Research On Monitoring Method Of Different Soybean Populations Growth Based On Chlorophyll Concentration And Leaf Area Index

In this paper,the experiment was conducted in Xinxiang City,Henan Province,and the Chlorophyll Concentration and Leaf Area Index(LAI)were taken as the research object which belong to Recombinant Inbred Lines(RIL)and natural wild soybean populations canopy leaf.In order to pass the inversion of the research object provide a method for monitoring the growth of field crops,with the support of field measured data,we using the PROSAIL model and BP neural network optimized by genetic algorithm carry out inversion research on the chlorophyll concentration and LAI of the canopy leaves of different soybean populations.The main work and conclusions of this research are as follows:(1)Through the analysis of the spectrum and physical and chemical characteristics of the canopy leaves of different populations,it can be seen that in the visible light band,the RIL soybean population has the highest reflectance in the R3 period and the lowest in the R6 period.The reflectance of the natural wild soybean population is the highest in the R7 period and the lowest in the R3 period.In the near-infrared band,the reflectance of the two soybean populations are the highest in the R4 period,and the lowest in the R1 and R7 periods.The chlorophyll concentration and LAI have the high similarity,they reached the highest level in R3,R4,R5 and R6 periods,and the lowest in R1 and R7 periods during the growth periods,which provided a certain auxiliary reference for establishing the correlation between them.(2)There is a strong correlation between chlorophyll concentration and LAI and canopy spectrum.Among them,chlorophyll concentration has a greater impact in the blue,green and red bands of visible light.The largest positive and negative correlations of RIL soybean population are 0.20 and-0.54,respectively,and the maximum positive and negative correlations are 0.11 and-0.45 in the natural wild soybean population.While the LAI of the two soybean populations have a greater impact in the near-infrared band.The maximum positive correlations are both greater than 0.8,and the maximum negative correlations are both less than-0.6,which provided a certain theoretical basis for the sensitive wavelength input of the BP neural network and the local sensitivity analysis of the PROSAIL model in the follow-up work.(3)Through the construction of the BP neural network model optimized by the genetic algorithm and PROSAIL model,the study found that the optimal number of nodes in the hidden layer of the chlorophyll concentration and the LAI inversion models in the RIL soybean population are 10 and 11,respectively,and the optimal number of nodes are10 and 9 in the natural wild soybean population’s hidden layer.In the local sensitivity analysis of the PROSAIL model,the chlorophyll concentration and LAI are more sensitive to the sensitive band in the visible light,followed by the leaf structure parameters and dry matter content,and the LAI has a significant influence on the spectrum in the near-infrared band,and the average sensitivity is around 60%.(4)The comparative analysis of the models found that the PROSAIL model has a better simulation effect in the canopy spectrum of the RIL soybean population during the whole growth periods and the typical growth periods,and the BP neural network model optimized by the genetic algorithm has a better simulation effect in the natural wild soybean population’s canopy spectrum.The two models have the highest similarity between the simulated spectrum and the measured spectrum in the R5 period,and the simulation effect is the best than others in the entire typical growth periods.(5)By linearly fitting the predicted value of the model with the measured value,it is found that in the inversion of chlorophyll concentration,the maximum R~2 of the PROSAIL model are 0.87 and 0.88,and the RMSE are 1.62 and 0.60,which in the comprehensive inversion of the whole growth periods and the typical growth periods of the RIL soybean population.The maximum R~2 of the BP neural network model optimized by the algorithm model are 0.85 and 0.87,and the RMSE are 1.73 and 0.65,which in the comprehensive inversion of the whole growth periods and the typical growth periods of the natural wild soybean population.While the applicability of the LAI inversion model in different soybean populations is the same as that of the chlorophyll inversion model,the R~2 of the predicted value is closer to the measured,the RMSE of the LAI inversion model is smaller and the estimation accuracy is higher than the former model.In the R5 period,the performance of the model reaches the best level,the R~2 and RMSE of the PROSAIL model in the RIL soybean population are 0.89and 0.11,and the R~2 and RMSE of the BP neural network model optimized by the genetic algorithm in the natural wild soybean population are 0.85 and 0.13.It shows that the LAI inversion model is more accurate,and the error is smaller between the predicted value and the measured.Through the inversion of the LAI in the R5 period,the growth of soybeans can be more accurately reflected.