Estimation Of Wheat Yield Based On WOFOST Model And UAV Image Assimilation

At present,in the context of "Internet+"agriculture,integrating information technology with modern agriculture to achieve intelligent and efficient agricultural production,cultivation and management is an inevitable trend in the development of modern agriculture in China.Wheat is one of the most widely distributed and highest-yielding grain crops in China.Due to its topographical conditions and climatic conditions,Jiangsu Province has become the main planting area for winter wheat in southern China.Therefore,improving the accuracy of winter wheat growth monitoring in Jiangsu Province is of great significance for the effective estimation of winter wheat.In this study,the growth and development of winter wheat was simulated by WOFOST crop model,and the assimilation of UAV image data and crop model was carried out by using the least squares optimization algorithm.Finally,combined with the actual measurement of winter wheat LAI and yield,the alalysis and evaluation of assimilation results provide the basis for agricultural estimation.The main research contents and results are as follows:(1)In order to make the model simulation conform to the growth law of winter wheat in the study area,this study used the meteorological data,soil data,crop data,etc.of the regional meteorological site in 2015-2017,and used the OAT method to analyze the sensitivity of the model parameters,combined with the minimum.The second-pass,"trial and error,method and based on the previous research results,based on different density and nitrogen fertilizer treatment levels,for winter wheat development parameters TSUM1(the accumulated temperature from emergence to flowering),TSUM2(flowering to mature accumulated temperature)and growth parameters SLATB(ratio Leaf area),AMAXTB(maximum CO2 assimilation rate)for winter wheat parameter adjustment,localization of WOFOST model.The results showed that the WOFOST model could better simulate the growth and development of winter wheat in the study area.The R2,RMSE and NRMSE(%)of the winter wheat LAI were 0.8178,0.58 and 27.9,respectively,and the simulated biomass R2 and RMSE(kg·hm-2),NRMSE(%)were 0.7832?0.9531,315.55?986.15,10.1?29.8,respectively.The R2,RMSE(kg·hm-2)and NRMSE(%)of simulated yield were 0.5852,799.96 and 15.9,respectively.(2)In order to construct the best UAV inversion LAI model,this study selects 11 color characteristic indexes constructed by R,G,B;from the whole growth period,planting density,nitrogen application rates,and various growth stages.Correlation analysis between color index and winter wheat LAI was carried out,and a significant correlation color index was selected to construct the inversion model.The results showed that the model construction was the best before the jointing stage,jointing stage,flowering stage and filling stage.The model construction R2 was 0.795,0.784;0.746,0.625,respectively.The model verification R2 was 0.781,0.807,0.718,0.697,RMSE.They are 0.325,0.470,0.364,and 0.256,respectively.(3)Based on the localization of the WOFOST model,the least squares optimization algorithm is applied to model the assimilation based on the inversion of the LAI data by the drone,and the sensitive parameter specific leaf area(SLATB0,SLATB0.5,SLATB2)is adjusted and corrected.And the maximum CO2 assimilation rate(AMAXTB1,AMAXTB1.3),the model simulation LAI and UAV data inversion LAI error is minimal.The results showed that the assimilated model could better evaluate the growth and development of winter wheat in the study area.The assimilated WOFOST model simulated the R2,RMSE and NRMSE(%)of winter wheat LAI were 0.8812,0.49,23.5,respectively,and R2 and RMSE of simulated yield.(kg·hm-2)and NRMSE(%)were 0.9489,327.06,and 6.5,respectively.The accuracy of model simulation of winter wheat was improved,which proved the feasibility of assimilation of crop model and UAV data.