Evaluation and analysis of upscaling of different land use/land cover products (FORM-GLC30, GLC_FCS30, CCI_LC, MCD12Q1 and CNLUCC): a case study in China

Abstract

A rich of global land use/land cover (LULC) products exist, such as DISCover, FORM-GLC30, CCI_LC, MCD12Q1, etc. These products play an extremely important role in the study of the earth’s biochemical cycles and climate change simulations. However, these LULC products vary in spatial and temporal scale. Scale conversion is one of the effective means to solve this difference, which can realize the spatial matching between different LULC products and complete the effective integration of LULC products with other geographic element products. Due to data acquisition limitations, this study adopted China as a case study for examining the accuracy of upscaling of five LULC products, FORM-GLC30, GLC_FCS30, CCI_LC, MCD12Q1, and CNLUC. The original resolution of FORM-GLC30 and GLC_FCS30 is 30 m, and that of CCI_LC, MCD12Q1, and CNLUCC is 300 m, 500 m, and 1 km, respectively. There are four different upscaling methods of grid center, maximum area, maximum aggregation, and nearest neighbor to amplify the resolution to 1 ∼ 8 km. Within the upscaling experiment, using area change index and Shannon index to assess accuracy of LULC Product upscaling. The results showed that: (1) the rank of the upscaling methods according to upscaling accuracy was nearest neighbor > grid center > maximum area > maximum aggregation. (2) The most suitable scales for upscaling of FORM-GLC30, GLC_FCS30, CCI_LC, MCD12Q1, and CNLUCC are 2 km, 2 km∼6 km, 2 km∼5 km, 2 km, and 3 km∼5 km, respectively. (3) The CNLUCC dataset in China was shown to be suitable for conducting relevant long-term time-series research due to its long time span, high data accuracy, and relatively higher accuracy when upscaled.