Avaliação de índices de vegetação no monitoramento da cultura do milho (Zea mays L.) com ARP's em diferentes alturas de voo

Abstract

The growing demand for sustainable agricultural practices has driven the use of drones (UAVs) for crop monitoring, enabling more precise management of crop development. In Brazil, the Robusto hybrid maize stands out for its hardiness and adaptability. RGB optical sensors allow the calculation of vegetation indices such as RGBVI and VARI, estimating plant density and vigor without the need for infrared, thus reducing operational costs. However, the accuracy of these indices depends on flight altitude, phenological stage, and image resolution. Therefore, this study aimed to evaluate visible-range vegetation indices for monitoring maize (Zea mays L.) using UAV-derived images at different flight altitudes and phenological stages at the experimental site of Chã do Jardim, Areia, Paraíba.The experiment began with mowing, plowing, and harrowing the soil, followed by marking the field heads for planting Robusto hybrid maize on 08/01/2024, with a spacing of 0.10 m × 1 m. Crop development was limited by lack of rainfall and the use of an older planter. Data collection occurred on 09/02/2024 and 10/09/2024, between 11:00 a.m. and 1:00 p.m., using a DJI Air 2S drone with the Drone Harmony app, performing three flights per day at altitudes of 20, 40, and 60 m. Issues such as wind and clouds were manually controlled. Captured images were processed in Agisoft PhotoScan to generate orthomosaics and analyzed in QGIS to calculate RGBVI and VARI indices. The orthomosaics were divided into 400 parts, and five samples were exported to .csv and organized in Excel. Results showed that a 20 m flight altitude provided higher pixel concentration and greater precision. The VARI index exhibited greater amplitude, while RGBVI was more stable, better tracking maize vegetative growth. The response of VARI and RGBVI vegetation indices varied according to flight altitude and maize phenological stage. Flights at 20 m provided greater detail of canopy cover, with VARI showing higher sensitivity to local variations and RGBVI maintaining stability. Higher altitudes (40 and 60 m) reduced precision, especially for RGBVI, due to lower spatial resolution. Results indicate that there is no single ideal flight altitude; it is necessary to balance resolution and crop stage to minimize errors, with VARI being more sensitive and RGBVI more consistent throughout maize development. Therefore, RGB drone monitoring of maize is effective, with the highest accuracy at 20 m due to high resolution. VARI is more sensitive to local variations, while RGBVI is more stable but loses detail at higher altitudes. Thus, low-altitude flights are ideal in early stages, using VARI to detect subtle variations and RGBVI to track overall growth.