Separabilidade das principais culturas agrícolas do estado de Mato Grosso, a partir de imagens multitemporais do sensor MODIS

• Ano de defesa: 2013
• Autor(a) principal: Souto, Roberto Nunes Vianconi
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Mato Grosso; Brasil; Instituto de Ciências Humanas e Sociais (ICHS); UFMT CUC - Cuiabá; Programa de Pós-Graduação em Geografia
• Palavras-chave em Português: Sensor MODIS; EVI; Separabilidade; Culturas agrícolas; CNPQ::CIENCIAS HUMANAS::GEOGRAFIA; Separability; Agricultural crops
• Link de acesso: http://ri.ufmt.br/handle/1/1262

Agriculture in Mato Grosso has ever intensifying its production and has excelled, reaching a level of lead in the country. In this context, remote sensing has been used in the mapping and monitoring of growing areas for acquisition of agricultural information updated. Despite its potential, there is also limiting the performance of sensor systems available, that due to issues such as their spatial resolution, but mainly by the temporal resolution and availability of good quality images during the period of cultivation. Thus, the present study aimed to investigate the separability of the main crops in the Planalto dos Guimarães, in the state of Mato Grosso, through the analysis of multitemporal images of the sensor system Terra MODIS. The methodology consisted in the systematic analysis of EVI of major crops, together with monitoring of phenological development by periodic work field. The similarity of the time profiles of EVI was examined from a hierarchical cluster analysis, and thus generated subclasses similar to each culture. The pixel selection of images, from the quality indicators Quality Assurance (QA) Reliability and resulted in the generation of eight sets of images that were used for the analysis of separability between subclasses aggregate. As indicators of separability were used to fashion the mean value t test T unpaired and counting pairs of values of EVI of subclasses with a significant difference between the averages for the image of each date and level of QA. Results show that a better separability requires the use of images with higher quality and MODIS images should not be classified without prior evaluation of the quality of the pixels. However, between the second half of December and first week of January, which usually happens as much vegetative growth of cultivars of the first season, there is great difficulty in finding pixels valid at levels higher quality, which are nonexistent levels QA 0001 and QA 0010.