MONITORING LANDSLIDES AND VOLCANIC DEFORMATION FROM INSAR 
TECHNIQUES 
V. Singhroy“, H. Ohkura, K. Molch', R. Couture? 
* Canada Centre for Remote Sensing, 588 Booth Street, Ottawa, Canada, K 1A 0Y7 
: vern.singhroy@ccrs.nrcan.gc.ca 
National Research Institute for Earth Sciences and Disaster Prevention, Tennodai 3-1, Tsukuba-shi, Ibaraki-ken, 305-0006 JAPAN 
ohkura@ess.bosai.go.jp 
* MIR Télédétection on assignment at CCRS, 588 Booth Street, Ottawa, Canada, K1A 0Y7 
4 Geological Survey of Canada, 601 Booth Street, Ottawa, Canada K1A OE8 
Commission V11 WG/5 
KEY WORDS: InSAR, Landslide motion, Rockslides, Volcanic deformation 
ABSTRACT: 
Our research has shown that interferometric SAR techniques can be used to monitor deformation at landslides and volcanoes under 
specific conditions. Special filtered interferograms are being used to reduce speckle and minimize errors created by vegetation. 
InSAR images were used to monitor current post slide motion along existing fault lines at the Frank Slide a 30 x 10* m? rock 
avalanche, in the Canadian Rockies. This information is used to understand the post failure mechanism and mobility of the slide. At 
Miyake-Jima Island in Japan a filtered interferogram was used to monitor post eruption volcanic shrinkage. The volcano erupted in 
July-August 2000 that led to the total evacuation of inhabitants from the island. Our differential InSAR results show subsidence at 
the summit after the eruption, and allow us to monitor the post eruption processes, thereby assisting in planning the return of local 
residents to the island. 
INTRODUCTION 
SAR interferometric techniques, which uses images acquired 
with two repeat passes over the same scene, are being used 
increasingly to monitor landslides and volcanic deformation. 
(Murphy and Inkpen, 1996; Singhroy et al., 1998; Singhroy and 
Mattar, 2000: Ohkura and Shimada, 2001; Lundren et al., 2001; 
Feretti et al., 2001). Recent research has shown that differential 
SAR interferometry (DIFSAR) is being used to monitor 
landslide motion under specific conditions (Vietmeier et al., 
1999; Rott et al., 1999; Singhroy and Molch, 2004). Provided 
coherence is maintained, as in non-vegetated areas, surface 
deformation of a few cm over the acquisition period can be 
measured. To avoid large spatial decorrelation, data pairs with 
small perpendicular baselines, and short time intervals between 
acquisitions are being used. In addition, correcting the effect of 
topography using GPS information and detailed topographic 
data are also essential in achieving useful measurements of 
surface deformation. 
MONITORING DEFORMATION 
AT THE FRANK SLIDE 
Our study focused on the Frank Slide, a 30 x 106 m? rockslide- 
avalanche of Paleozoic limestone, which occurred in April 1903 
from the east face of Turtle Mountain in the Crowsnest Pass 
region of southern Alberta, Canada. Seventy fatalities were 
recorded. Several investigations have focussed on post failure 
mechanism and mobility (Couture et al., 1998; Cruden and 
Hungr, 1986). However, a recent 6000 tons rock fall in 2001 
has renewed monitoring programs on the Frank slide by the 
Government of Alberta (Alberta Environment, 2000). Our 
570 
InSAR investigation is aimed at assisting in the location of in- 
situ motion instruments. 
ERS data were to be used for the INSAR analysis. In order to 
select a set of suitable scenes a thorough baseline analysis of all 
ERS-1 and ERS-2 ascending scenes acquired over the location 
(track 406, frame 989) during summer between 1992 and 2001 
was performed. It was of interest to find as many data pairs as 
possible during that time period, yet keep the perpendicular 
baselines below 100 m, thus reducing contributions of 
topography on differential phase values. Ascending orbit was 
chosen for the look direction (right) to correspond with the 
aspect of the slope. We selected an August '95 and August '97 
InSAR pair with a perpendicular baseline of 4.5 m to produce 
the results shown in Figure 1. For all data pairs processed, 
coherence values were generally high on the slide itself, even 
for temporal baselines of more than 700 days. Values e.g. for 
the Aug-95 / Aug-97 pair (736 days) are in the range of 0.73 to 
0.91. Singhroy and Molch (2004) published additional details 
on data selection and InSAR processing. 
Our interferometric images show deformation of 2 cm over the 
2-year period of along parts of the fault line. This suggests that 
the failure along the fault line may trigger secondary slides or 
rockfall and are the target for field instruments. In addition, our 
InSAR deformation maps are providing a more synoptic view 
of active areas on the old slide surface. A combination of the 
synoptic INSAR maps and measurements from field instruments 
will provide a more integrated monitoring system for these 
large active slides along strategic transportation routes, and also 
assist in current investigation on post failure mechanism and 
mobility at the Frank Slide