| Growth and health status of plants can be reflected by detecting their physiological information which is a foundation of intelligent cultivation. A base and key techniques of modern agriculture is how to realize that a new method of quickly, exactly and non-destructive detecting useful information of plant instead of routine way of time-consuming methods, complex analysis courses and high cost. Recent research on non-destructive testing technology of plant physiology information has been given more and more attention in all over the world especially in China. This work is supported by National High Technology Research and Development Program 863"Laser-induced Plant Physiological Information Detection Sensors and Diagnosis"(2007AA10Z203). A laser-induced chlorophyll fluorescence spectrum acquisition system was established firstly. Detection of chlorophyll fluorescence spectroscopy technique, chemometrics techniques combined with modern physical-chemical analysis techniques were applied to research on quantitative models for leaf moisture, chlorophyll content, fluorescence parameter Fv/Fm, net photosynthesis rate, transpiration rate, stomatal conductance and intercellular CO2 concentration for cucumber and poplar as objects. Qualitative models were established for cucumber pests and diseases diagnosis of base on chlorophyll fluorescence spectroscopy technique too.The main contents and conclusions in this paper were as follows:1. The relationship between chlorophyll fluorescence intensity and laser power was researched. Fluorescence spectrums, which were induced by four power (2.50mW, 5.00mW, 7.50mW, 10.00mW) of center wavelength 473nm and 660nm laser respectively, were analyzed in the experiment. Under this condition, fluorescence intensity variation of center wavelength 685nm and 732nm fluorescence in different plant physiology information (chlorophyll content, water content) were calculated and analyzed by MATLAB software. Results showed that there was a very significant linear correlation between fluorescence intensity of every peak position and laser power(R>0.91); chlorophyll content parameter impacted on the relationship between fluorescence intensity and laser power significantly, and a linear correlation between fluorescence intensity gradient of every peak position and chlorophyll content was found(R>0.86). Based on these results a mathematical model with chlorophyll content parameter of relationship between fluorescence intensity and laser power was established.2. Under different excitation conditions, the relationship between leaf chlorophyll content and chlorophyll fluorescence spectrum was analyzed quantitatively with PLSR (Partial Least Squares Regression, PLSR). The results indicated that model accuracy was the best in 473nm and 7.50mW laser, so this exciting condition is optimal.3. An acquisition system for plant physiological information determination was set up and the software system was developed in this paper. And based on these we applied an invention patent.4. Leaf moisture, chlorophyll content, fluorescence parameter Fv/Fm, net photosynthesis rate, transpiration rate, stomatal conductance and intercellular CO2 concentration of cucumber and poplar leaf samples were analyzed quantitatively by chlorophyll fluorescence spectroscopy technique.After spectra outliers being eliminated by Chauvenet theorem and concentration outliers being eliminated by leverage, student residual testing and COOK distance, the number of samples used for quantitative analysis of leaf moisture, chlorophyll content, fluorescence parameter Fv/Fm, net photosynthesis rate, transpiration rate, stomatal conductance, intercellular CO2 concentration were 194, 151, 149, 144, 143, 145, 148 for cucumber and 146, 146, 147, 147, 151, 143, 148 for poplar respectively.Comparison results of PLSR, BP (Back Propagation Neural Network, BP) and LS-SVM (Least Squares-Support Vector Machine, LS-SVM) with five wavelength and five pretreatment methods indicated that: (1) The established LS-SVM model with input was the best by the first ten principal components of 627-826nm combined with SNV (Standard Normal Variate, SNV) for leaf moisture. BP model with input was the best by the first eight principalμcomponent of full-band spectrum combined with SNV for chlorophyll content. LS-SVM with input was the best by the first nine principal component of full-band spectrum with WA (Wavelet Analysis, WA) for fluorescence parameter Fv/Fm. LS-SVM with input was the best by the first eight principal components of 627-864nm combined with WA for net photosynthesis rate. LS-SVM with input was the best by the first ten principal components of full-band spectrum combined with SNV for transpiration rate. BP with input was the best by the first ten principal components of full-band original spectrum for stomatal conductance. PLSR with input was the best by the fist twelve factor components of full-band spectrum combined with FDT (First-order Differential Treatment, FDT) for intercellular CO2 concentration.(2) The correlation coefficient of calibration set, root mean square error of calibration set, the correlation coefficient prediction set and root mean square error of prediction set were as follows: a) for cucumber leaf moisture, 0.9640, 0.4520%, 0.9442, 0.6116%, b) for poplar leaf moisture, 0.9546, 0.9941%, 0.9368, 1.2688%, c) for cucumber chlorophyll content, 0.9699, 0.0941mg/kg, 0.9519, 0.1511mg/kg, d) for polar chlorophyll content, 0.9775, 0.0944mg/kg, 0.9665, 0.1396mg/kg, e) for cucumber fluorescence parameter Fv/Fm, 0.9319, 0.0086, 0.9134, 0.0103, f) for poplar fluorescence parameter Fv/Fm, 0.9426, 0.0076, 0.9098, 0.0106, g) for cucumber net photosynthesis rate, 0.9519, 0.4632μmolm-2s-1, 0.9374, 0.5065μmolm-2s-1, h) for poplar net photosynthesis rate, 0.9439, 0.9034μmolm-2s-1, 0.9038, 1.1389μmolm-2s-1, i) for cucumber transpiration rate, 0.9786, 0.0349 mmolm-2s-1, 0.9448, 0.0449 mmolm-2s-1; j) for poplar transpiration rate, 0.9289, 0.3165 mmolm-2s-1, 0.9065, 0.3594 mmolm-2s-1, k) for cucumber stomatal conductance, 0.9378, 1.0123 mmolm-2s-1, 0.9244, 1.0841 mmolm-2s-1, l) for poplar stomatal conductance, 0.9554, 5.8858 mmolm-2s-1, 0.9097, 7.9426 mmolm-2s-1, m) for cucumber intercellular CO2 concentration, 0.9560, 3.1404μmolmol-1, 0.9249, 3.8901μmolmol-1, n) for poplar intercellular CO2 concentration, 0.9430, 5.7503μmolmol-1, 0.9080, 7.2289μmolmol-1.5. Two hundred and nineteen cucumber samples were used to discriminate among downy mildew disease, aphid disease, downy mildew and aphid disease and health by analyzing qualitatively:Classification results of PCA (Principal Component Analysis, PCA), DA (Discriminant Analysis, DA), DPLS (Discriminant Partial Least Squares, DPLS), BP and LS-SVM, and optimization results of five modeling bands (504-627nm, 627-826nm, 827-900nm, 627-786nm, 504-900nm) and five pretreatment methods (Savitzky-Golay smooth, FDT, FFT, SNV, WA) with input by different principal (factor) components were compared to indicate that established classification correctness model by BP with input by the first seven principal components of full-band spectrum combined with FDT was optimal. The optimal model had accurate rates of 100% for calibration and prediction set.Seven indicators of plant physiological information were analyzed quantitatively by chlorophyll fluorescence spectroscopy technique in this paper. Two kinds of plant sample (cucumber and poplar) were only taken as examples, so the results and conclusions were some specificity. A great variety of plant samples should be added into the models for enhancing accuracy and comprehensiveness in the future. The obtained results in this study indicated the potentiality of chlorophyll fluorescence spectroscopy technique to rapidly detect plant physiological information and discriminate plant diseases and insect pests. The study was a foundation of developing an instrument based on chlorophyll fluorescence spectroscopy technique for rapid detecting plant physiological information.
|
PDF Full Text Request