Publications

Qian, XY; Zhang, L (2022). An integration method to improve the quality of global land cover. ADVANCES IN SPACE RESEARCH, 69(3), 1427-1438.

Abstract
Land cover plays a major role in global climate change and biogeochemical cycles. It is widely used in ecological and Earth-system models to simulate atmosphere-biosphere exchanges of energy, water, and carbon. However, the current global land cover (GLC) products lack data consistencies in land cover classes, spatial scale, temporal extent, and the image source and have low accuracy. This study aimed to create a time-series GLC dataset with high accuracy by integrating the current multiple GLC products and correcting the spatial and temporal errors that did not conform to reasonable rules. A logistic regression indicator method was proposed to integrate CCI2012, MODIS2012, and GLCNMO2013 for spatial optimization and integration. As a result, the GLC2012 was generated with the user's accuracy of 74.4%, which was 3.4-10.9% higher than that of the above products. The GLC2012 product was sequentially superimposed by the yearly changes of the CCI-LC product to form a time-series dataset from 1999 to 2018. Then, an abnormal temporal change detection and reconstruction model (CDR) was developed to detect and correct data sequences. The pixels with any changes from 1999 to 2018 were divided into three types of real, possible, and pseudo change. The possible change was required to determine fault or truth and pseudo change was defined as a false change. The spectral similarity method was proposed to correct the false change class. Finally, a 2000-2017 improved GLC annual time-series dataset (IGLC) with a spatial resolution of 0.01 degrees was produced. Hence, the production of a set of high-precision, time-continuous GLC datasets is of great significance for monitoring global ecological environmental protection and urban development, and for proposing corresponding policies and recommendations. (C) 2021 COSPAR. Published by Elsevier B.V. All rights reserved.

DOI:
10.1016/j.asr.2021.11.002

ISSN:
1879-1948