基于GEE平台结合RF和SVM算法的茶园提取研究

RF and SVM Classification Algorithms of Tea Plantations Extraction Based on GEE platform

科学有效的茶园遥感监测技术为土地利用管理、茶园管理、产业政策制定奠定了基础。研究基于GEE平台利用Landsat 8/OLI分别结合SVM和RF两种分类算法对云南省普洱市和西双版纳自治州的茶园进行了提取,并将2种算法的提取结果进行了对比。结果表明,RF和SVM算法的分类总体精度(OA)分别为95.61%、95.56%,Kappa系数相同,均为87%。RF算法的分类总体精度高于SVM算法,而Kappa系数相同。RF算法的制图精度(PA)为65.44%,与SVM算法(65.75%)相比相差较小,RF和SVM算法的用户精度(UA)分别为63.08%和57.37%。基于GEE平台结合RF分类算法对茶园的提取精度高于SVM算法。采用不同的传感器以及不同的分类算法可实现对茶园准确、高效的提取,对制定科学的茶园管理政策、茶园灾害预警、地表覆被变化等研究具有重要意义。

Scientific and effective tea plantation remote sensing monitoring technology lays the foundation for land use management, tea plantation management and industrial policy formulation. Based on GEE platform, tea plantations in Pu′er City and Xishuangbanna Autonomous Prefecture of Yunnan Province were extracted by Landsat 8/OLI combined with support vector machine(SVM) and random forest(RF) classification methods respectively. The extraction results of the two classifiers were compared. The results showed that overall accuracy(OA) of RF and SVM were 95.61% and 95.56%, respectively, while Kappa coefficients were 87%. OA of RF was higher than that of SVM, but Kappa coefficients were the same. The producer accuracy(PA) of RF was 65.44%, which had small difference with SVM(65.75%). The user′s accuracy(UA) of tea plantations extraction by RF and SVM were 63.08% and 57.37%, respectively. Using different sensors and different classification algorithms can achieve accurate and efficient extraction of tea plantations, which is of great significance to formulation of scientific management policies, early warning of disasters, and research on land cover change in tea plantations.