Fluxo de calor latente em uma floresta tropical da Amazônia : uma análise de séries temporais com wavelets e do produto MOD16

Abstract

The land use changes occurred in recent decades, as well as the magnitude of the surface fluxes in the Amazon region, are preponderant elements for discussion regarding the atmosphere - biosphere interaction in this ecosystem. The present study aimed (i) to characterize the microclimate and (ii) apply temporal and spatial analysis tools to help in the understanding of latent heat flux (LE). For this, were used data from 2004 to 2010 of net radiation (Rn), air temperature (T), air relative humidity (RH), wind speed (u) and sensible (H) and latent (LE) heat fluxes measured in a micrometeorological tower in the Jaru Biological Reserve (REBIO Jaru), in southwestern of Amazonia. For the methodological approach were also used data products from the Moderate Resolution Imaging Spectroradiometer (MODIS), with focus on the algorithm for global evapotranspiration estimation (MOD16). The wavelet of Morlet transform was used for a time series analysis, in hourly and daily scales. With this study it was possible to identify the existence of well-defined seasonal patterns, with differentiation between the seasons and variations in the microclimate over the years, with an increase of the variables T and H and decreased of RH and LE. The wavelet transform allowed to identify the dominant periods as well as the fluctuations of the energy levels occurring in LE, Rn, T and RH, enabling the detection of variances consistent with the behavior of microclimate data, which demonstrated the applicability of this method to detecting patterns difficult to identify. With respect to LE estimated by MOD16, similarity was found between the annual averages of the model and the measured data. However, the variations over shorter periods were not well represented, except by the monthly averages of the dry season. One factor that may cause the differences mentioned is the fact that the LE MOD16 product, using eight days averages, do not capture the effects of fluctuations that occurs in shorter periods (12 and 24h, 2 and 4 days), previously identified in the wavelet analysis.