Monitoring Fast-Moving Landslide in the Three Gorges Area By Offset Tracking Method with High-Resolution SAR Data

Landslide has long been considered as a major geological hazard with great threats to local residents and social economic developments within the Three Gorges area, China. With the regular operation of Three Gorges Dam since 2009, the reservoir water level undergoes an annual cycle from 145 m in early summer to 175 m in early winter caused by storage and discharge operations. Consequently, many ancient landslides were activated by such large variations of water level. One of typical examples is the famous Shuping landslide located in Zigui County.

To reduce the risks of landslide collapse, long-term monitoring of ground surface deformation must be carried out. Terrestrial surveillance like GPS and extensometer have been installed at Shuping in 2003. Although these methods can achieve highly accurate displacement measurements, only sparse points can be observed, which is far from the requirement of monitoring vast reservoir area coverage.

For such a vast and inaccessible area, D-InSAR and PS-InSAR can measure long-term and continuous displacement trends. Nevertheless, it has been revealed by several studies that InSAR can hardly retrieve accurate deformation signal on fast-moving landslides such as Shuping.

Therefore, we further investigated the feasibility and effectiveness of pixel offset tracking (POT) method to detect fast-moving with high-resolution SAR data. One of the key points is to estimate displacement at each pixel through subpixel-level SAR image matching. In order to tackle the difficulties of using traditional POT technique, the point-like target offset tracking (PTOT) is developed. By making use of point-like targets with stable backscattering behavior over long time span, PTOT method is can achieve more reliable results with accuracies at 1/20 of SAR pixel resolution. More importantly, PTOT method can measure two-dimensional displacements, i.e. in both azimuth and slant range direction, while InSAR can measure displacement only along the line-of-sight direction. The experiment at Shuping landslide is carried out with TerraSAR-X datasets acquired in both Stripmap (SM) and High-resolution Spotlight (HS) mode. Preliminary results suggest that both SM and HS data can be used to measure the large displacement, while HS data measurements seems more reliable due to higher resolution than SM data.