Land cover classification system is the basis for urban landscape studies. In recent years

the rapid development of earth observation technique provides the researches of land cover with a great amount of remote sensing images with high resolution. Many scientists have established various types of systems base on the distinguishing features of different remotely sensed data

but none of which can be used pervasively or be widely recognized in the international academia. In this paper

we proposed a classification system based on the sharpened GF-1 data with 2 meter resolution and fit for the urban area. We took Guangzhou city as an example to illustrate the application of this classification system. In this classification system

we focus on the characteristics of urban ecosystem and make it a hierarchy system without overlapping within the classes. The first two layers of the system are solid and classified with certain scientific basis

and the third layer is open for different uses. We divided the first layer into 4 parts

which include: buildup area

vegetation

water and bare land. The second layer combined with residential and commercial area

industrial district

road; forest

bush

grass

paddy field; river

pond etc. The classification results reach the precision of 90.1% and cover all parts of layer two

which is technically practicable.