Agricultura de precisão por sensoriamento remoto: estudo aplicado na fertirrigação de cana-de-açúcar

• Ano de defesa: 2022
• Autor(a) principal: Gustavo Ferreira de Souza
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; Brasil; IGC - INSTITUTO DE GEOCIENCIAS; Programa de Pós-Graduação em Geografia; UFMG
• Palavras-chave em Português: Sensoriamento remoto; Geotecnologia; Agricultura; Fertirrigação; Agricultura de precisão; Cana-de-açúcar
• Link de acesso: http://hdl.handle.net/1843/47030

This work offers an applied methodological contribution of monitoring, control and technological evaluation by remote sensing in the agricultural management of areas irrigated with vinasse (Fertirrigation) in the production of sugarcane, to provide productivity with sustainability. The challenge of production in large areas and the limited geographic capacity of measurements in the field, the availability of satellite data was applied to the study, with a view to discussing monitoring in an agricultural production process. The fertigation process was the object chosen due to its high added value, with the execution of three processes in one: 1 - It transforms a residue into an input, 2 - irrigates the cane in the driest period 3 - fertilizes the soil. The production site occupies a large geographic area (20 thousand hectares), spatially distributed in nearby and interconnected farms, but difficult to control and monitor the quality of the process. The study aimed to analyze and develop a method with remote sensing (SR) techniques, spectral and object-oriented classification, to identify productivity correlation with spectral indices for monitoring fertirrigation; with different remote orbital sensors. Data from Sentinel 1 and 2 satellites were the basis for evaluating the mapping model and the impact of variation in fertigation correlated with Sentinel area productivity data. In the statistical evaluation, Person correlation, there was a correlation in the tested images (Sentinel 2, Pearson 0.4 and Anova 0.7) with vegetation indices extracted for those from 2017 to 2019. For the years 2019 to 2021, the test used the methods for evaluating only the month of April (sugarcane growth peak) had correlation (Pearson 0.98) and multiple regression (For NDVI and NDWI variables p<0.02). In the object-oriented classification, a model based on the LSMS (Large Scale Mean Shift) was defined, which separated the fertigated areas and the applied quality, based on spectral response and field experience. For classification, parameters were defined to assess the quality of fertigation (spatial radius 5, radius interval 200 and minimum size 50). The method represented a way to evaluate and measure the quality of the application of vinasse as a way to irrigate and fertilize at the same time. With the periodicity of the satellites of 5 days of revisit, it allows correction in situations of identification of failures in the application of vinasse. The work presented a tool that generated a product applied operationally for monitoring the quality in the coverage of vinasse release in sprinkler irrigation.