Quality Control in the AmeriFlux Network: Filling the Gaps

Quality control methods used for eddy covariance flux tower measurements will result in data record gaps. Data gaps in flux measurements are on average 65% (Falge et al. 2001). Reasons for filtered observations in flux-measurements include friction velocity (u*) filtering due to low turbulent flow, precipitation impact to open-path gas analyzers and sonic anemometers , spikes due to instrument errors and electronic noise, lack of turbulence stationarity, sensor malfunctions, or loss of power (Mauder et al. 2013). Long-term flux measurements, such as those taken across the FLUXNET networks, require significant maintenance to maintain research quality observations. The AMERIFLUX network subjects all data to additional quality control flagging, which results in additional gaps in data files. These gaps can pose difficulties to researchers interested in long-term questions pertaining to specific ecosystems and gap-filling methods are numerous, but subject to uncertainty and assumptions. To assess the quality and usability of individual sites, statistical tests were used to estimate missing data characteristics of half-hourly eddy fluxes across the AMERIFLUX network. Missing data statistics were stratified by land cover and land management to assess reliability of observational network as a function of ecosystem characteristics. The results help provide the researchers an estimate of overall data quality due to gaps in flux measurements, which is necessary prior to analysis of fluxes across sites and years and for estimate of uncertainty for model-data comparison.