Precision analysis and validation of multi-sources landcover products derived from remote sensing in China

Precise state and regional land cover mapping are foundational data in the studies of land surface processes and modeling, such as climate change, biodiversity conservation, ecosystem assessment, and environmental modeling. This study aimed to perform the validation and comparison of the four global land cover datasets, and examine the suitability and accuracy of different coarse spatial resolution datasets in mapping land cover categories across China. The results indicated that the four land cover products demonstrated the consistency in whole scale, but in local scale, display a conspicuous difference especially in southeast and southwest zones. GLC2000 had the best consistency in the relative index and overall accuracy, which were 0.92 and 55.86% respectively; the GLOBCOVER had the worst consistency compared to the reference data , the relative index and overall accuracy were 44.71% and 0.31. The four land cover products showed the spatial confusion especially in forest, shrub, cropland and grassland, which implied the coarse resolution remote sensing data was helpless in discerning the land cover with the leaf type. The result will provide a reasonable reference for using these datas.