Annual and interannual variability of NDVI in Brazil and its connections with climate

This study deals with the connections between vegetation cover and the annual and interannual climatic variability in Brazil. The analyses were based on the normalized difference vegetation index (NDVI) derived from the Advanced Very High Resolution Radiometer (AVHRR) sensor for the Global Inventory Modeling and Monitoring Study (GIMMS) continental dataset and monthly mean outgoing longwave radiation (OLR) data. The statistical technique of principal component analysis (PCA) was used for analysing the data for the January 1982-December 1993 period. The first component reveals that NDVI data have potential for characterizing the main patterns of vegetation types in Brazil. It was also possible to characterize the annual and interannual climate-related variability of the vegetation types. Specifically, the second and third leading PCA are associated with the summer/winter and spring/autumn portions of the annual cycle while their interannual variations are modulated by El Nino Southern Oscillation (ENSO) phenomena. The fourth component showed that 'non-climatic' factors, such as slash burning events, do affect NDVI values mainly in central Brazil. The high correlation coefficients between NDVI and OLR eigenvectors, corresponding to the summer/winter ( r =0.91) and spring/autumn ( r =0.70) modes, suggest that the annual variability of the vegetation coverage in Brazilian territory is largely modulated by the prevailing climatic conditions.