An Approach to Orbital Image Classification for the Assessment of Potato Plantation Areas

• Autor(a) principal: Boulomytis, Vassiliki Terezinha Galvão
• Data de Publicação: 2013
• Outros Autores: Alves, Claudia Durand
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Labor & Engenho (Online)
• Texto Completo: https://periodicos.sbu.unicamp.br/ojs/index.php/labore/article/view/164
• topic: Pixel-based Classification; OBIA; Assessment of Plantations; Preservation Area Interference; Water Contamination Risk.; Pixel-based Classification; OBIA; Assessment of plantations; Preservation area interference; Water contamination risk

In the city of Bueno Brandão, South of Minas Gerais State, Brazil, the Watershed of Rio das Antas is located prior to the public water supply and is susceptible to hydro-degradation due to the intensive agricultural activities developed in the area. The potato plantation is the most significant cropping in the city. Because of the possibility of interfering in the preservation areas, mainly the ones surrounding water courses and springs, it is very important to do the assessment of the plantation sites, in order to avoid the risk of water contamination. The procedures adopted by the agro activity farmers generally present the following features: intensive use of agro-chemicals, cropping in places with slopes which are higher than 20%, close to or in permanent preservation areas. The scope of this study was to develop the proper methodology for the assessment of the plantation areas, regarding the short time of procedure, as the period between the plantation and the harvest occurs in six months the furthest. These areas vary year in year out, as the plantation sites often change due to the land degradation. Because of that, geotechnologies are recommended to detect the plantation areas by the use of satellite images and accurate data processing. Considering the availability of LANDSAT medium resolution images, methods for their appropriate classification were approached to provide effective target detection.