Land cover distribution region based on multivariate data

It was very difficult to acquire satisfy classification result only using spectral information and textual information on the broad land of China. This paper divided the land cover regions based on multivariate data to improve classification accuracy which were Pa, BT, DEM, AVHRR NDVI time-series data and IM. Through principal components analysis, the information percent of first three principal components was 97.4% and the IM, Pa, DEM and NDVI were key factors to land cover distribution. There were nine land cover districts in the final result which was the bases to land cover classification in China.