales that control disaster-producing rainfall are neither fully resolved by global numerical weather prediction nor reliably preserved by current global artificial intelligence weather models. Global AI systems such as Pangu-Weather, GraphCast, GenCast, and related models have transformed medium-range forecasting skill and computational efficiency, yet they remain fundamentally constrained by coarse training targets, regression-induced smoothing, and limited direct representation of terrain-locked convection and local hydrometeorological extremes. This paper reconstructs and substantially extends an event-based manuscript on AI-driven regional forecasting into a submission-oriented framework centered on the more defensible idea of physics-aware regional extreme-weather forecasting or downscaling with open climate data. The core argument is that AI should not be treated as a wholesale substitute for high-resolution regional physics; rather, it should be used as a skillful large-scale predictor whose state can be physically harmonized and injected into a regional nonhydrostatic model. We therefore formalize an AI-initialized, physics-aware dynamical downscaling pipeline in which open global reanalysis and observation products are used to generate, constrain, and evaluate regional forecasts of extreme rainfall. The framework is instantiated using the published North China July–August 2023 extreme precipitation case, for which the original study compared WRF simulations driven by Pangu forecasts against WRF simulations driven by NCEP GFS forecasts across lead times of 0.5, 3.0, and 5.5 days. This paper contributes in three ways. First, it repositions the original study within the modern literature on AI weather forecasting, regional downscaling, and physically constrained machine learning. Second, it formulates the coupling problem mathematically, clarifies the state alignment needed to make AI forecasts dynamically usable by WRF, and introduces a coherent reliability-oriented evaluation logic based on error growth, threshold skill, and event-structure consistency. Third, it reorganizes the experiments and results into a rigorous narrative grounded in reproducibility. Using the published event-level metrics, the AI-initialized regional system outperforms the GFS-initialized counterpart at extended lead times. For the North China case, the maximum precipitation threshold retaining a Threat Score of at least 0.1 is 400 mm at 5.5-day lead for Pangu-initialized WRF, whereas the GFS-driven counterpart retains comparable skill only at 50 mm. At 0.5-day lead, both systems perform competitively, but the AI-driven system still exhibits stronger spatial correlation (0.76 versus 0.68) and lower RMSE (86.2 mm versus 96.4 mm). The evidence supports a restrained but important conclusion: physics-aware AI-initialized regional modeling is a promising route for long-lead extreme-weather forecasting, yet current evidence remains case-limited and should be interpreted as a strong event-based demonstration rather than universal proof of general superiority.
Mina Annetta, Shanti Purohit, Viljar Vagle et al.· International Journal of Inf...· 0 citations