Assessing the method-specific differences in quantification of CO2 advection at three forest sites during the ADVEX campaign

The newmethod for CO2 advective flux computation, based on the air mass-conservation principle, MCA (Montagnani et al., 2009) is applied to datasets collected at the three forest sites of Renon, Wetzstein and Norunda during the ADVEX campaign. Values of advective flux, calculated for 1 month at each site, are compared to those obtained using the more common method which computes the advective fluxes along vertical and horizontal CO2 gradients, GA (Feigenwinter et al., 2008). According to both methods, night-time CO2 advection values were found to be positive at the sloping sites of Renon (MCA, 8.88 mmol m2 s1, GA, 14.30 mmol m2 s1) and Wetzstein (MCA, 2.82 mmol m2 s1, GA, 3.07 mmol m2 s1) and negative at the flat site of Norunda (MCA, 3.00 mmol m2 s1, GA, 8.12, mmol m2 s1), where the occurrence of extremely high negative advection values was calculated at night according to both methods. Daytime advection was found to be generally small and negative at all sites following both methods, while standard deviations were found to be generally higher according to the GA method. Half-hourly calculated values were found to be similar during some periods, while in others, characterized by specific wind conditions, substantial differences were present. The coefficient of correlation (r2) between the two estimates was 0.15 for Renon, 0.55 for Wetzstein and 0.45 for Norunda. Three methodological aspects were considered to identify the reasons for the observed differences in CO2 advections estimates: the correction factor used to attain mass conservation, the air incompressibility assumption and the vertical interpolation of wind velocities were found all to be scarcely correlated to observed differences. These results indicate that general information concerning sign and daily courses of CO2 advection estimates can already be taken from direct measurements, but there are still unresolved theoretical and computational issues affecting their quantitative reliability. 2010 Elsevier B.V. All rights reserved.