Bin Lu

Machine learning (ML) methods have emerged as a promising approach for dissolved oxygen (DO) mapping. However, the uncertainties arising from model configurations such as hyperparameter setting and random initialization remain insufficiently characterized. Here we quantify this uncertainty through an efficient ensemble intercomparison. ML-based DO mapping uncertainty spatially peaks in the oxycline and temporally increases during the periods of historical observation scarcity. These patterns are primarily driven by intrinsic DO variability and extrinsic observing strategy. Our results further show that with the deployment of the Argo-O ₂ floats, global uncertainty has been reduced by 37.8% over the last two decades. However, uncertainty reductions are regionally uneven: substantial decreases occur mainly where sampling is spatially uniform, including the Atlantic, Indian, and Southern Oceans. These findings emphasize that enhancing spatial uniformity is equally important as increasing the number of observations, positioning uncertainty analysis as a bridge between understanding DO variability and designing future observing-systems.

Dissolved nitrate, phosphate, and silicate are fundamental drivers of marine primary productivity and the biological carbon pump. However, the development of continuous, long-term global datasets has long been severely hindered by extreme historical data sparsity and complex biogeochemical dynamics. Statistical interpolation methods struggle to simultaneously fill the severely sparse data gaps and capture the non-linear interactions, necessitating advanced artificial intelligence (AI) to explicitly learn and leverage their underlying relationships. Nevertheless, most existing AI methods reconstruct nutrients independently (i.e., Single-Task Learning), failing to exploit the synergistic effects inherent in cross-nutrient stoichiometry. In this study, we present Jingwei-Nutrients , a four-dimensional global monthly data product at 1 ° ×1 ° resolution from 0 to 2000 m depth spanning 1965 to 2023, reconstructed using a Transformer-based Multi-Task Learning (MTL) framework trained on a comprehensive, quality-controlled multi-source observational database. Evaluation on chronological K -fold cross-validation yields mean validation R ² values across folds of 0.991, 0.969, and 0.990, with RMSEs of 1.295, 0.163, and 4.109 µmol kg ⁻¹ for nitrate, phosphate, and silicate, respectively. Temporal K -fold cross-validation reveals that the MTL framework consistently achieves higher R ² and lower RMSE for all three nutrients compared to single-task models, with larger accuracy gains in data-sparse earlier decades such as 1965–1975. Our data product reproduces consistent global climatology patterns and seasonal cycles with World Ocean Atlas (WOA). Furthermore, independent evaluations against long-term monitoring stations (HOT and KERFIX) and GO-SHIP cruise sections (P16N, P16S, and P06E) demonstrate our effectiveness across multi-decadal temporal trends, spatial variability and vertical changes. Additionally, an ensemble-based uncertainty analysis reveals interpretable spatial heterogeneities and a long-term decreasing trend in global uncertainty, which directly mirrors the historical transition from sparse early sampling to modern observing networks. This product fills a critical gap in historical ocean biogeochemical observations, providing a reliable, physically consistent foundation for marine biogeochemical modeling and climate change studies. The product is openly available at https://doi.org/10.5281/zenodo.21066027 (Wang et al., 2026).

Dissolved nitrate, phosphate, and silicate are fundamental drivers of marine primary productivity and the biological carbon pump. However, the development of continuous, long-term global datasets has long been severely hindered by extreme historical data sparsity and complex biogeochemical dynamics. Statistical interpolation methods struggle to simultaneously fill the severely sparse data gaps and capture the non-linear interactions, necessitating advanced artificial intelligence (AI) to explicitly learn and leverage their underlying relationships. Nevertheless, most existing AI methods reconstruct nutrients independently (i.e., Single-Task Learning), failing to exploit the synergistic effects inherent in cross-nutrients stoichiometry. In this study, we present Jingwei-Nutrients, a global monthly dataset at resolution from 0 to 2000 m depth spanning 1965 to 2023, reconstructed using a Transformer-based Multi-Task Learning (MTL) framework trained on a comprehensive, quality-controlled multi-source observational database. Evaluation on the validation set yields values of 0.980, 0.961, and 0.983, with RMSEs of 2.21, 0.23, and 6.35 for nitrate, phosphate, and silicate, respectively. Temporal K-fold cross-validation reveals that the MTL framework consistently achieves higher and lower RMSE for all three nutrients compared to single-task models, with larger accuracy gains in data-sparse earlier decades such as 1965–1975. Our dataset reproduces consistent global climatology patterns and seasonal cycles with World Ocean Atlas (WOA). Furthermore, independent evaluations against long-term monitoring stations (HOT and KERFIX) and GO-SHIP cruise sections (P16N, P16S, and P06E) demonstrate our effectiveness across multi-decadal temporal trend, spatial variability and vertical changes. Additionally, an ensemble-based uncertainty analysis reveals interpretable spatial heterogeneities and a long-term decreasing trend in global uncertainty, which directly mirrors the historical transition from sparse early sampling to modern observing networks. This dataset fills a critical gap in historical ocean biogeochemical observations, providing a reliable, physically consistent foundation for marine biogeochemical modeling and climate change studies. The dataset is openly available at https://doi.org/10.5281/zenodo.19491198.

Ocean dissolved oxygen (O₂) is an essential climate variable crucial for sustaining the marine life; thus, changes of O₂ at various spatiotemporal scales should be quantified and understood. Here, we study the climatology and annual cycle of O₂ at regional to global scales using eight available gridded observational products. These datasets are generated by different groups using different primary data selection, quality control, bias correction, and interpolation methods, including statistical and machine-learning-based mapping methods. A common set of metrics was collaboratively developed by the community of the Gridded Observational Dataset Intercomparison Project-Dissolved Oxygen (GODIP-DO) to facilitate the inter-comparison. We find that global mean O₂ profiles are consistent among all products (±3 µmol kg⁻¹), with the well-established decrease from high surface values to a minimum ∼ 1000 m, and subsequent increase to higher O₂ at depth, although local differences could reach ±25 µmol kg⁻¹ (0–1000 m). The hemispheric O₂ annual cycle correlates strongly with ocean temperature changes, suggesting the key driver of temperature for the O₂ annual cycle. However, there is substantial variation in the global mean 0–100 m O₂ annual cycle, the magnitude ranges from −1 to 0.8 µmol kg⁻¹, with a standard deviation of the datasets of ∼ 0.3 µmol kg⁻¹. Average oxygen minimum zones (OMZ) volume among the products is 80.92 × 10⁶ km³ (±1.95 %) for a 60 µmol kg⁻¹ threshold and 152.00 × 10⁶ km³ (±1.72 %) for a 90 µmol kg⁻¹ threshold. Our results help to depict and understand the spread among the available O₂ gridded datasets.

Ocean dissolved oxygen (O₂) is an essential climate variable crucial for sustaining the marine life; thus, changes of O₂ at various spatiotemporal scales should be quantified and understood. Here, we study the climatology and annual cycle of O₂ at regional to global scales using eight available gridded observational products. These datasets are generated by different groups using different primary data selection, quality control, bias correction and interpolation methods including statistical and machine-learning based mapping methods. A common set of metrics was collaboratively developed by the community of Gridded Observational Dataset Intercomparison Project-Dissolved Oxygen (GODIP-DO) to facilitate the inter-comparison, which allows assessing the robustness of the uncertainties through the spread of the products. Global mean O₂ profiles are consistent among all products (±3 µmol kg^-1), with the well-established decrease from high surface values to a minimum ~1000 meters, and subsequent increase to higher O₂ at depth, although local differences could reach ±25 µmol kg^-1 (0–1000 m). The hemispheric O₂ annual cycle correlates strongly with ocean temperature changes, suggesting the key driver of temperature for the O₂ annual cycle. However, there is substantial variation in the global mean 0–100 m O₂ annual cycle, the magnitude ranges from -1 to 0.8 μmol kg^-1, with a standard deviation of the datasets of ~0.3 μmol kg^-1. Average oxygen minimum zones (OMZ) volume among the products is 80.92 × 10⁶ km³ (±1.95 %) for a 60 µmol kg^-1 threshold and 152.00 × 10⁶ km³ (±1.72 %) for a 90 µmol kg^-1 threshold. Our results serve as a starting point for resolving the uncertainty budget of the ocean O₂ changes.