Estimation of Anthocyanins in Whole-Fertility Maize Leaves Based on Ground-Based Hyperspectral Measurements

The estimation of anthocyanin (Anth) content is very important for observing the physiological state plants under environmental stress. objective this study was to estimate Anth maize leaves at different growth stages based on remote sensing methods. In study, hyperspectral reflectance and corresponding were measured critical nodulation, tasseling, lactation, finishing maize. First-order differential spectra (FD) derived from original (OS). First, spectral parameters highly correlated with selected. A total two sensitive bands (Rλ), five classical vegetation indices (VIS), six optimized (VIC) selected first-order spectra. Then, univariate regression models (Anth-UR models) multivariate estimating anthocyanins (Anth-MR constructed these It shown that conversion could effectively improve correlation between Rλ, VIC, Anth, VIC are usually more than VIS. addition, overall performance Anth-MR better Anth-UR models. Among them, combination three types (FD(Rλ) + OS_VIC FD_VIC/VIS) as inputs had best performance. Moreover, an impact models, tasseling lactation showing results. best-performing follows. For stage, model FD(Rλ) VIS-based SVM model, R2 0.868, RMSE 0.007, RPD 2.19. FD_VIC-based RF 0.797, 2.24. These results will provide a scientific basis monitoring using techniques.