Aplicação computacional AGDATABOX-RS: gerenciamento de dados de sensoriamento remoto

• Autor(a) principal: Conti, Giuvane
• Data de Publicação: 2021
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações do UNIOESTE
• Texto Completo: http://tede.unioeste.br/handle/tede/5718
• topic: Agricultura de precisão; Índices de vegetação; Plataforma web; Zonas de manejo; Precision agriculture; Management zones; Vegetation indexes; Web platform; Sistemas Biológicos e Agroindustriais

Remote sensing has the ability to assist in the evolution of agricultural practices, providing periodic information about the state of a crop over a harvest, at different scales and for different segments. Applications in precision agriculture use remote sensing practices, such as vegetation indexes, from multispectral images, to measure physical and chemical parameters of plants, along their development cycle. Technological advances allowed the development of innovative services for the agricultural sector, based on the internet and hosted in the cloud. Therefore, the objective of this research was to develop a computational module that integrates and provides remote sensing data for the AgDataBox precision agriculture platform. The developed application allows the persistence of a new area (field), the search for raster images of orbital satellites, the selection vegetation indexes, as well as vectorizing and inserting images of interest in the AgDataBox platform. The proposed module was tested with data from the 2018/2019 corn crop (summer harvest), in a study area at Céu Azul, Paraná. Twelve vectors were generated from Sentinel-2 satellite images, using the normalized difference vegetation index (NDVI), enhanced vegetation index (EVI) and enhanced vegetation index (EVI-2) of 11/12/2018, 12/16/2018, 15/01/2019 and 25/01/2019 and persisted in the AgDataBox platform. In addition, vectors were persisted with variables of productivity, altitude, sand, silt, clay, mechanical resistance to soil penetration at depths of 0-0.1, 0-0.2, 0.1-0.2, and 0.2-2.3 m. After autocorrelation analysis between the variables, with productivity as the target variable, EVI-2 and altitude were selected as the variables that showed the best cross autocorrelation with the target variable. Management zones (MZs) were delineated in the AgDataBox-Map module, using the fuzzy c-means method, for two, three, and four classes using three sets of input variables: (i) EVI-2_NM, (ii) Altitude, and (iii) EVI 2_NM + Altitude. After analyzing the results, it was concluded that the best design used the variable EVI-2 in the design of three classes of MZs. All designs for two classes showed statistical differences between their classes, with the best performance being obtained with the altitude variable. All designs with four classes were discarded, as there was no statistically significant difference between their classes.