Using near infrared spectrum analysis to predict water, chlorophyll content in tomato leaves

摘要

In this study, we developed a nondestructive way to analyze water and chlorophyll content in tomato leaves. A total of 200 leaves were collected as experimental materials, 120 of them were used to form a calibration data set. Drying chest, SPAD meter and NIR spectrometer were used to get water content, chlorophyll content and spectrums of tomato leaves respectively. The Fourier Transform Infrared (FTNIR) method with a smart Near-IR Updrift was used to test spectrums, and partial least squares (PLS) technique was used to analyze the data we get by normal experimentation and near infrared spectrometer, set up a calibration model to predict the leaf water and chlorophyll content based on the characteristics of diffuse reflectance spectrums of tomato leaves. Three different mathematical treatments were used in spectrums processing: different wavelength range, different smoothing points, first and second derivative. We can get best prediction model when we select full range (800-2500nm), 3 points for spectrums smoothing and spectrums by baseline correction, the best model of chlorophyll content has a root mean square error of prediction (RMSEP) of 8.16 and a calibration correlation coefficient (R~2) value of 0.89452 and the best model of water content has a root mean square error of prediction (RMSEP) of 0.0214 and a calibration correlation coefficient (R~2) value of 0.91043.