This paper deals with vegetation cover variability in Brazil using cluster analysis. The study was done using NDVI (Normalized Difference Vegetation Index) images from the AVHRR (Advanced Very High Resolution Radiometer) sensors for the January1982-December1993 period. The results show that the annual cycle of NDVI in the Amazon region is not well defined; the maximum values typically occur in June, two months after the rainy season, while the minimum ones occur in two distinct periods: February-March and September-November. In Central Brazil, the Savannas has a well defined annual cycle, showing maximum NDVI values around March and May and a minimum in September. On the other hand, the seasonal variability of the Northeast Brazil (NE) "Zona da Mata" (Atlantic Forest) and Savannas of Roraima vegetation cover show high NDVI values in June and July and low values between February and March, a few months before the rainy season onset. In the case of NE, the "Caatinga" (thorn shrub) shows a well defined annual cycle with a remarkable dry period, the highest NDVI values occur between April and May, which is the end of the rainy season, and the smallest values occur in September and October. In portions of Santa Catarina and southern part of Parana State, the annual cycle of the prevailing vegetation cover (open ombrophylous forest and mixed ombrophylous forest) is not well defined, while in Southern Brazil, the Steppe region does show a seasonal variability, with maximum NDVI values between March and June and a minimum one in August. Also, it was observed that ENSO events, independent of their intensity, do affect the different types of vegetation cover mainly the dense and greener forest types (e.g. the Amazon forest).
vegetation cover; El Niño; climate
