Evaluation of digital terrain model extracted from SAR data in X-Band of SAR R-99B / Avaliação do Modelo Digital do Terreno Extraído de Dados SAR Interferométricos na Banda X do SAR R-99B

AUTOR(ES)
DATA DE PUBLICAÇÃO

2007

RESUMO

In the Brazilian territory there are several areas with no topographic mapping coverage and the existent maps are not updated due to the continuous land use and land cover changes. The Sistema de Prote¸cao da Amazonia (SIPAM) uses airplanes with synthetic aperture radar which allows to image in several conditions. This radar was designed for recoinassance tasks with a civil and military perspective. The potential of this sensor for mapping has not been studied yet. The purpose of this dissertation was to evaluate the elevations of the interferometric digital terrain model supplied by SIPAM. To reach this aim, it was analyzed the possible factors wich affect the quality of the model extracted by interferometry such as the slope gradient, the aspect, the regions of imagery in range dimension, the land use and land cover. The study site at Sao Jos´e dos Campos, state of Sao Paulo, Brazil was divided in homogeneous regions (strata) which features were related to the factors that would affect the quality of the model. The sample collection in the strata were done with the GPS receivers performing relative positioning (sample points survey) and differential GPS (road survey). The digital terrain model showed shift and rotation problems that were fixed using the surveyed roads as reference. The fixed digital terrain model was classified according to the Brazilian mapping standards as compatible at scale of 1:100.000 (class A) on condition that bias will be removed. The digital terrain model errors have bias, with a mean error of -27,83 m. The strata had a small number of samples being used the nonparametric statistics. After the remotion of bias, the strata did not have significant differences at the regions of imagery in range dimension and at the land use and land cover, in general. The aspect and slope gradient were considered influential on errors. The mountain strata showed a larger variability and the mountain stratum with foreslope and anti-azimuthal aspects tend to underestimate the errors. The ondulated stratum with backslope and azimuthal aspects tend to overestimate the errors with intermediate variability. The stratum composed by grouping the ondulated with foreslope and anti-azimuthal aspects and flat does not show bias and it has intermediate variability. Although the SAR R-99B has a high spatial resolution in interferometric mode (3 m), the digital terrain model extracted was not compatible with scales higher than 1:100.000 (class A), with a result worse than the expected and had a planimetric error of 178,297 m that does not allow its use for mapping unless it is applied first an appropriate geometric correction and bias removal.

ASSUNTO(S)

cartographic evaluation sar r-99b interferometria modelo digital do terreno synthetic aperture radar digital terrain model interferometry radar de abertura sintética sar r-99b avaliação cartográfica

ACESSO AO ARTIGO

Documentos Relacionados