Analysis of ALOS/PALSAR polarimetric signatures and scattering mechanisms of forest types in Tapajós region, Brazil.
João Roberto dos Santos(1), Igor Silva Narvaes(2), Paulo Mauricio Lima Alencastro Graça(3) and Fábio Guimarães Gonçalves(4)
(1) National Institute for Space Research - INPE, Av. dos Astronautas, 1758, 12227-010 São José dos Campos - SP., Brazil
(2) National Institute for Space Research - INPE, Av. dos Astronautas, 1758, 12227-010 São José dos Campos - SP., Brazil
(3) National Institute of Amazonian Research - INPA, Av. André Araújo, 2936, 69060-001 Manaus, AM., Brazil
(4) Oregon State University, 321 Richardson Hall, Corvallis, OR 97331, United States
The objective of this work is to analyze the graphic representation of polarimetric signatures of ALOS/PALSAR data (L-band) in primary forest, secondary succession, forests with timber exploitation and those under recovery after a surface burning process. Additionally another objective is an exploratory analysis of scattering mechanisms of forest typology in accordance with the alternative procedure of SAR image classification by target decomposition. The area under study is located in the region of Tapajós, Brazil, located at geographical coordinates S 3º 01’ 59.85” to 3º 10’ 39.33” and W 54º 59’ 53.08” to 54º 52’ 44.96”. The PALSAR data were collected in 23 April 2007, at HH, VV and HV polarizations, with a spatial resolution of 4.50 m and 9.50m in range and azimuth, respectively, and incidence angle of 24.233º. At the representation of polarimetric signatures, the cross-section of the forest target was plotted on a three-dimensional graphic as a function of the combination from the orientation angle, ellipticity angle, referring to the ellipse of polarization. The resulting surfaces were compared visually and the diversity of configuration for each plot was analyzed supported by physiognomic-structural information collected during the field survey. Besides that the distinction between simple and multiple scattering due to the decomposition of targets was formed by the association of entropy values and the average angle for each sample area derived from eigenvalues and eigenvectors of the coherence matrix from polarimetric data. After the samples were introduced to the bi-dimensional (H, Alpha) classification space, statistical tests were used to verify significant differences at the classification by the decomposition of targets, using linear regression fitting models, in a comparison between samples. Some results can be mentioned: (a) at types of landscapes studied there is a predominance of scattering processes with low and medium entropy; (b) for all thematic classes studied the pixel distribution in the [H, Alpha] bi-dimensional space was more frequent at zones 4, 5 and 9. This study improves the understanding of the interaction mechanisms between L-band PALSAR signals and structural parameters of tropical rainforest, subsidizing the inventory and the monitoring of land cover in the Amazon region.