Abstract: The Amazon region is a very important place from meteorological and ecological point of view, with typical features, which are needed to be taken in account in its modeling. Therefore, the study of a good representation of the atmospheric boundary layer (ABL) evolution is fundamental for a good performance of numerical meteorological models in this region. The structure of ABL is linked with the representation of turbulent fluxes. The present work shows a comparison among many turbulent models: Smagorinsky, Mellor-Yamada and some first-order closure schemes. The first-order schemes are: Troen-Mahrt, Holstag-Moeng, Holstag-Boville and a parameterization based on Taylor's statistical theory of turbulence. A counter-gradient term is also used in the first-order closures, including a new formulation of this term. All these first-order approaches were recently introduced in the RAMS-CSU code, the mesoscale meteorological model employed. The experiments were perfomed over pasture simulating the whole diurnal cycle. The simulated experiments are compared with WET-AMC observations, where the following parameters are checked: potential temperature, vapor mixing ratio, relative humidity.