Some methodological questions concerning advection measurements: a case study

A dataset from two campaigns conducted at the Vielsalm experimental site in Belgium was used as a basis for discussing some methodological problems and providing intermediate results on estimating CO2 advection. The analysis focused on the horizontal [ CO2] gradient and on the vertical velocity w, the variables most affected by uncertainty. The sampling error for half-hourly horizontal [CO2] gradients was estimated to be 1.3 mu mol mol(-1). Despite this important random error for half-hour estimations of [CO2], the mean horizontal [CO2] gradients in advective conditions were shown to be representative at the ecosystem scale and to extend only to the lowest part of a drainage sub- layer, which developed in the trunk space. By contrast, under daytime conditions, this gradient was shown to be more sensitive to local source heterogeneities. The estimation of the short- term averaged vertical velocity ((w) over bar) was the greater source of error when computing advection terms. The traditional correction methods used to obtain (w) over bar are discussed and a (co)sine correction is tested to highlight the instrumental origin of the offset in w. A comparison of measurements by sonic anemometers placed close together above the canopy showed that the uncertainty on (w) over bar was 0.042 m s(-1), which is of the same order of magnitude as the velocity itself. In addition, as the drainage sub- layer is limited to the lowest part of the canopy, the representativeness of (w) over bar is questionable. An alternative computation using the divergence of the horizontal wind speed in the trunk space produced a (w) over bar estimation that was four times lower than the single-point measurement. However, this value gives a more realistic estimate of the vertical advection term and improves the CO2 budget closure at the site.