Land use and land cover change is one of the focuses in the study of global change in recent years. Land cover characteristics, at the global and regional scale, are very important for evaluating global environment status and simulating global environment scenario in the future. A special issue named "Global and regional land cover characterization from remotely sensed data" was published in 2000 in International Journal of Remote Sensing. Based on this, international advance of land cover characterization from remotely sensed data at global and regional scale was summarized and introduced in this paper, including data sources, dimension and land cover mapping methods (pre-processing, classification and accuracy assessment), etc. Two global land cover databases-IGBP DISover and University of Maryland 1 km products were introduced as examples.
The domain of land cover mapping spans the range between two extremes: "coarse" resolution at frequent time intervals, and "fine" resolution at long intervals. Major steps in extracting land cover information by using satellite data at fine and coarse resolutions include: geometric corrections, compositing, radiometric corrections, classification and accuracy assessment. Land cover information that can be gleaned from satellite images is the spectral and spatial attributes of individual cover types. Two types of numerical classification approaches for satellite image classification have been evolved for more than 30 years, which are unsupervised algorithm and supervised algorithm. In recent years, numerous variants of these two basic classification methods have been developed, including decision trees, neural networks, fuzzy classification and mixture modeling for supervised classification, and classification by progressive generalization, classification through enhancement and post-processing adjustment for unsupervised techniques. "No land cover classification project would be complete without an accuracy assessment". In addition to purely methodological considerations, accuracy assessment tends to be strongly constrained by the resources available. Thus, in practice, accuracy assessment is "A balance between what is statistically sound and what is practically attainable must be found".
On board of EOS-AM-1 satellite, MODIS sensor acquires data with 250 m，500 m，1000 m spatial resolution and 36 bands spanning 0.4~14 μm. MODIS data, free for users all over the world, is up to date and is becoming the most important data source for land cover classification at the global and regional scale.