ÍNDICES DE VEGETAÇÃO NA PREDIÇÃO DE PRODUTIVIDADE DE SOJA

Abstract

Revolution 4.0 is directing agribusiness to employ more technology to help manage agricultural crops. The use of drones for imaging offers a new perspective on plant health, turning possible to verity flaws and spots that are not observed in the soil. Thus, this work aimed to evaluate the ability to correlate soybean yield obtained with previous crop imaging, under the effects of different second crop alternatives prior to soybean cultivation. A drone with onboard multispectral camera was used, flying during a 2018/2019, 2019/2020 and 2020/2021 harvest at an experimental station in the city of Rio Verde, GO. The flights were carried out in four phenological stages of soybean (R3, R5, R6 and R7) and seven phenological stages of corn and other cover crops (V1, V3, V7, V8, R1, R4 and R6) at different heights in the first, second and third harvest, respectively. The experimental design was in DC, arranging treatments in strips, totaling 14 treatments and 12 replications, totaling 168 sample plots of 3 m2. The results obtained for the NDVI index (sum, mean and median) were evaluated in relation to soybean yield corrected by the Pearson linear correlation method at the level of 1 and 5% probability by the F test. The results obtained show signs of ability to use technology to predict soybean yield. However, the second harvest systems that proceed, height of flight, phenological stage, and time of flight (harvest or off-season) directly influence the response. It is concluded that there is a need of refinement in the parameters to use the images in agricultural experimentation, to obtain robust correlations through imaging.