Characteristics of the relative sampling error and its application to flux aggregation in eddy covariance measurements

Using eddy covariance (EC) measurements, exchanges of energy, water, and carbon dioxide between the atmosphere and terrestrial ecosystem have been studied worldwide. EC measurements fundamentally operate under the hypotheses of stationarity and ergodicity. However, these hypotheses are disturbed in the real world by various factors generated by varying atmospheric and land surface conditions. The relative sampling error-a parameter quantifying this disturbance estimated by the average convergence in a time series-from a tangerine orchard is investigated to determine its range in the real world, and its capabilities as a quality parameter for EC measurements. Our results showed that the relative sampling error ranges from 5% to infinity and displays a close relationship with the integral turbulence characteristic as a function of atmospheric stability. Therefore, we suggest that the relative sampling error represents a quality parameter for EC measurements and can be a weighting factor for statistical aggregations of exchanges observed by EC measurements.