Uncertainty analysis of eddy flux measurements in typical ecosystems of ChinaFLUX

Fluxes of CO2 (FCO2) and energy (latent heat, LE; sensible heat, H) exchange between ecosystems and atmosphere, as measured by the eddy covariance technique, represent a fundamental data source for global-change research. However, little is known about the uncertainties of flux measurements at an ecosystem level in China. Here, we use data from six eddy covariance tower sites in ChinaFLUX, including two forested sites, three grassland sites, and one agricultural site, to conduct a cross-site analysis of random flux errors (RFEs) of FCO2, LE, and H. By using the daily-differencing approach, paired observations are obtained to characterize the random error in these measurements. Our results show that: (1) The RFEs of FCO2, LE, and H in different ecosystems of ChinaFLUX closely follow a double-exponential (Laplace) distribution, presumably due to a superposition of Gaussian distribution for high flux magnitude. (2) The RFEs of FCO2, LE, and H are not homogeneous and appear to be a linear function of flux magnitude. (3) Except for H, the RFEs of FCO2 and LE exhibit a distinct seasonal pattern. For FCO2, the dependence of RFEs on wind speed varies somewhat according to vegetation type, whereas for LE and H, there is no such dependence. The effect of temperature on RFEs is not statistically significant (P < 0.05). Both the distribution and the relationship of RFEs with flux magnitude in ChinaFLUX are essentially in accord with those in AmeriFlux and CarboEurope.