Abstract
Rapid and accurate mapping of winter wheat using remote sensing technology is essential for ensuring food security. Most of the existing studies have failed to fully characterize the phenological features of winter wheat in mapping, resulting in low classification accuracy. To this end, this study developed a new multiple phenological spectral feature (Mpsf) and then used the generated new features as input data for a one-class classifier (One-Class Support Vector Machine, OCSVM) to map winter wheat. The main steps in this work are as follows: (1) Identifying key phenological periods. The spectral indices temporal profiles of winter wheat (after cloud masking) were drawn separately using different spectral indices, and the key phenological periods of winter wheat were identified with a priori knowledge of phenology. (2) Composition for a new feature. Composited the spectral features of winter wheat for each key phenological period to generate a new feature. (3) Training using a one-class classifier. The new feature was put into OCSVM for training, and the final winter wheat mapping result in the Beijing region was obtained. The cost of this new winter wheat mapping method is low and the accuracy is high. To verify the accuracy of this study, we compared the Mpsf map with three kinds of reference data, and all of them got good results. In comparison, with ground truth samples from Sentinel-2, the total accuracy was overall higher than 97.9%. The relative error of the 2019 winter wheat mapping result was only 0.51%, compared with the data from the Beijing Bureau of Statistics. In comparison, with an up-to-date available winter wheat-mapping product for Beijing (spatial resolution: 30 m), the Mpsf map has significantly fewer misclassifications. To our knowledge, this study produced one of the highest accuracy winter wheat-mapping products in Beijing for 2018 and 2019 to date. In general, we hope that this work can promote the development of winter wheat mapping and provide a reference for sustainable agricultural development and governmental decision-making.
Funder
National Natural Science Foundation of China
Beijing Outstanding Young Scientist Program
Subject
General Earth and Planetary Sciences
Cited by
1 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献