Momentum, sensible heat and CO2 correlation coefficient: what can we learn from 20 years of eddy covariance measurements

Long-term data series of carbon dioxide and other gas exchanges between terrestrial ecosystems and atmosphere
become more and more numerous. Long-term analyses of such exchanges require a good understanding of
measurement conditions during the investigated period. Independently of climate drivers, measurements may
indeed be influenced by measurement conditions themselves subjected to long-term variability due to vegetation
growth or set-up changes.
The present research refers to the Vielsalm Terrestrial Observatory (VTO) an ICOS candidate site located
in a mixed forest (beech, silver fir, Douglas fir, Norway spruce) in the Belgian Ardenne. Fluxes of momentum,
carbon dioxide and sensible heat have been continuously measured there by eddy covariance for more than 20
years. During this period, changes in canopy height and measurement height occurred.
The correlation coefficients (for momemtum, sensible heat and CO2) and the normalized standard deviations
measured for the past 20 years at the Vielsalm Terrestrial Observatory (VTO) were analysed in order
to define how the fluxes, independently of climate conditions, were affected by the surrounding environment
evolution, including tree growth, forest thinning and tower height change.
A relationship between canopy aerodynamic distance and the momentum correlation coefficient was found
which is characteristic of the roughness sublayer, and suggests that momentum transport processes were affected
by z-d. In contrast, no relationship was found for sensible heat and CO2 correlation coefficients, suggesting
that the z-d variability observed did not affect their turbulent transport. There were strong differences in these
coefficients, however, between two wind sectors, characterized by contrasted stands (height differences, homogeneity)
and different hypotheses were raised to explain it. This study highlighted the importance of taking
the surrounding environment variability into account in order to ensure the spatio-temporal consistency of datasets.