Vegetation indices in the estimation of common bean yield cultivated under nitrogen rates

Abstract

Remote sensing presents several applications in agriculture, highlighting the definition of specific management areas and estimation of crop yield. The aim of this study was to compare the accuracy of two vegetation indices (NDVI and IRVI) in the estimation of grain and biomass yield of common bean cultivated under nitrogen doses in four phenological stages. The experiment was carried out in the winter crop of 2018. The experiment had five treatments, under N rates in the topdressing (0, 50, 100, 150 and 200 kg ha-1), with four repetitions. The NDVI and IRVI vegetation indices were obtained by the Greenseeker active sensor. Regressions of the mean value of NDVI and IRVI of each plot were plotted with the biomass and grain yield of the crop. The regressions were submitted to the analysis (p < 0.05) and, when significant, the adjustments were tested by the coefficient of determination (R2) and root mean square error (RMSE). The accuracy of estimating common bean grain yield by means of vegetation index (R2 = 0.71, RMSE = 162 kg ha-1) is higher than for biomass yield (R2 = 0.52, RMSE = 898 kg ha-1). The phenological stage with higher accuracy in the estimation of yields is grain filling (R8).