grant NER/S/J/2004/13118

Accurate observations of the carbon cycle are essential as inputs to global climate models. Observations made by the micrometeorological technique of eddy covariance, whist widespread, may be incorrect if air is advected away from below the sensor system. This is potentially a serious issue for FLUXNET, a global network of eddy flux sites. The approach in this thesis to investigate this problem was twofold: A full micrometeorological mass balance using an instrumented 50 m long by 50 m wide by 6 m high Cartesian control volume (CV) covering the understorey vegetation of a 40 m high Eucalyptus forest was carried out; situated adjacent to the Tumbaruma eddy covariance site in Australia. At night positive (into the atmosphere) advection fluxes caused by down-slope katabatic drainage within the forest trunk space, dominated the CO2 flux budget of the CV, with both vertical and horizontal advection terms having predominantly positive values. The nighttime estimates of advection were subject to large systematic errors that were of the same order of magnitude as the advection signal. Nevertheless, the nocturnal respiration flux of the understorey vegetation was clearly resolved by the diurnal full mass balance flux curve that resulted from the experiment, having a typical value of 5 μmol m-2 s-1. A second experiment carried out at the Griffin forest in Scotland demonstrated the presence of sub-canopy katabatic/gravity flows at night that would be likely to cause scalar advection resulting in underestimation of the nocturnal respiration flux of CO2. Finally, it is recommended that the micrometeorological mass balance technique should not be deployed across FLUXNET because of financial cost and issues of systematic error.