ESTIMATING THE PLANT NITROGEN CONTENT OF FOXTAIL MILLET (SETARIA ITALICA L.) BASED ON CONTINUOUS WAVELET ANALYSIS

作者：Fuxai Xia, Meichen Feng, S.A. ZHU, C. WANG, Tianqi Mu, L.J. XIAO, W.D. YANG, M.J. ZHANG, X.Y. SONG, H. YANG, Ming-Yu Qin · 发表于：Applied Ecology and Environmental Research · 年份：2022 · DOI：10.15666/aeer/2005_43914407 · 被引用次数：1 · 研究领域：Greenhouse Technology and Climate Control

The variation in the plant nitrogen content (PNC) directly characterizes the growth of foxtail millet, and estimation of the PNC using hyperspectral techniques is important for effective evaluating the growth of this species. Effective statistical modeling methods can improve the accuracy and reliability of PNC estimates. In this study, field experiments were conducted under different gradients of organic fertilizer to develop an estimation model for determining the PNC in foxtail millet. The continuous wavelet transform (CWT) was used to process the collected reflection spectra and construct partial least square regression (PLSR), random forest (RF) and support vector machine (SVM) estimation models. Among the common wavelet families, Daubechies (db5), Coiflets (coif3), Biorthogonal (bior1.5), Symlets (sym8), haar and rbio3.1 were selected to analyze the correlation with the PNC, and all the wavelet functions had a good correlation with the PNC. The correlation coefficients were 0.834, -0.835, -0.973, -0.784, -0.789 and -0.770, respectively. The CWT technique can significantly improve the prediction accuracy of the PNC. The best PNC estimate was obtained using db5 (R 2 cal = 0.859, RMSEcal = 3.415), and the best decomposition scale was 2 4 . In addition, the validation data indicate that db5-RF can be used to estimate the PNC (R 2 val = 0.935, RMSEval = 1.311, RPDval = 3.32). This study provides a reference for the practical application of PNC analysis in foxtail millet.