Abstract: Remote sensing data of land cover is the basic information for largescale landscape pattern research. Land-cover data from different research teams adopt different data sources and classification methods, which leads to inconsistent results of landscape pattern research. Quantitative analysis of the consistency of land-cover data and its impacts on landscape pattern research is of great significance. Here, with two sets of global scale 30m resolution landcover remote sensing data GlobeLand30 and FROM-GLC as the research objects, we compared their differences in soil and water conservation areas in Beijing-Tianji-Hebei Region from two aspects of landcover type and landscape pattern index. The results showed that: (1) In the mountainous and hilly water conservation and soil conservation region in the eastern part of Taihang Mountains, theagreement in area size between the two datasets was the highest, with a consistency value of 96.8%. By contrast, the consistency was the lowest in the mountainous and hilly sand control and water conservation region in the northwestern Taihang Mountains, with a value of 3.6%; (2) The spatial consistency of those two datasets in the Jin-Yi-Lu Bohai Bay ecological maintenance area was the highest, with a spatial consistency value of 73.6%. In the mountainous and hilly area of the northwestern Taihang Mountains, the spatial consistency of sand control and water conservation area was the lowest, with a value of 25.17%; (3) In terms of landscape pattern, the complexity of GlobeLand30 dataset in the six ecological functional areas of BeijingTianjinHebei was lower than that of the FROM-GLC dataset, indicating that diversity of landscape in GlobeLand30 dataset was lower, and homogeneity of surface landscape categories was stronger, while surface landscape heterogeneity and fragmentation degree of FROM-GLC dataset were stronger.

Key words: GlobeLand30 data, FROM-GLC data, consistency, landscape pattern.