SYNERGETIC USE OF MODIS, ASTER, AND LANDSAT DATA FOR LAND COVER CLASSIFICATION IN ARID AND SEMI-ARID AREA OF NORTH CHINA

摘要

The Moderate Resolution Imaging Spectroradiometer (MODIS) Enhanced Vegetation Index (EVI) product has several excellences, which is effective to land cover mapping. In this study, we use 17-dimensional 16-day interval MODIS EVI data with a spatial resolution of 500m and fine-resolution (Landsat, ASTER) data to generate the land cover classification map in north China. The MODIS EVI data are used to mapping land cover map and the fine-resolution data are used to select reference data. The National Land Use/Cover dataset (NLCD-2000), the vector digital map of China located in this study area was referenced. We use the random algorithm randomized each category of reference data. In each category, 2/3 pixels are used for study, and 1/3 pixels are used for test data. The self-organizing map (SOM) neural network method was adopted for the EVI time-series data classification. The classification accuracy and kappa coefficient are 86% and 0.84. The result showed that synergetic use of the time-series MODIS EVI and fine-resolution remote sensing data can be used to produce timely and detailed land cover map in arid and semi-arid region with limited ancillary data.