USO DE IMAGENS MULTIESPECTRAIS OBTIDAS COM RPAS NA PREDIÇÃO DE PARÂMETROS BIOFÍSICOS DA CULTURA DO FEIJOEIRO

Abstract

With the advent of photogrammetry, widely used in precision agriculture, it is currently possible to predict the nutritional, phytosanitary and productive condition of different crops quickly, in a non-destructive way and with greater efficiency compared to conventional techniques, using RGB or multispectral images captured by specific cameras on board unmanned aerial vehicles or even satellites. In this context, the objective of this study was to evaluate the performance of eight vegetation indices (Multispectral: Normalized Difference Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI), Normalized Difference Red Edge Index (NDRE), Modified Chlorophyll Absorption in Reflective Index (MCARI), Leaf Chlorophyll Index (LCI) and Structure Insensitive Pigment Index 2 (SIPI2); RGB: Visible Atmospherically Resistant Index (VARI) and Triangular Greenness Index (TGI)), in the bean crop (Phaseolus vulgaris L.) subjected to the application of 4 foliar magnesium doses (0, 250, 500 and 1000 g.ha-1) in two different growth stages (V4 - third developed trefoil and R6 - full flowering). Correlations were established between these indices and the characteristics of the crop analyzed "in loco", as a function of the treatments tested. The experiment was conducted in a central pivot area and the images collected by a multispectral sensor attached to a drone. The data were submitted to regression analysis and the interaction of variables analyzed by statistical significance, correlation coefficient (r) determination coefficient (R²) and root mean square error (RMSE). The NDVI and MCARI vegetation indices stood out for yield prediction, both with r = 0.82 and RMSE of 330 and 329 kg.ha-1 respectively. TGI performed better for plant height (r = 0.73 and RMSE = 7.4 cm). The best index for detecting chlorophyll content was MCARI (r = 0.81; R² = 0.66 and RMSE = 10.14), followed by NDVI (r = 0.81; R² = 0.65 and RMSE = 10.19). The growth stage with the highest accuracy in estimating the variables was R9 (Physiological maturity) at its onset. GNDVI at R6 and R9 and VARI at R9, were significant at 5% for the treatment magnesium doses, the quadratic regression fits indicated the best dose at 500 g.ha-1.