ALTURA DE VOO INTERFERE NA DETECÇÃO ESPECTRAL DE Heterodera glycines EM SOJA?

Abstract

Driven by a growing demand for soy, Brazil has reached the leadership in the world ranking of crop production in the 2020/21 harvest. However, attacks by plant parasitic nematodes such as Heterodera glycines can reduce productivity. Non-destructive monitoring methods such as remote sensing can reduce the number of samples and the monitoring cost, turning possible to identify management methods that consider the location of the attack. Flights with drones at greater heights turn possible to cover a greater area, consequently with lower financial and time costs. However, its worse spatial resolution can confound the nematode effect due to the high munber of mixed pixels. Flights at low altitude generate images with smaller pixels, but it becomes a time consuming process as it demands a greater number of images, longer flight time and higher processing cost. The aim of this study was to determine an ideal flight height and spectral wavelengths to identify the plants attacked by H. glycines. With the aid of a portable GPS device, the nematological sampling points were demarcated. Four singles soil and root samples (0 to 20 cm) were collected and combined into a composite sample to quantify the nematode population in them. Drone flights were performed on the same day of nematode collection using a multispectral sensor at different heights: 40 m (3 cm pixel), 80 m (6 cm pixel) and 120 m (9 cm pixel). The evaluated variables were submitted separately to analysis of variance and the means were compared by the Forward model selection and Akaike's Information Criterion (AIC) and multivariate analysis of principal components (PCA). The different species of nematodes present in this study influence the choice of the spectral band and can be identified with changes in the mathematical model/association of bands and/or flight height.