Characterization of vegetation physiognomies in earstern Amazonon region through airbone videography and landsat 7 ETM + images / Caracterização de fisionomias vegetais na Amazônia Oriental através de videografia aerotransportada e imagens landsat 7 ETM+

AUTOR(ES)
DATA DE PUBLICAÇÃO

2003

RESUMO

Transformation of multispectral remote sensing data is needed for compression of data volume and for the inference of biophysical properties of the land cover. Linear Spectral Mixing Models (LSMM) and Normalized Difference Vegetation Index (NDVI) are transformations that result in indicators of vegetation structure. However these methods present limitations in relation of their sensitivity. There are few evaluations on the real meaning of the products derived with these techniques. Airborne videography allows a thorough evaluation of these transformations, for it allows a better understanding of the vegetation structure and of how it affects the scattering mechanism of the radiation detected by orbital sensors. The objective of this work is to evaluate the spectral response of several vegetation physiognomies in the municipality of Maraba, Para, in relation to the reflectance in Landsat -7 ETM+ and in their transformations and to compare the proportion of the Shadow component derived through the LSMM with the Shadow percent measured in the Videographic products. Based on the videography, five types of vegetation physiognomies were identified: Upland Forest, Floodplain Forest, Secondary forest, Babacu Forest and Grassland. Analysis of the digital number distribution of each physiognomy in each band was conducted and of their relations with NDVI and with the components Soil, Vegetation and Shadow extracted from the Landsat Bands through LSMM. It was also analyzed the relationship between the Shadow component and the percentage of shadow determined through the analysis of videographic data. The results demonstrated that the spectral signature of each physiognomy is associated to the amount of vegetation cover, canopy architecture, and dominant background and to the leaf angular distribution. The IDVN discriminated only Grassland, Secondary Forest and the remaining classes. The percentage of shadow in the videographic data allowed only the differentiation of Babacu Forest from the remaining classes. The Shadow component of the linear mixing model was the most efficient product for the discrimination of the vegetal physiognomies, allowing the discrimination among all the classes but between Grassland and Secondary forest. This result reassures the potential of LSMM to the study of natural vegetation cover in Tropical Forest Regions.

ASSUNTO(S)

landsat 7 video equipment comportamento espectral índice de vegetação por diferença normalizada (ndvi) landsat 7 spectral signatures índice de vegetação normalized difference vegetation index (ndvi) videografia mixture model sensoriamento remoto remote sensing vegetative index modelo linear de mistura espectral (mlme)

ACESSO AO ARTIGO

Documentos Relacionados