Estimating Chlorophyll And Nitrogen Contents Of Apple Tree Leaf Based On Hyperspectrum

Chlorophyll is an important material for plants’ production and photosynthesis. Nitrogen is one of the important elements for apple’s growth. The estimation of chlorophyll and nitrogen content in apple has great importance. Traditional nutrition diagnosis of apple trees, mostly taken samples back to the laboratory testing for analysis. Though the results are more accurate, it weast time and energy. Hyperspectral remote sensing is the technology that developed rapidly in recent years. It has high spectral resolution and its band continuity is strong. It can estimate nutrient elements of vegetation rapidly, accurately and nondestructive. Therefore, the research to explore the highspectrum estimation of chlorophyll and nitrogen content in apple tree leaves has important theoretical and practical significance for the growth and condition monitoring of the apple trees, and improve the scientific management of the apple tree.In this study, apple orchards in Qixia City and Taian Huangjiazhuang Experiment Station, Shandong Province are chosen as the experimental sites. Samples were collected in shoot growth phenophase, spring shoots pause growth phenophase, autumn shoots pause growth phenophase. The reflection spectrum of apple leaves is measured using ASD Field Spec4. Chlorophyll content and nitrogen content of leaves are determined in laboratory. Sensitive wavelength is chosen by correlation analysis between original reflectance spectra and the first derivative reflectance spectra and chlorophyll and nitrogen. Hyperspectral vegetation indices are established by difference, ratio and normalized measurement with sensitive wavelength. Estimation models for estimating chlorophyll content of apple tree leaves and estimation models for estimating chlorophyll contents of apple leaves based on different derivative windows and estimation models for estimating N content in apple leaves of different phenophases are established with the hyperspectral vegetation indices.Finally, the main conclusions are:(1) Initially found out the hyperspectral characteristics of apple leaves. It found that curve trend of apple leaves in different states are basically the same, but the level of reflectivity is changing. In 400—500 nm, the reflectance is low for absorb blue light. The reflectance become high after 500 nm and reached the highest at 550 nm in green light. The reflectance is low in 600—700nm for absorb red light. In 690—780nm the reflectance become high quikly. In 780—1300nm, the reflectance changes a little. In 1300—1950nm the reflectance is become low and reached the lowest at1450 nm. In 1450—1950nm the reflectance is become high first and then becom low, it reached the lowest at 1950 nm. In 1950—2500nm the reflectance is become high first and then becom low. With the trees growing, the reflectance is become lower in visible light. In short wave near-infrared the reflectance is become high first and then becom low. In long wave near-infrared the reflectance is become lower. With the chlorophyll content growth, the reflectance is become lower. With the N content growth, the reflectance is become lower in visible light. In short wave near-infrared the reflectance is become low first and then becom high. In long wave near-infrared the reflectance is become lower.(2) The study of estimate the chlorophyll content of apple tree leaves at autumn shoots pause growth phenophase, a hyperspectral-based chlorophyll content estimating model for apple tree leaves is proposed in this paper. The apple tree leaves are picked in autumn when they stop growing, and the spectral data and the chlorophyll content in the leaves of apple trees are measured. First derivative(FD) is used to process the spectral data, and choose sensitive parameters. Hyperspectral models for estimating chlorophyll content in the leaves of apple trees are established by single variable(use one variable to establish models) and partial least square(PLS) methods. Four sensitive parameters FDR530、FDR734- FDR 530、( FDR 734- FDR 530)/(FDR 734+ FDR 530)、 FDR 697- FDR 581 are chosen to establish hyperspectral estimating models using partial least square. The partial least square model y=2.936676-323.421911x1+43.289625x2+0.730393x3-104.999517x4 has the highest R2(coefficient of determination), lower RMSE(root mean square error) and RE%(relative error). The partial least square model is more appropriate for estimating chlorophyll content in the leaves of apple tree.(3) Find out the best first order differential window in 1 to 30 is 13. First order differential transformation of 1 to 30 windows is done to original spectral data respectively. Correlation analysis is done between apple leaf chlorophyll content and first order differential data. Two sensitive wavelengths are chosen under each window. With five differential consecutive windows as a group, the best differential window is selected in each group. After the conversion of two sensitive wavelengths in six differential windows, related analysis with apple leaf chlorophyll content is done and two new parameters with the largest correlation coefficient are chosen to establish estimation model. It found that the coefficient of the determination(R2) of estimation model under different derivative windows reduced after the first increase. Five consecutive windows of the different derivative windows within the range of 1-30 are chosen as a group. The best derivative window of each group is elected and the result is that 4, 7,13,17,22, 30 is the best derivative window of each group. The establish estimation model works best when the derivative window is 13. Testing the partial least squares model and the stepwise regression model established under differential window 13, it found that the R2 of the stepwise regression model is bigger than the partial least squares model. The RMSE and RE% of the stepwise regression model is smaller than the partial least squares model, which shows that stepwise regression model is more suitable to estimated chlorophyll content.(4) The study establishes the best estimation model by the hyperspectral data at different phenophases. The experiments were done in 20 orchards in the field, measured hyperspectral data and nitrogen contents of apple leaves at three phenophases in two years, which were shoot growth phenophase, spring shoots pause growth phenophase, autumn shoots pause growth phenophase. The study analyzed the nitrogen contents of apple leaves with its original spectral and first derivative, screened sensitive wavelengths of each phenophase. The hyperspectral parameters were built with the sensitive wavelengths. Multiple stepwise regressions, partial least squares and BP neural network model were adopted in the study. The results showed that 551 nm, 716 nm, 530 nm, 703 nm; 543 nm, 705 nm, 699 nm, 756 nm and 545 nm, 702 nm, 695 nm, 746 nm were sensitive wavelengths of three stages. R551+R716, R551*R716, FDR530+FDR703, FDR530*FDR703; R543+R705, R543*R705, FDR699+FDR756, FDR699*FDR756 and R545+R702, R545*R702, FDR695+FDR746, FDR695*FDR746 were the best hyperspectral parameters of each phenophase. Of all the estimation models, the estimated effect of shoot growth phenophase was better than other two phenophases, so shoot growth phenophase was the best phenophase to estimate the nitrogen contents of apple leaves based on hyperspectral models. In the three models, the 4-3-1 BP neural network model of shoot growth phenophase was the best estimation model. The R2 of estimated value and measured value was 0.6307, RE% was 23.37, RMSE was 0.6274.