Methods and uncertainties in the experimental assessment of horizontal advection

Different configurations of the measurement system were used to quantify the uncertainty of horizontal concentration gradients and to disentangle its sources. The discrete temporal sampling was the major source of uncertainty, accounting for 54% of the total uncertainty, while spatial sampling accounts for 39% of the total uncertainty in stable conditions and for 35% in unstable conditions, with the remaining uncertainty being explained by the accuracy of the instrumental set-up (analyser, pumps and valves) (7% and 11% for stable and unstable atmospheric conditions, respectively, corresponding to ∼0.2 ppm). Finally, we investigated the effect of buffer volumes on the uncertainty generated by a discrete temporal sampling in the estimation of horizontal advection. The use of volumes with a mean residence time equal to the turn-over time of the manifold reduced by half the standard deviation in the time series of horizontal advection (from 3.8 to 1.8 μmol m−2 s−1). The measurement system in its final configuration is currently used to quantify storage and advective fluxes at the site.