MAPEAMENTO DE NEMATOIDES NA CULTURA CAFEEIRA A PARTIR DE IMAGENS MULTIESPECTRAIS OBTIDAS POR AERONAVES REMOTAMENTE PILOTADAS

Abstract

Remote sensing has been increasingly applied in precision agriculture due to its strong correlation with agricultural variables. The spectral combination of visible bands has become an alternative cameras with absence of near infrared, due to a relative correlation with the vegetation and, consequently, with the agricultural variables. In this way it is possible to identify changes present in a plantation due to pest and pathogen attack, being possible to detect the presence of nematodes in the coffee crop with this tool. In this context low-cost remotely piloted aircraft (RPAS) allow the continuous monitoring of these crops being an increasingly used alternative for the generation of pest/pathogens management maps, assisting in monitoring productivity of the plantations, minimizing losses. In this work, the classification by the maximum likelihood method was used to discriminate areas of the coffee crop infested by nematodes. For this, the classification was applied in a combination of Normalized Difference Vegetation Index and analysis of the accuracy of the mapping was validated from the Kappa index. The confounding matrix presented 20% omission error (for nematode), 28.57% commission (for exposed soil), being the highest values obtained, in addition to the Kappa index of 0.75.