This product provides monthly ρ CO2data with a spatial resolution of 0.25 ° from 1993 to 2021, which can be used to study regional spatial differences, seasonal cycles, and ten-year variations of ρ CO2. The reconstruction based on observation data is trained using surface ocean carbon dioxide atlas (SOCAT) observation data as observation values and various satellite derived and reanalyzed environmental variables known to control sea surface ρ CO2as model inputs. This product has demonstrated high accuracy in model training, validation, and independent testing stages, demonstrating its robustness and ability to accurately reconstruct ρ CO2in areas or periods lacking direct observational data. Compared with all observed samples of SOCAT, the determination coefficient of the ρ CO2product is 0.92, the root mean square error is 12.70 µ atm, and the cumulative uncertainty is 23.25 µ atm. The ReCAD-NAACOM pCO2 product demonstrated its ability to analyze the seasonal cycles, regional scale variations, and ten-year trends of NAACOM coastal ρ CO2. This new product provides reliable ρ CO2data for more accurate research on carbon dynamics along the North Atlantic coast.
SOCAT database (2023 edition): Measurement of fugacity of CO in seawater (fCO2).
1. Model design
2. Model training
3. Model evaluation
4. Reconstruct the uncertainty of ρ CO2
5. Comparison with the products of global reconstruction of ρ CO2
The uncertainty of the Atlantic edge region calculated by this dataset is comparable to the 43.4 µ atm reported by Sharp et al. (2022) within 100 kilometers of the North American Pacific edge, indicating good performance of the data product.
This work is licensed under a
Creative
Commons Attribution 4.0 International License.
CSTR:11738.11.NCDC.ZENODO.DB6760.2025
0.5281/zenodo.14038561