Abstract
Global land cover data are necessary for global environmental monitoring and research studies. To classify land cover in Asia, first, ground truth (GT) data were collected as the training data. In the authors' previous study paddy in south China were misclassified into evergreen broadleaf and needleleaf forests, because paddy GT data were not collected in south China but in Japan. In this study, paddy GT data in China were newly collected from 1-km-grid China land use data, which were prepared by Chinese Academy of Science. Next, maximum likelihood method was applied to monthly Normarized Difference Vegetation Index (NDVI) data, which were calculated from 1-km-grid Systèm Pour 1'Observation de la Terre (SPOT) /VEGETATION data in 2000. Finally, land cover was classified into 19 items and it was found that paddy was correctly classified in south China (correct ratio : 100% on GT sites) .
