Li Zhiwen, Chen Xingyu, Qi Jie, Xu Chong, An Jiafu, Chen Jiandong
School of Economics and Management, Southwest Petroleum University, Chengdu, 610500, Sichuan, People's Republic of China.
School of Public Administration, Southwestern University of Finance and Economics, Chengdu, 611130, Sichuan, People's Republic of China.
Sci Rep. 2023 Aug 9;13(1):12936. doi: 10.1038/s41598-023-39963-0.
The accuracy assessment of land cover data is of significant value to accurately monitor and objectively reproduce spatio-temporal dynamic changes to land surface landscapes. In this study, the interpretation and applicability of CCI, MCD, and CGLS long time-series land cover data products for China were evaluated via consistency analysis and a confusion matrix system using NLUD-C periodic products as reference data. The results showed that CGLS had the highest overall accuracy, Kappa coefficient, and area consistency in the continuous time-series evaluation, followed by MCD, whereas CCI had the worst performance. For the accuracy assessment of subdivided land cover types, the three products could accurately describe the distribution of forest land in China with a high recognition level, but their recognition ability for water body and construction land was poor. Among the other types, CCI could better identify cropland, MCD for grassland, and CGLS for unused land. Based on these evaluation results and characteristics of the data products, we developed suitable selection schemes for users with different requirements.
土地覆盖数据的精度评估对于准确监测和客观再现地表景观的时空动态变化具有重要价值。在本研究中，以NLUD-C定期产品作为参考数据，通过一致性分析和混淆矩阵系统，对CCI、MCD和CGLS长时间序列土地覆盖数据产品在中国的解译能力和适用性进行了评估。结果表明，在连续时间序列评估中，CGLS的总体精度、Kappa系数和面积一致性最高，其次是MCD，而CCI的表现最差。对于细分土地覆盖类型的精度评估，这三种产品能够以较高的识别水平准确描述中国林地的分布，但它们对水体和建设用地的识别能力较差。在其他类型中，CCI对耕地的识别能力较好，MCD对草地的识别能力较好，CGLS对未利用地的识别能力较好。基于这些评估结果和数据产品的特点，我们为不同需求的用户制定了合适的选择方案。
