基于GEE云平台和Sentinel⁃2数据的普洱市森林覆盖制图

Remote Sensing based Land Cover Classification of Pu′er City Using GEE Cloud Platform and Sentinel-2 Data

快速准确获取森林的空间分布对评估森林资源和生态环境状况具有重要的意义。以云南省普洱市为研究区,基于Google Earth Engine(GEE)平台和Sentinel-2影像数据,结合实地调查数据、机载遥感数据及地形辅助数据,提取影像的光谱特征、纹理特征以及地形特征,通过特征筛选,得到适合森林分类的最优特征数据集。结合简单线性非迭代聚类(Simple Non-Iterative Clustering,SNIC)超像素分割算法,探究不同分类方法、特征变量等因素对分类精度的影响。结果表明:面向对象分类方法的分类精度要优于基于像元分类方法,分类总体精度为88.21%,Kappa系数为0.87,可以较为准确地对普洱市进行森林覆盖制图。面向对象方法可以有效减轻“椒盐现象”,特征优选避免了冗余信息对分类结果的影响,有效提高了分类效率。GEE平台与面向对象方法结合可以提供大区域、高精度的森林覆盖遥感快速制图。

Quick and accurate access to the spatial distribution of forests is of great significance for assessing the status of forest resources and ecological environment protection.Taking Pu'er City in Yunnan Province as the research area,Based on the Google Earth Engine(GEE)platform and Sentinel-2 image data,combined with the field survey data,airborne remote sensing data and terrain auxiliary data,the spectral features,texture features and topographic features were extracted.Through feature screening,the optimal feature set suitable for forest classification was obtained.Combining Simple Non-Iterative Clustering(SNIC)superpixel segmentation algorithmto explore the influence of different classification methods and characteristic variables on the classification accuracy.The results showed that the classification accuracy of the object-oriented classification method was higher than that of the pixel-based classification method,with an overall classification accuracy of 88.21%and the Kappa coefficient of 0.87.which can accurately map the forest cover of Pu'er City.The object-oriented method can effectively alleviate the“salt and pepper phenomenon”,and feature optimization avoids the influence of redundant information on classification results and effectively improves classification efficiency.The combination of GEE platform and object-oriented method can provide large-area,high-precision forest cover remote sensing rapid mapping.