Integrating Global Land Cover Products Based on Thematic and Spatial Accuracy Assessment

Accurate global land cover (GLC) product is citical for climate change and climate models. Different classification schemes and local accuracy are identified as two aspects for the discrepancies among GLC products. We proposed a methodology that combined thematic similarity and local classification accuracy to produced integrated GLC products for 2001-2005,2008-2010, and 2013–2015. The results show that the global characteristics of the land cover distribution are well captured in our integrated products. And, the integrated method successfully improves the accuracy of GLC products, with the overall accuracy of the integrated product increase 0.8%-18.3% validated by our independent ground truth. Additionally, the class-specific accuracy has been improved to varying degrees. The integrated GLC products show high consistency and stability for the three epochs. The proposed method is expected to apply for various specific applications.