Subsections

Monitoring Slow Moving Landslides in the Berkeley Hills with TerraSAR-X Data

Ling Lei and Roland Bürgmann

Introduction

Resolving the kinematics of landslides is a prerequisite for improving our understanding of the mechanics of these potentially hazardous features. We need to better understand how landslides destabilize during large rainstorms and seismic events. In the Berkeley Hills there are four large, slow moving, deep-seated landslides. All the landslides extend through residential areas and move on the order of cm/year, each covering an area of roughly 0.25-1.00$km^{2}$. These slides are located in a rapidly uplifting zone adjacent to the Hayward fault (Figure 2.13). A lot of damage to homes, breakage of underground utility pipes, and confusion over property lines was caused by landslides over the years, although deformation on these landslides is typically small and slow. It is currently not well understood how the landslides respond to seismic activity on the Hayward fault. DInSAR (Differential Interferometric Synthetic Aperture Radar) is a powerful tool for measuring movements of ground by exploiting the phase difference of SAR images taken at different time instances. In this project, we aim to monitor the Berkeley Hills landslides using SAR data from a number of different satellites, especially TerraSAR-X.

TerraSAR-X Data Set

The TerraSAR-X satellite, launched on June 15, 2007, carries an X-band SAR operating at 9.65 GHz. One of the main goals of the TerraSAR-X mission is to produce high resolution imagery with near optical quality. The antenna on TerraSAR-X can be steered in both elevation and azimuth and can be used to generate Spotlight, Stripmap and ScanSAR images. Spotlight images have a resolution of about 1 meter, while Stripmap data have a ground resolution between 2 meters and 6 meters, and ScanSAR images have a resolution of approximately 16 meters. The satellite operates in an 11 day repeat orbit at an altitude of 514 km. The rather short orbit repeat cycle and the electronically steerable antenna allow fast and frequent imaging of a particular site. The frequent interferometric coverage can help especially with monitoring events on shorter time scales. Fast events can be detected and atmospheric delay errors can be reduced by averaging many interferograms.

Figure 2.13: Map of active landslides of the Berkeley Hills (http://www.akropp.com/resources)
\begin{figure}
\centering
\epsfig{file=lei10_1_1.eps, width=5.5cm }
\setlength{\abovecaptionskip}{1pt}
\setlength{\belowcaptionskip}{0.5pt}
\end{figure}

Up to now, we have ordered and received 90 TerraSAR-X Spotlight Single Look Complex (SLC) images and a few Stripmap SLC images delivered by DLR. The TerraSAR-X images were acquired over the active landslides, coastal subsidence and shallow Hayward fault creep near the city of Berkeley. The data acquisition interval is from November, 2008 till now. Four types of Spotlight images and one type of Stripmap images in time sequence were ordered and acquired: spot_012, spot_038, spot_049, spot_075 and strip_003, with HH polarization, different look angles, and different pass directions.

Figure 2.14: Acquisition dates and perpendicular baseline
\begin{figure*}
\centering
\epsfig{file=lei10_1_2.eps, width=12cm}
\setlength{\abovecaptionskip}{1pt}
\setlength{\belowcaptionskip}{0.5pt}
\end{figure*}

Figure 2.15: Mean LOS velocity of the landslides area and Berkeley Marina
\begin{figure*}
\centering
\epsfig{file=lei10_1_3.eps, width=9cm}
\setlength{\abovecaptionskip}{1pt}
\setlength{\belowcaptionskip}{0.5pt}
\end{figure*}

We use ROI_PAC to do the Interferometric processing. We used a two-pass differential interferometry approach using SRTM 1-arc-second DEM heights as a reference to calculate the topographic phase. From the analysis of a map of phase coherence values, we find high coherence values in urban areas but low values in the hills and more vegetated areas which caused phase unwrapping problems in the landslides area. Thus, a time series of interferograms was used for the atmospheric correction. We used sequential interferograms to do the stacking, which means that every image appears in two interferograms except the first and last image. We obtained preliminary results from Stripmap interferograms and standard stacking of Spotlight interferograms. The results are generally consistent with southwest motion of the landslides.

Persistent Scatterer Processing

While standard InSAR measurements that rely on one or a stack of several interferograms can resolve the motion of large landslides, this method is still often hampered by significant noise introduced by atmospheric delays and by loss of coherence in vegetated or high-relief terrain. The persistent scatterer approach can enhance the ability to find suitable scatterers in relatively low-coherence terrains. We use Stanford Method for Persistent Scatterers (StaMPS) which was developed at Stanford University by Dr. Andy Hooper (Hooper et al., 2004). Since we do not have many Strimap acquisitions yet, we only used 12 scenes, which span the time interval from May 2009 to April 2010, for our analysis. We constructed eleven interferograms relative to the master scene of May 10, 2010 with Doris (Delft Object-oriented Radar Interferometric Software). All of these images were used to identify persistent and coherent pixels. Figure 2.14 shows the acquisition date and baseline information. The wrapped phase of the PS pixels is selected and the improved phase unwrapping algorithm is adopted. The negative mean Line-of-sight (LOS) velocity in the landslides area (Figure 2.15) is consistent with landslide motion, moving away from the satellite. Also, the negative LOS velocity reveals subsidence of the Berkeley marina area. For our future work, we hope the four beams of TerraSAR-X Spotlight data from different viewing geometries will significantly improve our ability to fully characterize the kinematics and temporal patterns of the landslides. We are still in the early stages of this investigation and will acquire more data to do the PS processing. Results from TerraSAR data will be carefully compared and integrated with InSAR data from other spacecraft, including the ERS-1/2, Envisat, RADARSAT-1, and ALOS satellites.

Acknowledgements

We thank the German Aerospace Centre (DLR) for providing TerraSAR-X data for this project. We thank Paul Lundgren, Eric J. Fielding, and Paul Rosen for providing TerraSAR reading codes and beneficial discussions. We thank Andy Hooper for supporting StaMPS software.

References

Hilley, G., R. Bürgmann, A. Ferretti, F. Novali and F. Rocca, Dynamic of slow-moving landslides from permanent scatterer analysis, Science, 304, 1952-1955, 2004. TerraSAR-X Ground Segment Basic Product Specification Document, TX-GS-DD-3302. Bürgmann, R., E. Fielding, and J. Sukhatme, Slip along the Hayward fault, California, estimated from space-based synthetic aperture radar interferometry, Geology, 26, 559-562, 1998. Eineder, M., N. Adam, and R. Bamler, Spaceborne Spotlight SAR Interferometry With TerraSAR-X, IEEE Trans. Geosci. Remote Sense., 5, 1524-1535, 2009. Hooper, A., H. Zebker, P. Segall and B. Kampes, A new method for measuring deformation on volcanoes and other natural terrains using InSAR persistent scatterers, Geophys, 31, 2004.

Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94720-4760
Questions or comments? Send e-mail: www@seismo.berkeley.edu
© 2007, The Regents of the University of California