Study On Inversion Model Of Farmland Film Residues Based On Multispectral Unmanned Aerial Vehicle Data

There are some problems in using manual field collection of farmland plastic film residue,such as large labor consumption,long time and error in manual collection of plastic film residue.In order to obtain the residual film amount data conveniently,low cost and short cycle,a method of obtaining the residual film amount of farmland in the later harvest stage by using multispectral UAV was proposed in this study.Taking Yuli County in southern Xinjiang and Manas County in Northern Xinjiang as the study area,10 test fields were selected for data collection.Three sample boxes were set in each test field,and the data of surface residual film amount and overall residual film amount,i.e.0-30 cm residual film amount,were obtained manually.Then,the multispectral data of each test field were obtained by flying at three altitudes of 25 m,50 m and 100 m by multispectral unmanned aircraft,Then,the correlation analysis between multispectral data and surface residual film is carried out to obtain the best spectral characteristic parameters for estimating farmland surface residual film and the inversion model of surface residual film.Then,the correlation between surface residual film and overall residual film is analyzed to finally obtain the correlation and inversion model between the best spectral characteristic parameters and overall residual film.The research results are as follows:(1)It is concluded that when the flight altitude of UAV is 25 m,it is the best flight altitude to identify the amount of residual film on the farmland surface.(2)The spectral characteristic parameter GNDVI can identify the amount of farmland surface residual film in Manas County and Yuli County.The inversion model of surface residual film in Manas County is:y=-1.589x~2-20.632x+8.2245,R~2 has the highest accuracy,reaching 0.4319.The inversion model of surface residual film in Yuli County is y=458.4x~2-48.451x+2.5993,and R~2 has the highest accuracy,reaching0.5752.The spectral characteristic parameter OSAVIcan identify the amount of farmland surface residual film combined by Manas County and Yuli County,y=-83.684x~2+40.143x+0.9133,and the highest R~2accuracy is 0.5056.(3)The spectral characteristic parameter GNDVI can be used to retrieve the total amount of residual film in farmland.The model used in Manas County is y=-156.43x~2+19.631x+10.991,R~2 is 0.5199,the model used in Yuli County is y=572.58x~2-34.78x+1.9396,R~2 is 0.5207,and the spectral characteristic parameter OSAVI can be used to retrieve the total amount of residual film in farmland combined in Manas County and Yuli County is y=-346.21x~2+111.61x+1.427,with R~2 accuracy of 0.5251.(4)GNDVI can identify the amount of farmland surface residual film and total residual film in Manas County and Yuli County respectively,and OSAVI can identify the combined amount of surface residual film and total residual film in Manas County and Yuli County.Therefore,it is feasible to use UAV to monitor farmland residual film.