Detecção de estradas não pavimentadas com o uso de imagens orbitais de média resolução espacial.

• Autor(a) principal: Fernandes, Angelo Florentino
• Data de Publicação: 2008
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFG
• Texto Completo: http://repositorio.bc.ufg.br/tede/handle/tde/2928
• topic: Sensoriamento remoto.; CBERS.; Spring.; Estradas rurais.; Erosão.; AGRONOMIA::CIENCIA DO SOLO

Unpaved roads are essential for the municipal districts, mainly for those that have its economy based on the agricultural. However, not always the public power of these municipal districts has control on the amount and characteristics of the existent highways. These information are essential for the organizes of the net and maintenance of the highways of the municipal district, which are usually built without approaches of engineering that it increases the risk of degradation of those highways, what happen basic. It is of fundamental importance that the municipal district has it controls so much about the amount of not paved roads, as on the problems that eventually appear and they need solution. The present work had for objective to verify the adequate of the use of images of medium space resolution for identification of the mesh of not paved highways and recognition of features originated erosive of these. The work was accomplished tends as study object the microbasing of the river Chaves, in the municipal district of Morrinhos, GO. For the accomplishment of the work got up referring field data to not paved highways and erosions to its margins, using a receiving GPS. Images of sensor CCD of the satellite were used CBERS, referring to the months of April and September of 2007, and the software Spring, of the National Institute of Space Researches (INPE). In the study made use of the bands 2, 3 and 4, referring to the areas of the green, red and infra-red close of the spectrum electromagnetic. Vectorial data of the studied area obtained SIEG close to also used. The images were spatial coordinates, being defined the area of the study, being these analyzed band the band, by means of colored compositions. They were also accomplished analyses using supervised classification and filtrated for it enhances of borders. The obtained results showed that, the best form of extraction of that feature type is to manual and that, highways of reduced traverse section and of low volume of traffic they are not recognition passives, as well as features erosive of small load, in the bed, borders or in marginal areas to these. The artifices of digital processing with the use of filters and supervised classification didn't supply satisfactory results for the identification of those features. It was also verified that the production of map of erosion risk with the available scale for the place of the study, didn't present correlation adapted with the features lifted up erosive to field. It is concluded like this, that the use of images of medium space resolution for extraction of highways should be accompanied of field maps being able to not, however, to serve as support for neighborhood studies to these features.