The rapid development of economy and explosive growth of population have caused large decrease of water resources and vegetation coverage in the earth so that soil exhaustion

land desertification and salination become worse and worse

especially in arid to semi-arid area. It is very difficult to carry out the truth ground inventory and monitoring using the routine methods due to large and wide terrain and poor natural environments. Remote sensing technology has the ability of rapid information acquisition of large land expansion

and will supply useful information for monitoring the surface features in these regions. In recent years

imaging radar interferometric technique with its all weather

day and night capabilities

can generate the digital elevation model （DEM） and monitor surface change using amplitude and phase information from radar signal. So it has become a potential tool to acquire more resource and environmental information. The repeat-pass interferometry acquires two images by using one antenna for repeat passes over the same area at two different times. The two images can be used for further information extraction only while they have somewhat coherence.Arid to semi-arid landscapes are characterized by its sparse vegetation cover and lower soil moisture

and make the scattering origination in the radar image mainly from the surface. The coherence of the radar data acquired by repeat-pass interferometry is generally high

even for a long acquisition time intervals. However

the geometric changes from the surface can be generated with the time. These changes may originate from vegetation growth and erosion caused by wind or water. Any change can be detected by the decrease of coherence

hence the decorrelation of interferometric data can supply the basis for surface change monitoring.This paper presented the results of discrimination and classification of surface land types in Kashi test site

Xinjiang Province of northwestern China using the repeat-pass interferometric data

acquired by European Resource Satellite 1 and 2

based on the interferometric coherence estimation. Six types of land were discriminated and classified

including bare soil

salina

bush

bare rock/Gobi

marsh and water body. Then the backscatter and coherence characteristics of these land types were analyzed

and the relationship between coherence and surface features in arid and semi-arid area was also discussed. Finally

the temporal decorrelation model was made

and will supply the theoretical support for monitoring surface change in arid and semi-arid area using interferometric data.