Accuracy Evaluation of High Resolution Global Land Cover Products: A Case Study of Nine Urban Agglomerations

Large scale land cover data is an important basis for global land surface process research, ecosystem assessment, environmental modeling and other scientific research. Studying the characteristics of existing data sets has guiding significance for data users and the production of new data sets. Taking nine urban agglomerations (Yangtze River Delta Urban Agglomeration, Beijing Tianjin Hebei Urban Agglomeration, Guangdong-Hong Kong-Macao Greater Bay Area, Northwest Urban Agglomeration, Japan’s Pacific Coastal Urban Agglomerations, Greater London Urban Agglomeration, West Coast Urban Agglomeration, Atlantic Coast Urban Agglomeration and Great Lakes city Urban Agglomeration) as the study area, this study analyzed the spatial consistency of four land cover datasets, namely, GlobeLand30, FROM-GLC, GLCFCS30 and CCI-LC and evaluated the accuracy of the four land cover datasets using Google Earth high-resolution images. The results show that the spatial distribution patterns of the same land cover features in different land cover data sets are quite different, and the overall consistency coefficient among different land cover data sets is low; FROM-GLC has the highest overall precision and kappa coefficient, while CCI-LC has the lowest overall precision and kappa coefficient.