Spatiotemporal evolution of CO2 concentration, temperature, and wind field during stable nights at the Norunda forest site

Unusually high CO 2 concentrations were frequently observed during stable nights in late summer 2006at the CarboEurope-Integrated Project (CEIP) forest site in Norunda, Sweden. Mean CO 2 concentrations inthe  layer  below  the  height  of  the  eddy-covariance  measurement  system  at  30 m  reached  up  to500 mmol mol 1 and large vertical and horizontal gradients occurred, leading to very large advectivefluxes with a high variability in size and direction. CO 2 accumulation was found to build up in the secondpart of the night, when the stratification in the canopy sub-layer turned from stable to neutral. Largestvertical gradients of temperature and CO 2 were shifted from close to the ground early in the night to thecrown space of the forest late at night, decoupling the canopy sub-layer from the surface roughness layer.At the top of the canopy at 25 m CO 2 concentrations up to 480 mmol mol 1 were observed at all fourtower locations of the 3D cube setup and concentrations were still high (<400 mmol mol 1 ) at the 100 mlevel of the Central tower. The vertical profiles of horizontal advective fluxes during the nights underinvestigation were similar and showed largest negative horizontal advection (equivalent to an additionalCO 2 -sink) to occur in the crown space of the forest, and not, as usually expected, close to the ground. Themagnitude of these fluxes was sometimes larger than 50 mmol m 2 s 1 and they were caused by thelarge horizontal CO 2 concentration gradients with maximum values of up to 1 mmol mol 1 m 1 . As aresult of these high within canopy CO 2 concentrations, the vertical advection also became large withfrequent changes of direction according to the sign of the mean vertical wind component, which showedvery  small  values  scattering  around  zero.  Inaccuracy of the sonic anemometer at such low wind velocities is the reason for uncertainty in vertical advection, whereas for horizontal advection, instrument errors were small compared to the fluxes. The advective fluxes during these nights were unusually high and it is not clear what they represent in relation to the biotic fluxes. Advection is most likely a scale overlapping process. With a control volume of about 100 m × 100 m × 30 m and the applied spatial resolution of the sensors, we obviously miss relevant information from processes in the mesoscale as well as in the turbulent scale.