new approach for mapping regional land cover and the application of this approach in Australia

Detailed and accurate land cover data are widely used for various purposes, such as global land use change detection. This study aimed to investigate Australian land use and cover change by using 774 Landsat scenes in 2000 and 2010. The reference data included pictures or high-resolution images from Google Earth, global land cover data, Shuttle Radar Topography Mission data, and other literature. Australian land cover was automatically classified into nine main classes, namely cropland, forest, grassland, shrubs, wetland, water, artificial covering, bare land, and others, by using a global land cover decision tree developed for computer-based image classification. Obvious mistakes were subsequently modified manually. Finally, the classifications were found to have overall accuracies of 93.24% and 92.79% evaluated by using 2820 and 3820 sample points, respectively.