Extraction of Winter Wheat Planting Area in Jining City Based on Sentinel-2 Images
The single vegetation index was often used in remote sensing extraction of winter wheat,but a single vegetation index cannot fully reflect the differences between different vegetation,which can easily lead to missed and incorrect classification,affecting the extraction accuracy.To effectively extract the planting area of winter wheat,this article is based on the PIE Engine remote sensing platform,using Sentinel-2 remote sensing image data,combined with sampling points and original feature bands and exponential features as input features.Random Forest(RF)and Support Vector Machine(SVM)classification methods are used to extract the planting area of winter wheat in Jining City.The extraction accuracy of different classification methods under the same input features and the classification accuracy of different input features under the same classification method are evaluated,and the best classification feature and classification method are ultimately obtained.The results show that using exponential features as input features and combining with random forest classification method to extract the winter wheat planting area in Jining City in 2023 has the best accuracy,with a validation matrix ACC coefficient of 0.984 and a validation matrix Kappa coefficient of 0.974.It can be seen that the random forest index feature model based on Sentinel-2 remote sensing images can accurately extract the winter wheat planting area in Jining City.This study can provide an effective method for extracting the winter wheat planting area,providing auxiliary basis for regulating agricultural production,rational utilization of natural resources,achieving precise agricultural management,and ensuring effective food supply.
国家科技期刊平台
NSTL
万方数据
