Study On Detection Methods Of Physiological And Ecological Parameters Of Sugar Beet Based On Hyperspectral Imaging Technology

The acquisition of crop physiological and ecological information is one of the key and core issues in precision agriculture.Due to the destructiveness and the poor timeliness and effectiveness of traditional methods,it can't meet the real-time and non-destructive dynamic detection of crop physiological and ecological parameters.However,the rapid and non-destructive detection technology for the physiological and ecological information of sugar beet provides the technical support for the refined and scientific management in the production process of sugar beet.It has great significance for the development of green agriculture and the realization of high yield and sugar content in sugar beet.The study took sugar beet as the research object,and the field experiments campaigns in 2014,2015 and 2018 were conducted at three different growth stages(the rapid growth stage of leaf cluster,sugar growth stage and sugar accumulation stage),across different sites in Inner Mongolia(the sugar beet planting area of Taipingdi town in Chifeng,the teaching farm of Inner Mongolia Agricultural University and Maheli village of Tumote zuo banner in Hohhot),for different cultivars(KWS1676,KWS9147 and KWS1231),nitrogen(N)application rates(0-200 kg/hm²)and planting patterns(transplanting and direct seeding).The canopy ground-based hyperspectral data was collected used hyperspectral imager.Meanwhile,the physiological and ecological information(leaf nitrogen content,SPAD value and plant above-ground biomass)of sugar beet was tested with biochemical analysis.The rapid non-destructive detection methods for the physiological and ecological parameters of sugar beet at canopy scale using hyperspectral imaging technology were explored in this study.The main research contents and conclusions are as follows:(1)By analyzing the changes of nitrogen content,chlorophyll content and above-ground biomass(AGB)of sugar beet and the corresponding change trend of the spectral response curve at canopy level during the growth process of sugar beet,the optical detection mechanism for physiological and ecological parameters of sugar beet was explored.The optimal nitrogen application rates for each experimental area were determined,which was 108kg/hm²for the sugar beet planting area of Taipingdi town in Chifeng,120kg/hm²for the teaching farm of Inner Mongolia Agricultural University and130 kg/hm²for the Maheli village of Tumote zuo banner in Hohhot.The results verified the fitness of hyperspectral imaging technology in the assessment for physiological and ecological parameters of sugar beet.(2)Based on the meticulous sampling method,the normalized spectral index(NDSI)and the soil regulation spectral index(SASI)with all possible band combinations in the range of 390-990 nm were constructed,and the vegetation canopy adjustment parameter(L)of SASI was optimized using the particle swarm optimization algorithm(PSO).By analyzing the correlation between NDSI and SASI and the canopy nitrogen content of sugar beet respectively,as well as the influence of different modeling algorithms on the model for the canopy nitrogen content of sugar beet,the best spectral index and estimation model for each growth stage were determined.The performance of proposed spectral index,SASI1(R₄₁₈,R₆₈₆),SASI2(R₈₂₀,R₆₅₅)and SASI3(R₈₇₄,R₈₈₉),were the best for the prediction accuracy of canopy nitrogen content in sugar beet for the validation set with the coefficient of determination(R²)of 0.82,0.74and 0.80,the root mean square error(RMSE)of 2.30,2.71 and 2.26 g/kg,and the relative root mean square error(RRMSE)of 7.11%,10.21%and 8.75%,for the rapid growth stage of leaf cluster,sugar growth stage and sugar accumulation stage,respectively.(3)The competitive adaptive reweighted sampling(CARS)algorithm was applied to select the most sensitive wavelengths to AGB.This was followed by developing a novel modified differential evolution grey wolf optimization algorithm(MDE-GWO)by introducing differential evolution algorithm(DE)and dynamic non-linear convergence factor to grey wolf optimization algorithm(GWO)to optimize the parameters C and?of a support vector machine(SVM)model for the prediction of AGB.The prediction performance of SVM models under the three GWO,DE-GWO and MDE-GWO optimization methods for CARS selected wavelengths and whole spectral data was examined.The best prediction accuracy for the prediction of AGB in sugar beet was achieved by the SVM model optimized by the MDE-GWO algoriithm with the most sensitive wavelengths,independent of growing stage,years,sites and cultivars.(4)By analyzing the correlation between the existing chlorophyll estimated spectral indexes and the canopy chlorophyll content of sugar beet and the construction methods of above indexes,the modified chlorophyll index(MCI)was proposed by introducing the adjustable parameter M into the traditional chlorophyll index(CI).The prediction accuracy of the model for canopy chlorophyll content of sugar beet with different input parameters using partial least squares(PLS)and MDE-GWO-SVM algorithms was examined.MCI was found to be the best spectral index for estimating the canopy chlorophyll content of sugar beet in each growth stage.The performance of proposed spectral index,MCI(R₇₄₇,R₈₃₉),MCI(R₈₆₁,R₈₈₄)and MCI(R₉₃₁,R₇₇₀),were best for the prediction accuracy of SPAD value in sugar beet for the validation set with the R²of 0.85,0.73 and 0.79,RMSE of2.20,2.97and 2.55,and RRMSE of 4.62%,5.78%and 5.28%,for the rapid growth stage of leaf cluster,sugar growth stage and sugar accumulation stage,respectively.(5)With hyperspectral image data,the canopy nitrogen content and chlorophyll content of sugar beet corresponding to each pixel in the hyperspectral image using the optimal spectral index were calculated for each growth stage.Combined with the optimal estimation model,as the spectral signals under all pixels in the canopy leaves were converted into the nitrogen content and chlorophyll content at corresponding positions,the visual distributions of the canopy nitrogen content and chlorophyll content of sugar beet in each growth stage are realized respectively.It directly shows the changes of canopy nitrogen content and chlorophyll content at time and space level during the entire growth process of sugar beet.