| Maize is one of the main food and cash crops in China.Maize growth and development monitoring and yield simulation forecast are of great significance to ensure food security and develop the national economy,so it is necessary to improve the accuracy,timeliness and regional improvement of maize monitoring and forecast.Crop model has strong mechanical and temporal continuity for the simulation of crop growth and development and yield formation,while UAV remote sensing technology has better spatial continuity and immediacy for crop growth detection.To improve the effect and accuracy of crop simulation,based on WOFOST(World Food Studies)model and UAV remote sensing data,this paper simulated the growth index and yield of summer maize on ecological unmanned farm of Shandong University of Technology,Linzi District,Zibo City,Shandong Province.At the same time,to achieve the ecological agriculture goal of precise fertilization,weight loss and efficiency improvement in ecological unmanned farms and reduce agricultural pollution,different fertilizer application treatments were set in this paper to improve the robustness of crop models to simulate summer maize under different fertilizer application conditions in the study area,and to provide data guidance for the determination of reasonable and efficient fertilizer application,high precision and scientific farmland management measures.In this study,the vegetation index with a high correlation with the LAI of summer maize in the study area was first selected through correlation analysis,and then the LAI inversion model was constructed to determine the optimal LAI inversion model under each treatment.Secondly,Markov Monte Carlo(MCMC)parameter optimization method was used to optimize and adjust the parameters with high sensitivity of the WOFOST model,determine the optimal parameter set,realize the localization study of the WOFOST model,and on this basis,simulate the LAI,biomass and yield of summer maize under different fertilizer application rates.Finally,the ensemble Kalman filter algorithm was used to conduct assimilation research on the multi-spectral remote sensing inversion LAI of UAV and the simulation LAI of the WOFOST model,to improve the simulation accuracy of the model.Based on the assimilation model,the influences of different fertilizer application rates on the growth index and yield of summer maize were analyzed.The main work and conclusions of this study are as follows.(1)Through the correlation analysis of vegetation index obtained from multi-spectral remote sensing data of UAV and LAI of summer maize in the study area,the results showed that OSAVI,NDVI and RVI had a high correlation,and the correlation index was higher than0.8,and the correlation was significant.The linear regression method was used to construct LAI inversion models of summer maize from the trumpet stage to the maturity stage under different treatments,and the model accuracy was verified.The results showed that,under the four different treatments,the inversion precision of the multiple linear regression model(MLR)was higher and the error was smaller.The MLR model predicted that the R~2 of LAI under different treatments was higher than 0.8,RMSE was about 0.6,and R was higher than 0.9.In conclusion,MLR models under different treatments can quickly and accurately invert the LAI of summer maize,providing data support for subsequent assimilation with models.(2)To improve the adaptability of the WOFOST model to summer corn simulation in this study area,11 parameters with high sensitivity of the WOFOST model to summer corn simulation were determined based on the current research.The WOFOST model was calibrated using the Markov Monte Carlo parameter optimization method based on the meteorological data,soil data and CK treatment inter-field test data in this study area,and the growth,development and yield of summer maize under different fertilizer treatments of N1,N2 and N3 was simulated.The results showed that the WOFOST model was effective in simulating the growth,development and yield of summer maize under different nitrogen application treatments in the study area.The NSE of stem dry weight,ear dry weight,above-ground dry matter weight and LAI of summer maize simulated by the model were about 0.8 or above,and the highest was 0.992.The simulation accuracy of leaf dry weight was low,but NSE was higher than 0.5.In conclusion,the WOFOST model can be used to simulate summer maize growth in this study area.(3)To improve the simulation accuracy of the model on the growth,development and yield formation of summer maize,the ensemble Kalman filter algorithm(En KF)was used to assimilate LAI inversion of UAV multi-spectral remote sensing data and LAI simulation of WOFOST model.The results showed that assimilation effectively improved the simulation accuracy of summer maize in the study area.The RMSE of simulated and measured values of leaf dry weight,stem dry weight,ear dry weight,above-ground dry matter weight and LAI of summer maize under N1,N2 and N3 nitrogen application treatments decreased by 8.371,8.214,27.911,16.585 and 0.015,respectively,compared with those before assimilation.NSE was increased by 0.029,0.005,0.004,0.055 and 0.019,respectively,and the RE of N1,N2 and N3 yields were decreased by 1.243%,1.757% and 0.189%,respectively.After a comprehensive analysis,the assimilated model can more accurately simulate the growth and development of summer maize and yield formation in the study area.(4)The assimilation model was used to simulate summer maize under different nitrogen application levels to analyze the effects of different nitrogen application levels on the growth,development and yield of summer maize.The results showed that LAI,biomass and yield of summer maize were N2,N1,N3 and CK from high to low under different treatments,which would increase with the increase of nitrogen application rate.However,when the nitrogen application rate reached a certain level,LAI,biomass and yield of summer maize were reduced due to nitrogen stress due to the high nitrogen application rate. |
