Discriminação de classes de teores de nitrogênio e de clorofila foliares e de produtividade do feijoeiro usando sistema de sensoriamento remoto de baixo custo

• Ano de defesa: 2011
• Autor(a) principal: Abrahão, Selma Alves
• Orientador(a): Pinto, Francisco de Assis de Carvalho
• Banca de defesa: Araújo, Geraldo Antônio de Andrade , Paula Júnior, Trazilbo José de , Varella, Carlos Alberto Alves
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Viçosa
• Programa de Pós-Graduação: Doutorado em Engenharia Agrícola
• Departamento: Construções rurais e ambiência; Energia na agricultura; Mecanização agrícola; Processamento de produ
• País: BR
• Palavras-chave em Português: Agricultura de precisão; Índices de vegetação
• Palavras-chave em Inglês: Precision Agriculture; Vegetation indices
• Área do conhecimento CNPq: CNPQ::CIENCIAS AGRARIAS::ENGENHARIA AGRICOLA
• Link de acesso: http://locus.ufv.br/handle/123456789/707

This study aimed to develop classifiers based on different combination of spectral bands and vegetation indices of original, segmented and reflectance images for nitrogen and chlorophyll contents and yield for discriminating common beans. To define the best time for obtaining the crop spectral characteristics was also aimed by this study.The first phase of work was to evaluate a radiometric calibration system and to study the relationship between vegetation indices and leaf area index.The research was developed with the treatments arranged in a factorial 3 x 4, three N rates applied at planting (0, 20 e 40 kg ha-1) and four N rates applied sidedress(0, 20, 40 e 60 kg ha-1), according to the randomize block design with five replicates, in the Agronomy field of UFV/Florestal, Florestal-MG, from March to June of 2010, considered to be the dry common beans season. It was used a remote sensing platform with a helium balloon and two small-format digital cameras. In addition to the spectral bands alone, it was tested four vegetation indices: simple ratio (RS), red normalized difference (NDVI), green normalized difference (GNDVI) and modified chlorophyll absorption (MCAIR1). The radiometric calibration system was able to correct for variations in the intensities of the pixels caused by changes in lighting conditions. The regression models between the vegetation indices and logarithm of the leaf area index were those which tended to present the best fit to explain the relationship under study. This procedure resulted in higher values for the test F of regression and higher values of coefficients of determination.The classifiers proved to be efficient in the discrimination of leaf nitrogen and chlorophyll contents and yield. The best time to determine leaf N content was at 13 DAE (days after emergence), with Kappa ranging from 0.26 to 0.31. This result was obtained when using two vegetation indices from segmented images, an index related to the canopy structure and another related to chlorophyll. The best time to discriminate chlorophyll content was at 21 DAE, with Kappa of 0.47, when using red and blue bands from original images.The better times to discriminate yield were at 13 and 26 DAE. The best time to discriminate yield was at 13 DAE, with Kappa of 0.22, when using green band from segmented images.The best time to discriminate yield was at 26 DAE, with Kappa of 0.21, when using red and green bands from original images.