Extraction Of Oat Planting Area In County Based On Multi-source Remote Sensing Data

Remote sensing can quickly obtain ground information through long-distance detection,and use remote sensing to monitor and collect farmland information in a large range and all-weather,providing favorable information support for food supply,prediction,early warning and so on.The rapid acquisition of crop planting area and spatial distribution is the basis and premise of growth monitoring,pest prediction,yield estimation and other agricultural monitoring.In this paper,sentinel-2 and landsat8 are used as data sources to construct a time-series image data set.Combined with vegetation index,spatial features and texture features,decision tree method is used to extract multiple features of oat area in Youyu County,and the accuracy of crop classification and oat area extraction under different feature combinations is compared and analyzed(1)using the multi-source remote sensing images from different sensors to construct the time series data sets,compared with the simultaneous interpreting of single time images,the multi temporal images were more easily to distinguish the crops with similar growth period.The 4 time phases in June 3rd,July 30 th and August 6th were selected as the best time to extract the planting surface of oats.(2)NDVI is the best vegetation index to extract oat planting area,and the difference of NDVI in different time phases is more conducive to distinguish different types of ground objects;the multispectral image of growth period has a lot of texture information,and the combination of vegetation index and texture information can improve the accuracy of information classification;the selection of features depends on crops and planting environment,and the simple superposition of multiple features It can not improve the classification accuracy.(3)The classification results of different feature combinations are different.Using single NDVI feature to build decision tree,the classification accuracy is 78.22%,the area accuracy of oat extraction is 91.90%,the classification accuracy is 79.49% combined with texture features and spatial features,and the area accuracy of oat is 94.35% The error of samples will cause the spatial matching degree between crop classification results and actual planting situation to decrease,and the position accuracy of oat extraction is only 68.74%.(4)According to the spatial distribution map of crop classification in Youyu County,oats are widely distributed.Except for yangqianhe Township,dingjiayao Township in the northwest and niuxinbao Township in the northeast,oats are widely distributed in other townships,and the planting structure is diversified.Oats are intercropped with other crops.