Assessment of fine-resolution land cover mapping products in the Changbai Mountain Range, China

Land cover data are essential for modelling land surface processes in mountainous regions. Fine-resolution land cover products have provided unprecedented spatial detail of the land surface. The performance of land cover classification can be affected by various causes, propagating uncertainty through modelling. In this study, six openly accessible fine-resolution land cover data products for 2020, namely GlobeLand30, CLCD, GLC_FCS30, Esri land cover, WorldCover, and DynamicWorld, were cross-compared for consistency and accuracy in the Changbai Mountain Range. The study particularly highlighted the consistency and accuracy across various geomorphic types. The findings reveal that all the six data products exhibit an overall similarity in spatial patterns dominated by cropland and forest. Validation using a ground reference dataset revealed that WorldCover and GlobeLand30 provide the highest and lowest accuracy, respectively. Cropland and forest present considerable spatial consistencies among the data products and advanced accuracies. The middle-mountain regions with low landscape diversity and fragmentation showed the highest consistency and accuracy. Data products, however, are difficult to represent the distribution of grassland, wetland, and bare ground categories reliably due to limited consistency and accuracy. This study provides quantitative analysis for selecting appropriate land cover data products towards application implementations.