The role of sampling in the disparity between CO2 fluxes inferred from GOSAT vs. surface measurements

Abstract:

Several recent studies using GOSAT total column CO₂ measurements have reported a systematic redistribution of carbon fluxes compared to inversions considering only surface-based measurements. Significantly higher emissions are seen from the Tropics, particularly from tropical Africa, and the sink inferred for Europe is consistently larger than what the surface-based measurements suggest, and difficult to reconcile with bottom-up estimates. Controversy has surrounded this issue, with some suggesting that the flux patterns are, in fact, real, while others have argued that it is a result of undiagnosed biases in the measurement or long-range transport. This study presents evidence that the redistribution of fluxes may be explained by the increased spatial density of the satellite measurements, compared to the relatively sparse surface network. Inversions are performed with subsets of the satellite measurements within a given distance of the existing surface sites. As that distance is decreased, the retrieved fluxes converge towards that of the surface-only result. This suggests that the spatial pattern of fluxes inferred from the surface sites is, in fact, biased. This hypothesis is tested with a variety of pseudo-data experiments, to help explain why this might be. Chemical and transport processes leading to this discrepancy are discussed.