Evaluation of six global high-resolution global land cover products over China

ABSTRACT

Land cover is an important variable for climate, hydrology, and ecology studies. With the availability of various high-resolution global land cover (GLC) products, conducting a comprehensive assessment on their accuracy and consistency is important. In this study, we compared the performance of three latest 10-m-resolution GLC products, which include FROMGLC10 in 2017, ESA’s Worldcover10 in 2020, and ESRIGLC10 in 2020, and three latest 30-m-resolution GLC products, which include FROMGLC30 in 2017, GLC_FCS30 in 2020, and Globeland30 in 2020, in China. The consistency of these products was investigated in terms of spatial consistency and area consistency. Though the six GLC products demonstrate similar overall distribution patterns, their detailed spatial distributions are quite different, especially for the three 10-m-resolution products. Evidently, the cropland, forest, grassland, and bareland exhibited high inconsistencies than the other types. The classification accuracy of the six GLC products was also quantitatively assessed based on a visual-interpretation-based reference dataset. FROMGLC10 exhibits the highest overall accuracy of 65.57%, followed by FROMGLC30 (64.96%) and Worldcover10 (62.74%). ESRIGLC10 (49.79%) exhibits the lowest accuracy. The accuracies of shrubland, wetland, and tundra were relatively low. This study provides a valuable reference for selecting appropriate GLC products for potential users.

KEYWORDS:

• Land cover products
• high-resolution
• spatial consistency
• accuracy evaluation
• China

SUBJECT CLASSIFICATION CODES:

• Land cover
• remote sensing
• global data bases

Acknowledgements

The authors would like to thank Tsinghua University, European Space Agency, Environmental Systems Research Institute, Inc, Chinese Academy of Sciences and National Geomatics Center of China for providing GLC data. We would like to appreciate the Google Earth Pro for providing the platform and high-resolution images to select sample points and do visual-interpretation. The authors would also like to thank reviewers for their valuable comments on the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from the author, Y.W., upon reasonable request.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China under Grant 41871028, 42271351 and 42171101.