Study on Crop Classification Based on Sentinel-2 Time Series Data and Random Forest:A Case Study of Wuhan
This study takes Wuhan city,Hubei province as the research area,takes Sentinel-2 remote sensing images on the PIE-En-gine platform as the basic data source,makes sample datasets based on three open datasets,constructs spectral features,remote sens-ing index features,and NDVI time series feature sets,and selects random forest classifiers to carry out the training and verification of two-category classification of cropland and noncropland and four-category classification of single cropping rice,double cropping rice,other crops and noncropland in Wuhan from 2018 to 2021.The final statistics,comparison and analysis of the results of the two-cate-gory classification and the four-category classification show that the accuracy of the two-category classification is better than the four-category classification,and the overall accuracy is higher than 80%;the correctness of four-category classification is obviously better than that of two-category classification,and the conclusion is based on the results of four-category classification;in Wuhan,single cropping rice is less planted,and double cropping rice is widely planted;the planting area of both single cropping rice and double cropping rice first increased significantly in 2019,then decreased significantly in 2020,and then increased significantly in 2021.
Wuhansample datasetsSentinel-2NDVI time series characteristicsrandom forest
NETL
NSTL
万方数据
