Master's Thesis Abstract by Stephanie Ann Bohlman (1995)

Satellite images, which integrate individual tree canopies and cover a large spatial extent, can provide information on stand-level canopy phenology. In tropical forests, these seasonal changes are especially difficult to determine from ground studies because of the high species diversity and low phenological synchrony. The variability in the near infrared bands (bands 4 and 5) of several distinct vegetation types was used to detect seasonal changes in a series of three Landsat TM images from the wet season to the dry season in Marabá, Brazil (eastern Amazon basin). Despite different atmospheric and instrumental conditions among the three images, changes in characteristics of the forest types were clearly distinguishable. Changes in the spectral properties of the vegetation suggest: (1) upland terra firme forest has younger mean leaf age as a result of increased leaf exchange in the dry season, (2) lowland liana forest has younger mean leaf age that may correspond with liana leaf flush in the west season and (3) other riparian vegetation shows an increased amount of exposed wood due to deciduousness in the dry season. Chronological changes, indicating increased leaf biomass over a period of 14 months, are also apparent in terra firme forest that had suffered fire penetration from the burning of adjacent pasture. This study shows that time series of high-resolution images can provide detailed information on temporal changes in primary vegetation and guide field studies to investigate seasonal changes that may not be detectable from the ground.