THE LONG-TERM MONITORING OF VEGETATION COVER IN THE AMAZONIAN REGION OF NORTHERN BRAZIL USING NOAA-AVHRR DATA

We have analysed monthly composites of normalized difference vegetatio n index (NDVI) calculated from NOAA's Advanced Very High Resolution Ra diometer (AVHRR) for the Amazonian region of northern Brazil across a decade (August 1981 to June 1991) to ascertain if the dominant vegetat ion types could be differentiated, and to seek inter-annual climatic v ariation due to changing environmental conditions. The vegetation type s observed included dense forest (submontana and terras baixas), open forest (submontana and terras baixas), transitional forest, seasonal f orest (caatinga), and two types of savanna (cerrado). We found that mo nthly NDVI composites revealed seasonality in cerrado and especially i n caatinga cover types, which can be used in their identification, whi lst the phenology of other forest cover types varies little throughout the year. Additionally, yearly composite NDVI values showed a clear a nd significant reduction (p>0.95) in dry years, such as those with El Nino Southern Oscillation events. These results indicate the potential use of multi-temporal NDVI data for the environmental characterizatio n and identification of forest ecosystems. Our research found NDVI ima ges from NOAA AVHRR offer a long-term data set that is unequalled for monitoring terrestrial land cover. However, these data have to be used with a degree of caution, especially in regards to atmospheric interf erence, such as cloud contamination and volcanic eruptions, and post-l aunch changes in calibration.