Reducing the effects of weather on the sampling bias in tall tower eddy covariance flux measurements

Tall tower eddy covariance (TTEC) flux measurements are employed to estimate turbulent matter and energy fluxes at landscape scales (e.g. within 1 to 5 km radius around a tower). Virtually all landscapes feature horizontal surface heterogeneity. One main complication for the interpretation of TTEC is the sampling bias by the varying local meteorological conditions. While the wind direction bias can only be considered by the choice of the location of the TTEC, we examine here how the effects of atmospheric stability can be alleviated by sampling from different measurement heights (z_m). The objective is to define an optimal set of measurement heights to minimize sampling bias from variation in atmospheric stratification for TTEC long-term flux observation. To our knowledge, this problem has not yet been addressed in the scientific literature.

We used a two years’ dataset from the 122 m tall tower at Risø (Denmark, 55°41'39.15" N, 12°5'17.93" E) and two flux footprint models to develop an objective statistical approach for the definition of a set of measurement heights for optimal sampling of the landscape heterogeneity. The tower is equipped with 3D ultrasonic anemometers in five different heights. The evaluations concern the Eastern sector, which is comprised of a mosaic of land uses, small settlements and comparably sparse road infrastructures.

We define the criteria for optimal landscape flux sampling from the distributions of the source weights (contribution to the measured flux per unit area) sampled in a number of stability classes relative to a frequent unstable stability class as reference. The upper sampling height is set a priory to match measurements and the targeted area; here z_m equal to 120 m for the 70% cumulated footprint to stay within a 5 km radius around the tower.

Theoretical analysis with the footprint model shows the limitation of the attempt to compensate for lateral footprint extension at different stabilities, while the longitudinal sampling of the landscape heterogeneity can be maintained more homogeneously by the systematic choice of the measurement height according to atmospheric stability and wind speed.

The results rely on the accuracy of the footprint estimation, which is generally an essential criterion for the interpretation of TTEC measurements in heterogeneous landscapes.