The present work studies the forest-atmosphere turbulent exchange, that is, the estimation of turbulent fluxes and the accuracy associated with their statistical calculation. This should provide important subsidies for a better understanding of aspects related to the accuracy of the investigation of the energy balance and carbon cycle in the Amazon. Furthermore, the calculation of flux error estimates offers subsidies for studies on the statistical characteristics of turbulent vortices, particularly the integral scale I, a parameter that indicates the ability of the signal to retain the "memory of turbulence". The data used resulted from collections provided by the LBA, MILÊNIO-LBA and PRONEX-SECTAM projects, which are from the meteorological logic tower installed in the Caxiuanã National Forest, and were treated according to the methodologies of Vickers and Mahrt (1997) and Wyngaard (1983). Preliminary results show that the forest behaves as a carbon sink and that there is great variability in the calculation of CO2 fluxes (from 20% to 500%) and that this is due, at least in part, to the occurrence of non-stationary conditions. The hours in the afternoon-night transition periods showed the least accurate fluxes. In these, the dry season data showed much higher error estimates for the calculation of fluxes than those observed in the wet season. Among the 460 cases already studied, it was not possible to calculate the error in 34% of them due to the difficulty of obtaining a value for I, because under very unsteady conditions the auto-correlation function for the sign of w'c' is not estimable, which reflects a dominant pattern of non-stationarity at certain times of the day. The incidence of large errors in the fluxes at the beginning of dry season nights coincides with the occurrence of internal cold air gusts, which, when penetrating the canopy, induce the rapid upward displacement of the air trapped below the canopy.
|
