Terrain Change Detection Using ASTER Optical Satellite Imagery Along the Kunlun fault, Tibet

Terrain changes are manifested in satellite images as pixel offsets, which represent the apparent difference in the position of corresponding pixels in two time-separated images of the same portion of the Earth's surface. We present terrain change detection results using Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) imagery. The change detection methods employed are Fourier analysis, image window cross-correlation ("imageodesy"), and a wavelet analysis. These methods were applied to the detection of seismic displacement due to the November 14, 2001 Kokoxilli earthquake. During this Ms = 8.1 event, the Kunlun fault experienced a maximum of 16.3 m of displacement. Three ASTER image pairs along the Kunlun fault were used, each with different time separation windows. Each time separation window represents a different time interval, from two years to two months. Our results show that terrain change detection was most successful using the Fourier and wavelet techniques. The Fourier analysis detected left-lateral slip of 4.5 m along a fault oriented 270° in the proximal area of the Kunlun fault. These results are in good agreement with field observations. Wavelet analysis was very effective at delineating fault scarps caused by the 2001 Kokoxili earthquake.