Spatial scaling transformation of data from National Geographical Census

In order to realize the cross-validation of different scale of land cover classification in the National Geographical Census,10 test sites were selected in the paper,including Zhengzhou and Xi'an.The spatial multi-scale characteristics of different land cover types on landscape metrics and accuracy loses were analyzed,and taking the forest class as example,the spatial scaling transformation model was constructed within the spatial resolution range of 6~2 000 m,which had the estimate precision of 92.63%.Result showed that the precision loss would occur once the spatial scale of land cover data is changed,and the spatial model established by the resolution and landscape metrics could effectively correct the error caused by changing spatial scales,which would provide a reference for the quality control for the National Geographical Census and Monitoring.