VARIABILIDADE SAZONAL E INTERANUAL DA DIMENSÃO FRACTAL DE SÉRIES DE TEMPERATURA E UMIDADE RELATIVA DA AMAZÔNIA E PANTANAL

The rise in global average temperature, change in rainfall regime, the occurrence of more stringent droughts and floods, as well as the accelerated deforestation, all stress the need of describing and understanding the ecosystems. The time series analysis approach of the dynamic systems theory allows to characterize the nonlinear behavior of forest systems. The aim of this study was to analyze data from micrometeorological towers, measured above the forest canopy in experimental sites in the Amazon and the Brazilian Pantanal, searching for climatic attractors with low fractal dimensions of the temperature and relative humidity variables in the period 2005-2010. The annual and inter-annual sensitive variation of the estimates of fractal dimension, the exponent correlation dimension D^sub 2^, and the embedding dimension m strongly suggest the existence of periodic attractors and low dimension chaotic attractors governing the dynamics of these biomes. The exponent D^sub 2^ of the attractors varied in phase season- ally and was sensitive to forest fires.