Influência da vegetação na previsão do tempo utilizando o modelo BRAMS

Abstract

Considering the surface-atmosphere interactions exerted on vegetation and consequently on the climate, is that it becomes important in numerical prediction, the surface data update. The growing concern about the predictability of the models, combined with the research conducted in this field, is which enables the advancement in weather forecasting. Therefore, this study aimed to analyze the change in vegetation through of different Indexes of Normalized Difference Vegetation (NDVI), on the Brazilian Regional Atmospheric Model System Modelling (BRAMS) for the Northeast Region of Brazil, focusing on the Alagoas State, which analyzed the following variables: rainfall, Sensitive and Latent Heat Flux, Mixing Ratio, Potential Temperature, longwave and shortwave radiation incident, for all biomes detected by the model, which were: Open Evergreen Forest, Deciduous broadleaf tree and Savanna, Semi-desert I (Savannah Park) and II (Savannah Grassy-Woody) and Shrubby Grassland. The selected period were the years 2008 and 2009, in the month of June, which was chosen based on the analysis of spectral-temporal profile of NDVI data observed for the two years selected. The analysis considered the behavior in coordinates referring to the capitals of the Northeast Region, and in the case of Alagoas State, 10 cities distributed, which had the highest rates in the behavior for both years in both grids. There were two sets of simulation, the denominated control (using MODIS NDVI climatological) and the modified (using SPOT NDVI). We also used two grids, the first covers the Northeast Region with 20km spatial resolution (G1), and the second is a "nesting" of the first and comprises the State of Alagoas, with 5km spatial resolution (G2). For assimilation in the model, vegetation data observed were segmented. The results showed that although the control simulation have represented well the variables analyzed, was the modified simulation to that performed better behavior in rainfall, comparing their results with GPCC (G1) and automatic stations of INMET (G2). Regarding biomes, Shrubby Grassland, showed the greatest change between simulations and Savanna biomes I and II, showed the highest congruence. On the differences between simulations, the variable Shortwave radiation, showed the least change, and in all other variables presented substantive changes.