207 Rain Gauges (RG) of the Brazilian National Agency for Water (ANA) were analyzed using statistical non-parametrical tests. The Pettitt’s test identified ruptures in the chronological rainfall series, while the Mann-Kendall’s test detected annual and seasonal tendencies in rainfall indexes and a linear regression analysis identified slight gain or loss in precipitation. Pettitt’s test indicated 16% of ruptures in the chronological rainfall series at the same time as Mann-Kendall’s monthly test put in evidence 41% of the RG having negative trends in transition seasons (onset and offset of the rainy season). Lastly the linear regression analysis showed 63% of data having negative trends. Additionally the dates of onset and offset of the rainy season were identified and its results submitted to Mann-Kendall’s and the linear regression approach. The data suggests strong contrasts between the Southern Amazon and the Northern Cerrado showing a delay on the onset of the rainy season for 84% of the RG, a premature offset for 76% and a reduction in the rainfall seasonal extend for 88%. An exponential ordinary kriging analysis of RG in deforested areas also revealed major chances of deforestation areas working as an adjuvant in the weakening of the rainy season- especially in highly deforested areas of the Mato Grosso State and the northern Rondônia. Aiming to build a tool to detect interactions between land surface and rainfall patterns the207 RG were correlated through a buffer zones analysis with land use data acquired from satellite LANDSAT 5. The time frame previously selected was divided into three periods of forest cover (before 1997, between 1997-2010 and acumulated for 2010). The cross-related buffer zones analysis (1-50km) indicated at local level that precipitation patterns are not well correlated to forest cover. Yet the buffer zones methodology suggested that as larger the forest areas are, larger are the probabilities of those influencing precipitation at regional scale, contrary to forest fragments in local level. Despite the climatic data in the buffer analyzes do not reveal significant correlation to forest cover, the statistic Pettit and Mann-Kendall tests, the linear regression analyzes and the identification of the rainy season, confirmed a fine linkage with recent findings which focus large-scale circulation models including forest cover as a variable
