On March 18, 2026, marking the 60th anniversary of the founding of the Second Institute of Oceanography, Ministry of Natural Resources (SIO), the State Key Laboratory of Satellite Ocean Environment Dynamics officially released its independently developed Seamless Integrated Oceanographic Dataset ("SEEDS").
For a long time, Chinese marine observation system has been constrained by bottlenecks including poor temporal and spatial matching between satellite and sea-based data, inadequate data collaborative fusion, and incomplete linkage of the research and development chain. In addition, the discontinuation of updates to several major international marine databases has made it an urgent task to build high-quality independent and controllable marine datasets.
In response to the national strategic initiative to improve the integrated space-air-ground-sea observation network, and as a domestic pioneer in three rounds of technological innovation in marine observation, the laboratory leverages its technical strengths in satellite remote sensing, Argo floats and intelligent agile observation to launch the SEEDS multi-scale comprehensive marine dataset series.
Dedicated to realizing full-chain development covering mechanism research, observation monitoring and early warning, this dataset series provides solid data support for national marine strategies, local economic development and global ocean governance.
The SEEDS (Seamless Integrated Oceanographic Dataset) series achieves seamless coverage across global, ocean basin, regional and estuarine-bay scales, marking a major breakthrough in China’s development of independent and controllable marine datasets.
Global Scale
-
BOA Argo Global Gridded Temperature and Salinity Dataset
Built on global Argo float observations including those deployed by the Second Institute of Oceanography, this dataset features a horizontal resolution of 1°×1°, 58 vertical layers, a depth range of 0–2000 meters, and monthly mean products dating back to 2004. Adopting the improved Barnes successive correction method, it retains mesoscale oceanic signals to the maximum extent compared with international counterparts.
-
Global Lagrangian Eddy Dataset (GLED)
It accurately depicts the life cycle and three-dimensional structures of coherent ocean eddies from 1993 to 2020, covering eddy position, moving velocity, temperature-salinity anomalies and other key variables. Constructed based on satellite-derived geostrophic currents and Lagrangian framework, it overcomes the limitation of traditional Eulerian datasets in accurately identifying and tracking coherent ocean structures.
-
Global Sea Surface Nitrate Dataset
Combining in-situ nitrate measurements from BGC-Argo and WOD2023, MODIS satellite products and CMEMS model data, this daily global dataset with 8 km spatial resolution (2002–2024) is established via the XGBoost machine learning algorithm. Its retrieval accuracy is approximately 50% higher than similar products released by the European Copernicus Marine Service.
Ocean Basin Scale
-
Annual Full-Coverage Polar Ocean Color Remote Sensing Dataset (NN-LAT50)
Based on remote sensing reflectance data and neural network training, this monthly polar ocean color dataset with 4 km resolution spans 2003 to 2022. It fills the observational gaps of mainstream international products such as MODIS/Aqua, VIIRS and OLCI in polar regions during winter.
-
Indian Ocean Dissolved Oxygen Dataset (TransOxygen_Indian)
Using physically constrained Transformer deep learning models and multi-source environmental data fusion, it generates monthly dissolved oxygen products for the Indian Ocean, with 0.5° spatial resolution, 150 vertical layers and 0–2000 m depth range (2005–2023). Its RMSE is reduced by 10.5% compared with state-of-the-art baseline machine learning models.
-
Northwest Pacific Sea Surface Nitrate Dataset
Integrating multi-source fused ocean color data (OC-CCI) and sea surface temperature data (OISST), daily nitrate products at 4 km resolution (1998–2024) are retrieved via independently developed stacked random forest algorithm. It boasts better spatio-temporal integrity than single-satellite products and can precisely capture the impacts of mesoscale processes on nutrients.
Regional Scale
-
3D Temperature-Salinity Gridded Dataset for Bohai, Yellow and East China Seas (BYES_Atlas v2)
Fusing observational data from 18,959 in-situ stations across 46 cruises, monthly climatological products with 0.25° horizontal resolution and 17 vertical layers (2005–2009) are produced by optimal interpolation objective analysis. It shows significantly enhanced capability in characterizing typical marine phenomena such as coastal estuarine upwelling compared with domestic and overseas peer datasets.
-
East China Sea Surface Chlorophyll-a Dataset (LMC)
Based on over 2,000 field observation stations, localized remote sensing correction for inshore water reflectance is adopted to generate daily, monthly and climatological sea surface chlorophyll-a products at 4 km resolution (1998–2024). It maintains high accuracy in both Case I and Case II waters of the East China Sea, more completely captures nearshore features and Changjiang Diluted Water signals than standard SeaWiFS and MODIS chlorophyll products, and supports quantitative research on ecological dynamic processes.
Estuary & Bay Scale
-
Remote Sensing Products of Ocean Color Anomalies at Coastal Drainage Outlets
Integrating in-situ data covering diverse discharge scenarios and Sentinel-2 imagery, it establishes a library of 14 types of in-water optical spectra. Combined with unsupervised clustering and random forest algorithms, it realizes rapid automatic identification of water color anomalies. To date, 41 abnormal water color zones have been successfully detected. With 10 m spatial resolution and 5-day temporal resolution since 2022, it covers more than 4,000 coastal drainage outlets along Zhejiang coast.
-
Bay Eutrophication Remote Sensing Dataset
Fusing Sentinel-2 satellite imagery and field water quality measurements, high-precision retrieval models are built via machine learning. It supports the generation of eutrophication grading thematic maps and trend analysis reports, providing refined and dynamic scientific support for the construction of Beautiful Bays. Key variables include dissolved inorganic nitrogen, reactive phosphate, chemical oxygen demand and eutrophication index. Covering six typical bays in Zhejiang Province (Hangzhou Bay, Xiangshan Bay, Sanmen Bay, Taizhou Bay, Leqing Bay and Wenzhou Bay), it provides 5-day and monthly mean products starting from 2022 at 10 m resolution.
-
Offshore Water Quality Classification Remote Sensing Dataset
Multi-source remote sensing data including Sentinel-3 and GOCI are integrated for parameter retrieval via deep learning methods. DINEOF (Data Interpolating Empirical Orthogonal Functions) is innovatively applied for data gap filling to improve spatio-temporal continuity. Complying strictly with China’s
Sea Water Quality Standard, it delivers standardized water quality services covering dissolved inorganic nitrogen, reactive phosphate and water quality classification. With 300 m spatial resolution, it covers Zhejiang offshore waters and provides 5-day and monthly products since 2022.
The SEEDS dataset series has demonstrated prominent application values in multiple fields. Globally, its global temperature-salinity gridded products support numerous studies on global ocean heat content and sea level variation assessment. In global marine governance, its offshore China temperature-salinity datasets facilitate the laboratory’s participation in the Third World Ocean Assessment initiated by the United Nations, offering fundamental data for evaluating long-term dynamic and ecological changes in the Changjiang Estuary.
For institutional business services under the Ministry of Natural Resources, it underpins the operational hypoxia and ocean acidification early warning system and provides core data for the official China Marine Ecological Early Warning and Monitoring Bulletin. At local service level, its datasets on coastal drainage discharge and offshore water quality in Zhejiang serve as core data resources for Zhejiang’s provincial digital government platform Zheli Blue Ocean, winning high recognition from provincial authorities.
Adhering to the core philosophy of Integrated SEEDS Wisdom, Guarding the Blue Ocean, the State Key Laboratory of Satellite Ocean Environment Dynamics will continue to fulfill its mission of serving national strategic demands, independently develop high-level and controllable marine datasets, contribute core technological strength to building a powerful maritime nation, and lay a solid marine foundation for realizing the Chinese Dream of national rejuvenation.
