For climatological and hydrological investigations

the areas covered by snow and their spatial variability are important parameters

particularly in alpine regions. Snow and ice also play important interactive roles in the regional climates through its effect of increasing surface albedo. The interferometric SAR technique not only can produce a high resolution digital elevation models

but also can detect the changes in the surface over the orbit repeat cycle from the correlation properties of the radar echoes. With or without the snow cover

the measurement of interferometer correlation will be shown great difference over

which time describes processes occurring on the time scales of the orbit repeat time and size scales on the order of a radar wavelength. Except backscattering intensities in each scene and their changes between two passes

the coherence measurement between repeat pass will provide another useful measurement for mapping snow covers over large areas. We compare four ERS 1 repeat pass SAR image in the Tibet plateau test area

the first one in March（with snow）

the second one in May（without snow）

the rest two is a tendem in September （without snow）. The result shows that if the ground is completely undisturbed between viewings the signals will be highly correlated. Otherwise

the decorrelation will occur. Lake and snow cover have very low coherence between two data tades with and without snow cover

and the coherence measurements from the bare surface and the short vegetation are significantly higher. On other way

the entire target may show very low coherence if there are great decorrelations due to other parameters

such as spatial baseline

rotation and temporal. We will demonstrate the method and result for snow mapping by using both backscattering and cohe rence measurements with repeat passes ERS 1 image data at the Kunlunshan Mountain

the Tibetan plateau（36°03’N

91°00’E）. For validation of this method

we compared classification result with that derived from TM image. An accuracy of better than 82% can be achieved if we consider the classification result from TM image as the ground truth. The results show that the coherence measurements provide an effective way to map snow covered area.