Seasonal Bias in OCO-2 XCO2 Satellite Observations

Accurately monitoring atmospheric carbon dioxide (CO₂) is vital for understanding global carbon fluxes and shaping climate mitigation policies. This study explores the seasonal variability of biases between satellite-derived OCO-2 XCO₂ observations and ground-based TCCON XCO₂ measurements at the Caltech TCCON station over nine years (2014–2023). The study categorized the data by observation mode (nadir or glint) and month and investigated the distributions of deviations from the mean bias.Distinct seasonal patterns emerged in the bias variability. Nadir mode observations demonstrated consistent median deviations, ranging from -0.6 to 0.4 ppm, indicating minimal bias variability. In contrast, glint mode observations showed substantial variability, with absolute median deviations surpassing 1 ppm during January, March, and September. Skewness analysis revealed asymmetries in the data distributions and the presence of significant outliers. A strong correlation was observed between monthly Normalized Difference Vegetation Index (NDVI) values and glint mode skewness (R² = 0.76), highlighting its sensitivity to surface reflectance and vegetation dynamics. In comparison, nadir mode skewness demonstrated greater stability with minimal correlation to NDVI.The study underscores the need to consider environmental factors, such as vegetation coverage and observation mode differences when interpreting OCO-2 data. By identifying the role of seasonal variability in satellite-ground measurement discrepancies, these findings contribute to refining retrieval algorithms and enhancing satellite-based XCO₂ monitoring accuracy. Improved accuracy supports the development of more reliable carbon flux models, which are essential for effective climate policy and mitigation strategies. Future studies should replicate this analysis at other TCCON stations and incorporate additional environmental variables to further elucidate the drivers of seasonal biases in OCO-2 observations.