Scaling nighttime turbulence intensity for correcting carbon dioxide fluxes
Acevedo, UFSM, email@example.com
Moraes, UFSM, firstname.lastname@example.org
Campos, UFSM, email@example.com
Fitzjarrald, SUNY, firstname.lastname@example.org
Sakai, SUNY, email@example.com
The determination of nocturnal ecosystem respiration rates using the eddy covariance technique constitutes a major challenge for both micrometeorological and ecological communities. The difficulty in properly quantifying the total carbon dioxide transfer becomes critical when there is not enough turbulent mixing. For this reason, gap-filling methods have been proposed and are extensively used for the determination of ecosystem carbon budgets. These methods typically consist in replacing the data from nights with low turbulence intensity for those originated in nights with similar soil characteristics, but enough mixing. The turbulence scale used for this classification is the friction velocity, u*.
In a recent study, it was shown that there is organized turbulent mixing in very stable conditions, which can be properly determined if the appropriate averaging windows are used. These windows can be as small as 5 or 10 seconds in the most stable conditions. For larger time scales, the exchange has a more variable character, being generically associated to “mesoscale processes”.
In the present study we call the attention to the fact that the most common scale used to classify nighttime turbulent fluxes, the friction velocity is, indeed, a flux and, therefore is subject to large variability in the mesoscale range. As a consequence, nights with little turbulent exchange can be frequently classified in the traditional gap-filling schemes as turbulent, if the mesoscale transport is large enough. On the other hand, turbulent nights can be classified as non-turbulent if the mesoscale momentum transfer is positive. As a result, the classification of turbulent fluxes as a function of u* is smoothed: the exchange is overestimated for low turbulence and underestimated for cases with enough mixing. As an alternative, we propose the usage of the standard deviation of the vertical velocity fluctuations for this classification. It is shown that, not constituting a flux, this scale is not subject to the problems associated to the friction velocity.