EA-OWT

EA-OWT (East Asia Offshore Wind Turbine Dataset) provides georeferenced offshore wind turbine records for East Asian waters from 2019 to September 2025. The dataset is distributed in CSV and ESRI Shapefile formats using WGS 84 (EPSG:4326). The annual layers are cumulative: once a turbine is assigned a First_Year, it is retained in subsequent annual layers.First_Year represents the earliest year in which a turbine can be consistently identified in the available satellite imagery. It should be interpreted as an imagery-inferred first-appearance year rather than a verified installation, commissioning, or grid-connection date.The principal EA-OWT/ folder contains seven annual data groups, EA-OWT_2019 through EA-OWT_2025. Each annual group is provided as a CSV file and the corresponding ESRI Shapefile components. The released records use the same core fields across years: Turb_ID, Class, Country, Type, Cap_MW, Rotor_m, First_Year, Longitude, and Latitude. Unavailable attribute values are left blank.As of September 2025, EA-OWT contains 7,919 offshore wind turbine locations. Country is available for all records, while Type, Cap_MW, and Rotor_m are available for 5,260 (66.4%), 7,917 (99.97%), and 7,287 (92.0%) turbines, respectively. These attributes were manually compiled and cross-checked from manufacturer, government, project-owner, environmental-assessment, and other public technical records. Rotor-diameter values obtained from The Wind Power are redistributed with written permission under CC BY 4.0 with attribution.The Validation/ folder contains the 2025 product-evaluation dataset, including 2,781 EA-OWT predictions, 2,781 independently interpreted reference turbines, 30 validation cells, a validation README, and two reproduction scripts. Under the primary 200-m one-to-one matching criterion, the evaluation produced TP = 2,780, FP = 1, and FN = 1, corresponding to Precision, Recall, and F1 scores of 99.96%. At a 50-m matching radius, all three metrics were 99.89%. For the 2,780 matched turbines, the median geodesic separation was 7.20 m, the mean was 8.83 m, the 95th percentile was 21.04 m, and the maximum was 89.37 m. These results describe point-level agreement within the selected validation cells and should not be interpreted as a measure of complete spatial coverage across the full study domain.The Validation-19/ folder contains the evaluation subsets used specifically for the 2019 cross-dataset comparison. It includes the EA-OWT evaluation subset, DeepOWT, GOWT v1.3, Global_OWT, 1,193 independently interpreted reference turbines, and 21 validation cells. These files correspond only to the common evaluation area used for the cross-dataset comparison and are not intended to duplicate the full regional EA-OWT_2019 release. All spatial layers in this folder use WGS 84 (EPSG:4326).The cumulative annual turbine counts in EA-OWT are 1,714, 2,488, 5,039, 6,120, 6,981, 7,544, and 7,919 for 2019–2025, respectively. These values represent cumulative imagery-inferred turbine presence and should not be interpreted as independently verified annual commissioning totals.EA-OWT is intended to support regional analyses of offshore wind development, spatial distribution, temporal change, marine spatial planning, and related environmental assessments. Users conducting project-level analyses or requiring exact construction or commissioning dates are encouraged to cross-check the imagery-inferred records against authoritative project documentation.