6533b872fe1ef96bd12d4100

RESEARCH PRODUCT

Sorption-Caused Attenuation and Delay of Water Vapor Signals in Eddy-Covariance Sampling Tubes and Filters

Pekka KekäläinenJussi TimonenErkki SiivolaTimo VesalaIvan MammarellaAnnika Nordbo

subject

Atmospheric ScienceMaterials science010504 meteorology & atmospheric sciencesMeteorologyAttenuationEddy covarianceHumidityOcean EngineeringSorption04 agricultural and veterinary sciencesMechanics01 natural sciencesVolumetric flow rateFlux (metallurgy)Desorption040103 agronomy & agriculture0401 agriculture forestry and fisheriesWater vapor0105 earth and related environmental sciences

description

AbstractAdsorption and desorption (together called sorption) processes in sampling tubes and filters of eddy-covariance stations cause attenuation and delay of water vapor signals, leading to an underestimation of water vapor fluxes by tens of percent. The aim of this work was (i) to quantify the effects on sorption in filters and tubes of humidity, flow rate, and dirtiness and (ii) to test a recently introduced sorption model that facilitates correction of fluxes. Laboratory measurements on the transport of water vapor pulses through tubes and filters were carried out, and eddy-covariance field measurements were also used.In the laboratory measurements, the effects of sorption processes were evident, and filters caused a similar attenuation and delay of the signal as tubes. Filters could have a larger impact than a long tube, whereas the flow rate had a much smaller impact on the flux loss than the sorption processes (Reynolds numbers 2120–3360). The sorption model represented well the water vapor pulses in a wide range of conditions. As for the field measurements, the transfer function (TF) derived from the sorption model represented well the observations. Fitting parameters were found to depend strongly on the relative humidity and correlate with the signal delay. Having a more complex shape, TF of the sorption model represented much better the measured TFs than, for example, a Lorentzian or adjusted Gaussian TF, leading on average to a 4% unit difference in the flux corrections. Use of this more complex TF is recommended and its implementation is assisted by the codes provided in appendix B.

https://doi.org/10.1175/jtech-d-14-00056.1