Fraction images for monitoring intra-annual phenology of different vegetation physiognomies in Amazonia

Taylor & Francis
Publication Type:
Journal Article
International journal of remote sensing, 2011, 32 (2), pp. 387 - 408
Issue Date:
Full metadata record
Files in This Item:
Filename Description Size
Thumbnail2010000710OK.pdf1.27 MB
Adobe PDF
In this study we investigate the potential of fraction images derived from a linear spectral mixture model to detect vegetation phenology in Amazonia, and evaluate their relationships with the Moderate Resolution Imaging Spectroradiometer (MODIS) vegetation indices. Time series of MODIS 250-m data over three contrasting land cover types in the Amazon were used in conjunction with rainfall data, a land cover map and a forest inventory survey to support the interpretation of our findings. Each vegetation physiognomy was characterized by a particular intra-annual variability detected by a combination of the fraction images. Both vegetation and shade fractions were important to evaluate the seasonality of the open tropical forest (OTF). The association of these results with forest inventory data and the literature suggests that Enhanced Vegetation Index (EVI) and vegetation fraction images are sensitive to structural changes in the canopy of OTF. In cerrado grassland (CG) the phenology was better characterized by combined soil and vegetation fractions. Soybean (SB) areas were characterized by the highest ranges in the vegetation and soil fraction images. Vegetation fraction and vegetation indices for the OTF showed a significant positive relationship with EVI but not with Normalized Difference Vegetation Index (NDVI). Significant relationships for vegetation fraction and vegetation indices were also found for the CG and soybean areas. In contrast to vegetation index approaches to monitoring phenology, fraction images provide additional information that allows a more comprehensive exploration of the spectral and structural changes in vegetation formations.
Please use this identifier to cite or link to this item: