Análise do potencial de imagem TerraSAR-X para mapeamento temático no sudoeste da Amazônia brasileira

Abstract

ResumoO presente trabalho tem como objetivo analisar o potencial de imagens SAR polarimétricas do sensor TerraSAR-X, no modo StripMap, para mapear o uso e cobertura da terra na região sudoeste da Amazônia brasileira. No procedimento metodológico imagens de amplitude nas polarizações A HH e A VV , A <HH.VV*> derivada da matriz de covariância, bem como da entropia A Entropia derivada da decomposição de alvos por auto-valores fizeram parte, de forma individual ou combinada, do conjunto de dados investigados. Na classificação das imagens foram empregados dois classificadores: um baseado nas funções estatísticas de máxima verossimilhança (MAXVER); e outro, o método contextual (Context). Os resultados temáticos dessas classificações foram avaliados através da matriz de confusão e pelo índice Kappa. De forma sintetizada pode-se afirmar que as componentes A <HH.VV*> e A Entropia , têm significativa contribuição no procedimento classificatório, sobretudo pelo método Context, cujo desempenho alcançou com 78% de exatidão global e índice Kappa de 0,70.PalavRas-chave: mapeamento florestal, radar, classificação polarimétrica, Amazônia. Analysis of the potential use from TerraSAR-X images for thematic mapping in SW Brazilian Amazon region abstRactThe objective of this work was to analyze the potential use of SAR polarimetric images from the TerraSAR-X sensor system, at StripMap mode, to map land use and land cover in SW Brazilian Amazon. Amplitude images at polarizations A HH , A VV , A <HH.VV*> , derived from the co-variance matrix, as well as the entropy A Entropia , derived from the decomposition of targets by eigenvalues, are parts of the datasets investigated individually or in combined form. Two classifiers were used: the first is based on statistical functions of maximum likelihood (MAXVER), and the second is the contextual method (Context). The thematic results from these classifications were evaluated by a confusion matrix and by the Kappa index. Summarizing we can state that the components A <HH.VV*> and A Entropia , gave a significant contribution to the image classification procedure, considering specially the Context method, whose performance reached 78% of Global Accuracy and a Kappa index of 0.70.

Tropical land use land cover mapping in Pará (Brazil) using discriminative Markov random fields and multi-temporal TerraSAR-X data

“…While the classification of clean pasture and shrubby pasture remains challenging, TS-X data constitutes an adequate data source for forest / non-forest mapping. The PA and UA for forest are higher compared to the accuracies achieved for the other classes, and are in accordance with the accuracies of comparable studies (Schlund et al, 2013;Garcia et al, 2011). While we did not perform specific analysis on the differences of HH-HV and VV-VH polarized data sets, Table 3 shows the OA of the internal HH-HV polarized scenes (06-30) to especially benefit from the multi-temporal integration, and also its neighboring scenes to benefit disproportionately.…”

“…Figure 3 visualizes the classes considered in our classification scheme. The considered LULC classes match comparable studies using TS-X data in Brazilian, or tropical settings, respectively (Garcia et al, 2011;Schlund et al, 2013). As the time period of our study falls into the dry season between June and September, corresponding multispectral remote sensing data could be interpreted sufficiently well.…”

Dual-season and full-polarimetric C band SAR assessment for vegetation mapping in the Amazon várzea wetlands

“…Overall, our results confirm that PolSAR images are more efficient than dual-pol SAR images in mapping wetland land cover and vegetation types. PolSAR classification accuracies reported in the literature vary from 64% (Brisco et al, 2011) to 95% (Brisco et al, 2013), but more frequently between 75% and 90% (Ainsworth, Kelly, & Lee, 2009;Garcia, Roberto, Mura, Johann, & Kux, 2012;Lee, Grunes, & Pottier, 2001;Millard & Richardson, 2013;Qi et al, 2012;Sartori, Imai, Mura, Novo, & Silva, 2011). The accuracies yielded in this study ranged from κ~0.5 to 0.83, AD from~30% to 2% and QD from 15.3% to 3.33%.…”

Applications of Sar Data to Estimate Forest Biophysical Variables in Brazil

“…The polarization of the SAR signal, which refers to the orientation of the electric field emitted and received by the SAR sensor in the vertical (V) or horizontal (H) axis, and can be co-polarized (VV and HH -emitted and received vertically or horizontally, respectively), or cross-polarized (VH, and HV -emitted in one orientation and received in another), will have a great effect on the signal backscattered by vegetation (SOUZA et al, 2019). When additional information from backscattering phase is available, the magnitude of the polarimetric response makes possible the characterization of the objects scattering mechanisms as well (GARCIA et al, 2012).…”