Satellite imagery now provides an opportunity to complement traditional field instrumentation.
Monitoring landslide movements provides useful information that can potentially prevent catastrophic loss of life and property and assists in the development of guidelines for sustainable land use planning. Due to the significant time and cost associated with field instrumentation, many areas of high landslide risk are not monitored until damaging movements occur. As a result, our fundamental understanding of landslide triggering mechanisms and our ability to model landslide movements accurately has been hampered by limited field data. Satellite imagery is now providing an opportunity to complement traditional field instrumentation to fill this data gap.
Geotechnical engineering professor Ellen Rathje and graduate student Oscar Suncar recently performed a study using optical image correlation to measure deformations of the Portuguese Bend Landslide (PBL), which moves 1 to 7 m/yr south of Los Angeles. The study aimed to affirm that digital imagery from satellites provides displacement estimates consistent with traditional field measurements but also provides a more detailed picture of the displacement patterns across a landslide.
The PBL is located along the Pacific coast in Rancho Palos Verdes and consists of an elevated block of marine sediment with a core of metamorphic basement rock. The area has a history of landslides going back 250,000 years and the landslide complex was mapped by geologists starting in the 1920s. It was not until 1956 that it was fully investigated due to movements initiated by 20 m of fill placement associated with the construction of a road. The construction fill destabilized the slope, and more than 130 homes were destroyed, resulting in the City of Rancho Palos Verdes being sued.
Because of the critical nature of this landslide, the movements of the PBL have been monitored by the City of Rancho Palos Verdes and the Abalone Cove Landslide Abatement District (ACLAD) for the last 20 years. This monitoring consists of 72 GPS monuments distributed across the landslide complexes, each of which is surveyed every year. These monuments provide valuable information about the movements in and around the landslide.
The intent of Rathje and Suncar’s NASA-ROSES-sponsored study was to compare field data from the GPS stations to data from optical image correlation, which is a powerful remote sensing technique for estimating deformations of slow and rapidly moving landslides. This technique provides deformations at a spatial resolution that generally cannot be obtained with traditional field instruments or other remote sensing techniques.
Optical image correlation analysis consists of identifying shifts of small segments of pixels between two optical images using cross-correlation. For each segment of pixels, an East-West and a North-South shift are estimated along with a measure quality of the correlation. The aerial images must be geometrically corrected and precisely co-aligned before the correlation analysis can be performed. The correlation analysis is performed for hundreds to thousands of pixel segments, providing a detailed picture of the pattern of deformation across the landslide.
For Rathje and Suncar’s study, two 0.50 m resolution images from a satellite were used. The deformations computed from the optical images over the time period of August 31, 2010, to May 29, 2011, were compared with those measured by the GPS stations within the Portuguese Bend Landslide. Yearly GPS measurements from 2009, 2010 and 2011 were used to estimate the expected deformations during the time period.
Approximately 20 GPS stations are located over the ~1 km2 of the PBL. Most of the GPS stations recorded approximately 0.5-m of displacement towards the coast over this 9-month time period, while one station recorded over 6 m of movement near the coast. However, because the GPS stations are spaced at 200-300 m from each other, they do not provide detailed information on how the deformations vary across the entire landslide. The deformations from optical image correlation show the same levels of deformation as the GPS monuments but are spaced about every 30 m, improving the spatial density of measurements by almost an order of magnitude. This improved spatial density provides a clearer pattern of the deformation patterns, in particular showing how the deformations are concentrated within the central and southeast sections of the landslide.
The team concluded that the correlation of high-resolution optical imagery can provide accurate estimates of ground deformations and displacement fields over time. The remote sensing results provide more detail of the landslide’s deformation patterns with a high level of accuracy and reliability. Overall, the study shows that field-based monitoring is complemented by enhanced satellite monitoring.
“We hope that remote sensing becomes a common tool used to monitor landslides and better understand landslide mechanics, such that landslide risks are reduced in the future”, says Rathje.
Rathje has utilized remote sensing for earthquake damage assessment and mapping all over the globe. She was among a team of engineers and scientists commissioned by the United Nations to create the first detailed soil map of Port-au-Prince, Haiti, in the aftermath of the January 2010 earthquake. This map used field measurements of soil stiffness and remote-sensing-based damage estimates to identify the soil conditions that lead to enhanced damage in some parts of the city. Suncar also accompanied Rathje to Haiti on the earthquake reconnaissance surveys performed just weeks after the earthquake.
The focus of Rathje’s research is understanding and predicting geotechnical earthquake hazards, with particular emphasis on the evaluation of earthquake-induced ground failure (slope instability, soil liquefaction), the effect of soil conditions on earthquake ground motions, and the use of remote sensing to document earthquake hazards. She has received many awards for her work, such as the Huber Research Prize from ASCE, and is the Co-Chair of the NSF-sponsored Geotechnical Extreme Events Reconnaissance (GEER) Association.
