| Intercropping is an important method for increasing crop acreage and yield under decreased cultivated land conditions. Intercropped soybean has played key role in guaranteeing food security in the region and alleviating contradiction between supply and demand of domestic soybean to a large extent. Hyperspectral remote sensing, as advanced technology in modern agriculture, which could quantitatively reflect crops growth situation and nutrition, has broad research potential. Among the key indicators of soybean’s growth and development, carbon nitrogen metabolism parameters and photosynthetic fluorescence characteristics are used for reflecting soybeans response to the external environment. Hyperspectral remote sensing breaks through the damage of the conventional chemical methods to soybean leaves, and provides data support for rapid and nondestructive monitoring of soybean photosynthetic nutriture, thus plays a certain role to push several aspects including agricultural informatization.Based on a three-year field trial which contains five different spatial allocation under maize-soybean relay intercropping, this paper systematically analyses the quantitative relation between leaves’main parameters and hyperspectrum parameters during soybean V3. R1 and R4 period. The results showed, during the relay intercropping symbiotic period, the transmittance and R/Fr of low crop soybeans’ canopy leaves were decreased with the increasing of narrow-row spacing of maize, and less than the sole soybean canopy light environment obviously. Compared with sole soybean, relay intercropped soybean leaves carbon content and carbon nitrogen ratio in A1 treatment were decreased notably (declining range 1.082%-2.151% and 7.634%～2.151% respectively) while nitrogen content was increased notably (increasing range 7.094%～11.694%), and chlorophyll parameters were decreased notably while fluorescence parameters were first decreased then increased during V3 and R1 period; under different spatial allocation, leaves’main parameters of relay intercropping soybean showed the same change pattern with the increasing of narrow-row spacing of maize. And with the growth stage being put forward, leaves’ chlorophyll parameters, carbon content and carbon nitrogen ratio were increased while nitrogen content was decreased under the same allocation; and combining regression analysis, nitrogen content contributes more than carbon content in carbon nitrogen ratio, and the regression coefficient R2 was 0.979. At the same time, the changing characteristics of soybean leaf spectral reflectance and the first derivative spectrum were the same with leaf nitrogen content on green peak band. And the variation tendency of spectral reflectance in near-infrared band and red edge amplitude were the same with leaves’carbon nitrogen ratio. Compared with relay intercropping soybean in Al treatment, red-shift could be observed in the red edge position of first derivative spectrum in monoculture conditions, the average wavelength from 710 nm to 714 nm at V3 period, the average wavelength from 715 nm to 718 nm during Rl period; also with the increasing of maize’s narrow row distance, blue-shift appears in the red edge position; and putting forward growth stage will lead to continuous red-shift of the red edge position of soybean leaves’spectrum. And there was no significant difference among relay intercropping soybean under different spatial allocation in leaves’spectrum characteristic and main parameters during R4 period, in which maize have been harvested (P>0.05)According to the correlation analysis between carbon nitrogen parameters, chlorophyll fluorescence parameters of soybean leaves and original spectrum, the first derivative spectrum, spectrum characteristic area, vegetation index, spectrum index based on optimum wave band selection and wavelet energy coefficient after continuous wavelet transformation respectively, it could be found that multi-band spectral index of the free combination of the original spectral reflectance, first derivative spectra sensitive information extraction, and the spectral index deal with wavelet functions had a better correlation with main parameters of leaves than other methods. The correlation coefficients of chlorophyll parameters and spectral index based on a combination of multiband were more than 0.9.Linear and non-linear model and partial least squares were used to build the optimal estimate model for soybean leaf main parameters under different spatial allocation. Carbon content and C/N ratio of soybean leaves used D721, SDr, RVI(662, 617), bior3.9(94,893) and D731, GNDVI, DVI(736,713), db5(68,738) respectively as the spectral variables, using partial least squares model had high prediction accuracy, but the best estimate nitrogen content of the model was a quadratic function by the wavelet coefficients sym2 (96,669) construction. The best estimate chlorophyll b content of the model was a quadratic function by the spectrum index DVI (D734, D530), and estimate chlorophyll a and total chlorophyll of the model was obtained using partial least squares by the spectrum index D731, GNDVI, RVI(D7i8, D710), sym2(84,576) and D732, GNDVI, DVI(736,723), bior3.9(132,659). Fluorescence parameters, NPQ, F’q/F’m, and qP respectively used partial least squares estimation obtain optimal parameters of model by the spectrum index D707, SIPI, DVI(428,434), db5(13,858) and D481, RG, RVI(D691, D681), bior3.9(43,653) and D479, RG, RVI(D689, D679), coif2(24,670). The determination coefficients and test values were all relative high, after regression test, it could be found the RPD value and R2 of the model were around 1.665～2.411 and 0.471-0.934 respectively; also compared with other models, its RMSE value was smaller than those of model. Therefore, these results will provide a support of diagnosing the response of soybean to shading conditions by quantitative analysis and monitoring carbon nitrogen parameters and chlorophyll fluorescence parameters based on hyperspectral remote sensing data. |
PDF Full Text Request