| Crop growth information reflects the status and trend of crop growth,is an important part of agricultural information,timely understanding of crop distribution overview,growth status,fertilizer and water market and insect pest dynamics,is conducive to crop production managers or management decision-makers to provide timely and accurate data information platform.Unmanned Aerial Vehicle(UAV)remote sensing technology has the characteristics of real-time,timely and flexible monitoring time,and the use of UAV remote sensing data to dynamically monitor regional crop growth has incomparable advantages.In this paper,a fly ash filling area in Huaibei City,Anhui Province was selected as the research area,and the chlorophyll,biomass,plant moisture content and plant height of winter wheat during the overwintering period were used as the research objects,and the multispectral image data obtained by unmanned aerial vehicles were used to invert the growth parameters of winter wheat and monitor the comprehensive growth of winter wheat.The main contents and results are as follows:(1)According to the growth parameters of winter wheat,the response relationship between the growth parameters of winter wheat and the multispectral reflectance of unmanned aerial vehicles was analyzed,the sensitive bands of SPAD,AGB,PWC and VH of winter wheat were determined,and a univariate linear regression model of winter wheat growth parameter index was established.The chlorophyll-sensitive bands determined by the significance of Pearson’s correlation coefficient are blue,red and green,the sensitive bands of plant height and biomass are red and blue,and the sensitive bands of plant moisture content are red.The overall accuracy of the winter wheat growth parameter model established by using sensitive wavelength bands is low,and the model R~2 of the four growth parameters is up to 0.42.And from the RPD values of the SPAD,AGB,VH and PWC optimal models of winter wheat,it can be seen that the RPD values of the four models are lower than 1.4,which belongs to the class A model,and the accuracy of the model is poor,so it is difficult to effectively predict the growth parameters of winter wheat by using the spectral reflectance of the drone.(2)In view of the fact that it is difficult to effectively predict the growth parameters of winter wheat using the sensitive wavelength band of the UAV,22 multispectral vegetation indices are constructed by using the image spectral reflectance feature,and three machine learning algorithms are selected:Multiple Linear Regression(MLR),Random Forest(RF)and Extreme Learning Machine(ELM)to established optimal model of growth parameters of winter wheat.After linear and nonlinear operations,the existing multispectral reflectance can be combined to improve the correlation with the growth parameters of winter wheat,and comparing the 12 models established,it can be seen that the optimal prediction model of chlorophyll is the ELM-SPAD model.Its model R~2 is0.76,RMSE is 1.31,the optimal prediction model for biomass is RF-AGB model,its model R~2 is 0.76,RMSE is 0.21,the optimal prediction model for plant height is MLR-VH model,its model R~2 is 0.87,and RMSE is 0.67,and the optimal prediction model for plant moisture content is ELM-PWC model,its model R~2 is 0.78,and RMSE is 0.01;it can be seen from the RPD values of the model that the four growth parameter models of winter wheat belong to class C models This shows that the multispectral vegetation index greatly improves the accuracy of the model due to the integration of reflectivity information of multiple bands.(3)Since it is difficult to fully reflect the growth of winter wheat by a single growth parameter,this paper uses the basic principle of the coefficient of variation method to construct a comprehensive growth monitoring index(CGMI)of winter wheat.Compared with the MLR,RF and ELM models of winter wheat,the accuracy of the CGMI model is significantly improved compared with the MLR,RF and ELM models of the single growth parameters of winter wheat,and the CGMI model R~2 constructed by MLR is the highest,which is 0.91,and the 3 models belong to the C model,indicating that CGMI can improve the prediction accuracy of the winter wheat growth model.The results of winter wheat growth grading obtained by the confidence interval method show that the proportion of winter wheat with average growth in the whole study area was the largest59.59%,and the winter wheat with poor growth rate accounted for only 0.01%,and the overall growth of winter wheat in the study area was average.Since the study area is a cultivated land filled with fly ash,the growth of winter wheat is bound to be affected by soil.Therefore,GPR data are used to extract different soil cover thickness,and the response relationship between different soil cover thickness and winter wheat CGMI is analyzed.The results showed that with the increase of soil cover thickness,the CGMI value of winter wheat gradually increased and finally stabilized.CGMI reached the maximum value of 0.59 when the thickness of soil cover was 40-50 cm.Figure[23]Table[6]Reference[102]... |
PDF Full Text Request