Study On Remote Sensing Estimation Method Of Vegetation Coverage In Winter Wheat Summer Maize Planting Area

Accurate estimation of vegetation coverage is of great significance for monitoring crop growth and providing effective information for food security.Wheat and corn are mainly planted in the northern field,both of which are important food crops in China.With the wide application of remote sensing technology in the estimation of vegetation coverage in large areas,there is a lack of continuous monitoring of winter wheat and summer maize rotation areas.Winter wheat and summer maize rotation areas are mainly distributed in humid and semi-humid areas.The time period is long and the climate is changeable.A single optical image cannot be used for remote sensing dynamic monitoring of rotation areas.In view of the above problems,this paper chooses the combination of microwave and optics.The method of estimating the vegetation coverage of Sentinel-1 and Sentinel-2 satellites was studied under clear and cloudy conditions.Under clear sky conditions,Sentinel-2 satellite image and UAV image are combined to study the method of Sentinel-2 satellite image to extract vegetation coverage with higher accuracy by using pixel dichotomy,threshold substitution method and spectral conversion method.Under the condition of cloud and rain,based on microwave data,the vegetation coverage of microwave method was explored from two aspects:the selection of sensitive bands and the construction of estimation model,and applied in the local rotation area of the eastern piedmont plain of Mount Tai.The main results are as follows:(1)The study of sensitive parameters shows that the sensitivity of polarization parameters to summer maize coverage is higher than that of backscattering coefficient to summer maize coverage.The sensitive parameters of polarization parameters are different in different growth stages.Under the linear relationship,the most sensitive parameter of summer maize at jointing stage and tasseling stage is C11,and the most sensitive parameter at bell stage is C12＿imag.(2)The results of model construction show that Sentinel-2 satellite images combined with UAV data can improve the estimation accuracy of winter wheat and summer maize coverage under clear sky conditions.The coverage estimated by the normalized difference vegetation index NDVI and the ratio vegetation index RVI combined with the threshold substitution method,the single-band conversion method,and the vegetation index conversion method is compared with the accuracy of the coverage of winter wheat and summer maize estimated by the traditional pixel dichotomy method.It is concluded that the vegetation index conversion method combined with the UAV image has the best effect.The R~2of winter wheat is 0.9372,the MAE is 0.0514,and the RMSE is 0.0685.The R~2 of summer maize is 0.9259,the RMSE is0.0680,and the MAE is 0.0511.The multiple stepwise regression model constructed by the polarization parameters of the Sentinel-1 satellite image at the same time phase is better than the coverage model constructed by the backscattering coefficient.Under the same time phase and different spatial variations,the summer maize coverage model constructed by the BP neural network method is better than the multiple stepwise regression method.The R~2 of jointing stage,bell stage and tasseling stage were 0.6653,0.5929 and 0.7985,respectively.The R~2 of bell stage and jointing stage,tasseling stage and bell stage,tasseling stage and jointing stage were0.6847,0.8187 and 0.7830,respectively.The model has good estimation accuracy.(3)The vegetation index method and BP neural network method combined with Sentinel satellite data can be used for continuous monitoring of winter wheat and summer maize in the rotation area of the eastern piedmont plain of Taishan.The coverage of winter wheat and summer maize in the eastern region is higher than that in the western region.In this paper,through the combination of UAV and Sentinel-1 and Sentinel-2 satellite images,the coverage estimation of winter wheat and summer maize experimental areas was carried out from two conditions of clear sky and cloud rain.The vegetation index method for improving the accuracy of vegetation coverage estimation of Sentinel-2 satellite images and the BP neural network method for vegetation coverage estimation of Sentinel-1 satellite images were selected.The combination of the two methods and Sentinel satellite image data can be used for continuous monitoring of winter wheat and summer maize rotation areas.