Quantifying and reducing the differences in forest CO2-fluxes estimated by eddy covariance, biometric and chamber methods: A global synthesis

- Global synthesis shows NEP overestimate by EC mainly from Re underestimate.
- NEP estimates are affected by site terrain, forest canopy and methods.
- Cross-validating CO2-fluxes by various methods improves estimation of C budgets.

Carbon dioxide (CO2) fluxes between terrestrial ecosystems and the atmosphere are primarily measured with eddy covariance (EC), biometric, and chamber methods However, it is unclear why the estimates of the CO2-fluxes, when measured using different methods, converge at some sites but diverge at others. We synthesized a novel global dataset of forest CO2-fluxes to evaluate the consistency between EC and biometric or chamber methods for quantifying the CO2 budget in forest ecosystems. The EC approach, compared with the other two methods, overestimated net ecosystem production (NEP) by 25% (0.52 Mg C ha−1 yr−1), and underestimated ecosystem respiration (Re) by 10% (1.39 Mg C ha−1 yr−1) and gross primary production by 3% (0.48 Mg C ha−1 yr−1). The differences between EC and the other methods were greater at the sites with complex topography and dense canopy than at the sites with flat topography and open canopy. Forest age also influenced the differences mainly through changes in leaf area index. Open-path EC system induced large positive bias in the NEP estimated by EC, presumably due to its surface-heating effect. These results suggest that EC method likely produce biased estimates of NEP and Re in forest ecosystems. A global extrapolation suggests that the differences in the forest CO2-fluxes measured with different methods be consistent with the global overestimation of NEP and underestimation of Re by EC method. Accounting for these differences would substantially improve our estimates of the forest carbon budget. The uncertainties involved in each method were also discussed. To reduce uncertainty in quantifying both local and global carbon budgets, we recommend cross-validation of forest CO2-fluxes measured by different methods with more accurate measurements and careful data processing strategies.