Estimating Nitrogen Status of Plant by Vis/NIR Spectroscopy and Mathematical Model

Abstract

This paper investigated the potential of Vis/NIR spectroscopy and chemometrics to estimate N status of plant. Chemometrics was used as Vis/NIR spectroscopy analysis method to establish models to estimate N status of rapeseed and tea plant. In the research of rapeseed plant, a hybrid estimation model, artificial neural network (ANN) combined with partial least square regression (PLS) method, has been developed for diagnosis of nitrogen nutrition of rapeseed plant. 5 optimal PLS principal components were were selected as the input of BP neural network to establish the prediction model. The result showed that the prediction performance was excellent with r=0.95405, and the accuracy of prediction reached 95%. In the research of tea plant, PLS method was used to look for the fingerprint wavelengths (488, 695 and 931 nm). The PLS model for predicting the N status with r=0.908, SEP=0.21 and bias=0.138, showed an excellent prediction performance. Thus, it was concluded that chemometrics was a good tool for the spectroscopic estimation of plant N status based on Vis/NIRS.