A global dataset of high-resolution CO₂ enhancements derived from OCO-3 measurements

Abstract. We present a novel global dataset of CO₂ enhancements (ΔXCO₂) derived by fusing NASA’s OCO-3 satellite and NOAA ground-based observations. CO₂ enhancements quantify the spatially resolved excess in atmospheric CO₂ concentrations arising from anthropogenic emissions, biospheric CO₂ exchanges, and atmospheric CO₂ transport. Leveraging decades of monthly CO₂ measurements from eight remote stations strictly selected from NOAA ESRL network, such as the Mauna Loa station, we address the critical challenge of isolating localized CO₂ signals from background concentrations by developing a latitude-dependent global CO₂ baseline model that effectively captures spatial and seasonal variability in background CO₂. The developed baseline model demonstrates near-perfect hemispheric predictive accuracy (Northern: R²=0.988, RMSE=1.78 ppm; Southern: R²=0.995, RMSE=1.09 ppm). Spatially explicit ΔXCO₂ is then estimated by removing the column-corrected background CO₂ from co-located OCO-3 observations. Validations of the estimated ΔXCO₂ against tropospheric NO₂ (R²=0.896) and prior in-situ urban CO₂ measurements, along with the dataset's high spatiotemporal resolution (~ 3 km²), demonstrates its potential for tracking anthropogenic and biospheric CO₂ dynamics. Global ΔXCO₂ maps reveal mean CO₂ enhancements of 0.58 ± 1.81 ppm, with urban areas exhibiting 1.5-fold higher enhancements (1.43 ± 2.04 ppm). North Hemisphere land areas exhibits an approximately 81 % higher ΔXCO₂ average (0.67 ± 1.98 ppm) compared to the South Hemisphere (0.37 ± 1.32 ppm), with urban enhancements amplifying this hemispheric contrast up to 95 %. Comprising 54 million observations across more than 200 countries, this open-access dataset provides an alternative metric for monitoring complex atmospheric CO₂ variability and actionable insights for regional climate policies, available at https://doi.org/10.5281/zenodo.15209825.