UTILIZAÇÃO DE VEÍCULOS AÉREOS NÃO TRIPULADOS PARA ESTIMAR A DIVERSIDADE ESPECTRAL DE VARIEDADES DE SOJA

Abstract

The work aimed to evaluate the efficiency of sensors embedded in UAVs, in terms of spectral differentiation between soybean varieties. The experiment was carried out at IF Goiano-Rio Verde, where 32 soybean varieties were evaluated in 3 replications using randomized blocks and generating the spectral bands: NIR, REDEDGE, PRI, NPRI, NDVI, NDRE and Red wavelengths, Green, Blue, 586, 615, 661, 775, 825. Productivity, maturation group and mass of 100 grains were evaluated. To verify if there was a spectral difference between the genotypes, the data were subjected to analysis of variance (ANOVA) by the F test at 5% probability, from the analysis of variance the residual variance and covariance matrix was obtained and for this the dissimilarity matrix by the generalized Mahalanobis distance (D2), aiming to evaluate the spectral distance between genetic materials. The relative contribution of wavelengths and indices (S.j) to the separation of genotypes was estimated according to Singh (1981). The wavelengths / index that most contributed to the dissimilarity of the genotypes were BLUE, GREEN, NIR, RED and REDEDGE, respectively. The materials with the greatest spectral distance were G14 in relation to G31. The materials with the shortest distance were the G18 in relation to the CD 2728. The CCC was 0.7569, which shows a good precision in the generated cluster, through the generated data it was possible to observe the presence of spectral variability between the varieties .