Roessner, Sigrid 
  
this tectonically active area landslides are caused by complex interactions between endogenic and exogenic factors 
including tectonic structures, lithology, relief and climate. 
Since the Kashgarta landslide is only one prominent example of a large number of landslides occurring in a 50km wide 
belt along the eastern rim of the Fergana Basin, the goal is to develop effective satellite remote sensing based methods 
for an inventory of present landslide activity (last 10 years) and for a regionalisation of landslide-causing factors. The 
main emphasis is put on determination of relevant geological, tectonic and geomorphological indicators for landslide 
activity using GIS-based visual and quantitative analysis. For this purpose a spatial database consisting of topographic 
maps (1:100.000), MOMS-2P based DEM, MOMS-2P orthoimages, geometrically fitting Landsat-TM and SAR (ERS- 
1/2) data, geological information and field investigations has been established (Roessner et al., 1999). In the following, 
results of GIS-based analysis are discussed for the Kashgarta landslide. 
In the Upper Maili Suu river basin the dominance of geological structures initiating landslide processes requires 
detailed 3-D analysis of structural elements developed by the interaction between tectonic (endogenic) and mass 
wasting (exogenic) processes. For this area the derived MOMS-2P 25m-DEM and orthoimage are the first digital 
topographic database of sufficient spatial resolution allowing detailed GIS-based analysis by perspective visualizations, 
surface profiling, derivation of quantitative parameters (e.g., slope) and their analytical combination (Fig. 2). For 
structural investigations perspective DEM visualizations were overlaid by RGB-orthoimages and geological map 
information. They allowed identification of structural and tectonic features, such as fracture or shear elements and faults 
forming the basis for genetic interpretation of structural elements influencing slope failure (Wetzel et al., 2000). In the 
area of the Kashgarta landslide shear zones could be identified which cut the older N-E anticline. Within these shear 
zones young extension has developed showing a close spatial relationship to the landslides. Based on these observations 
it is assumed that the landslide was tectonically initiated leading to a primary displacement of the weakly consolidated 
Upper Cretaceous sediments (area above Profile A in Fig. 2). A more detailed analysis of landslide related tectonic 
structures can be found in Wetzel et al. (2000). The results represent one of the inputs for future GIS-based factor 
analysis of landslide hazard. 
Besides regional analysis of larger structures, high resolution MOMS-2P stereo products also allow detailed analysis of 
a single landslide which is demonstrated in Fig. 2 showing a perspective visualization of the DEM overlaid by the 
orthoimage and the courses of the profiles A, B, C. The profiles are the result of a combined GIS analysis including 
DEM, orthoimage and geological map information. Profiling the DEM overlaid by the orthoimage allowed precise 
differentiation between displaced masses (high albedo due to missing vegetation) and stable (vegetated) parts of the 
profile. This change is also reflected in the topographic profile in the form of small morphological steps (profile A). 
Slope discontinuities within displaced masses indicate secondary movements. Integration of geological information led 
to the final cross-sections of Fig. 2. In the middle part of the valley (Profile A) Mid-Quaternary (Q;) loesses were 
partially displaced by the landslide. In contrast, slopes within Tertiary units (Pc and N;/?) situated further downstream 
are stable and only the bottom of the valley is filled by displaced masses (Profile B). Profile C shows a cross-section 
which has not yet been affected by accumulation. Additional characteristics of the landslide could be obtained from 
slope classification. Areas of landslide deposition are dominated by slopes between 0 and 9 degrees reaching maximum 
values between 10 and 14 degrees in the area above Profile A. The main scarp of the landslide can clearly be identified 
at its upper rim by high slope values between 20 and 34 degrees. Stable flanks (Profiles B, C) are characterized by 
slopes greater than 15 degrees. 
6 CONCLUSIONS 
GIS-based satellite remote sensing in mountainous terrain requires digital topographic information (DEM) of a spatial 
resolution comparable to the satellite data. In Kyrgyzstan stereo satellite data are the only possible source for effective 
generation of such a DEM for a larger area. Stereo data obtained by MOMS represent an ideal opportunity due to 
threefold along-track stereoscopic imaging technology. The previously developed approach for stereo processing of 
MOMS-2P 3-line imagery was successfully applied to three scenes (60-62) of MOMS-2P data take T0906 covering the 
transitional zone between Fergana Basin and Tien Shan which is intensely affected by landslides. In the result a state 
coordinate based DEM of 50m raster size could be generated for 150 by 60km large area. The achieved absolute height 
accuracy of 1.1 &/- 17.3m represents a good quality DEM taking into account the limited number of control points 
which could be determined in the field by geodetic GPS measurements. In this process highest uncertainty was 
introduced during identification of corresponding locations for GCP in the image data and in the field which is caused 
by the combination of a complicated field situation (lack of larger man-made structures) and the limited spatial 
resolution of the MOMS-2P data. 
  
1264 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.