基于高分辨率人工识别地表类型的GlobeLand30产品精度评价  

Precision Evaluation of GlobeLand30 Products Based on High-resolution Artificial Identification Land Type

作　　者：马延慈 明艳芳[1] 王凯 苏庆华[1,2] 

机构地区：[1]山东科技大学测绘科学与工程学院,山东青岛266590 [2]曲阜师范大学地理与旅游学院,山东日照276826

出　　处：《山东科技大学学报（自然科学版）》2018年第5期1-10,26,共11页Journal of Shandong University of Science and Technology(Natural Science)

基　　金：国家自然科学基金项目(41771408);山东科技大学研究生科技创新基金项目(SDKDYC170209)

摘　　要：我国国家地理基础信息中心2014年推出的全球首套30 m分辨率地表覆盖遥感制图数据产品(GlobeLand30)在土地资源利用、地理国情监测、气候变化研究等领域具有重要作用。为评价该产品精度,采用2m分辨率人工识别土地类型产品,选取两个典型研究区域,通过空间统计分析、空间一致性分析和空间混淆分析手段,结合高程等因素对GlobeLand30数据产品中的各地表覆盖类型进行精度评价,综合分析影响精度的原因。研究表明:产品总体一致性稳定在70%左右,各地类判别精度具有较大差异,影响精度的因素主要在于地形地貌特征与光谱相似性质。obeLand30, the world’s first set of 30 m resolution surface coverage remote sensing mapping data product released by National Geomatics Center of China in 2014, plays an important role in the fields of land resource utilization, geographic condition monitoring and climate change research. In order to evaluate the accuracy of this product, two typical study areas were selected and the 2 m high-resolution artificial identification of land type products was used to analyze the precision of the GlobeLand30 data products through spatial statistical analysis, spatial consistency analysis, and spatial alienation analysis in combination with elevation. In this way, the factors influencing precision were analyzed synthetically. The results show that the overall consistency of the product keeps about 70%, and the discrimination precision is quite different among different land types. The factors influencing precision mainly lie in the topographic and geomorphological features and spectral similarities.

关 键 词：土地覆被产品 精度评价 人工识别土地类型产品 GlobeLand30 

分 类 号：P237[天文地球—摄影测量与遥感]