«LANDSLIDE DEFORMATION CHARACTER INFERRED FROM TERRESTRIAL LASER SCANNER DATA A DISSERTATION SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF ...»
LANDSLIDE DEFORMATION CHARACTER INFERRED FROM TERRESTRIAL
LASER SCANNER DATA
A DISSERTATION SUBMITTED TO THE GRADUATE DIVISION OF THE
UNIVERSITY OF HAWAIʻI AT MĀNOA IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHYIN
GEOLOGY AND GEOPHYSICSDECEMBER 2013 By Arjun Aryal
Benjamin A. Brooks, Chairperson James Foster Janet Becker Mark E. Reid Geno R. Pawlak Keywords: Terrestrial Laser Scanner, Cleveland Corral landslide, Particle Image Velocimetry (PIV), Iterative Closest Point (ICP), Elastic Dislocation, Balanced Cross-section, Slip Surface
Mark not only shared me the valuable field observations he made, but also provided very insightful comments about the research.
I would also like to thank Gerald Bowden from USGS who shared me his 2005-2007 TLS data from Cleveland Corral landslide I used here in my first chapter. Ben and Mark acquired 2010TLS data. I would also like to thank my co-authors Ben, Mark, Geno and Gerald for their insightful and valuable comments.
All the faculties and staff in the department of Geology and Geophysics were always supportive.
In particular, Neil Frazer and Steve Martel were always available when I knocked their door with some questions. GG friends Jonathan, Silke, Maxim, Sam, Adrian, Brian, Nancy, Asdis and others were always helpful. It has always been pleasure working with PGF team Ben, James, Jonathan, Todd, Jon and Austin. I would like to thank GG and HIGP departments and people Pete, Greg, Chip, Grace, Susan, Evelyn and Leona.
Last but not least, I would not be able to complete this dissertation without the continuous support and sacrifice from my family. I had good and bad times during this journey but my wife Mandira was always with me. She has been instrumental in instilling confidence in me. My 3.5 years old daughter Manasi has mostly been with me when writing this dissertation and she always fueled me with some energy. Finally, my late father Minanath, who never attended a formal school but he was an educator and he sent me school envisioning a better life for me. My special thanks go to my 82 year-old mother who can not read or write anything in any language but she always encouraged me to go to school since my early days.
ABSTRACT Landslides are ubiquitous and cause thousands of deaths and injuries each year. Achieving a better understanding of landslide stability and governing processes requires good knowledge of ground surface displacements but acquiring this information is challenging. Three dimensional point-cloud data from terrestrial laser scanning (TLS) show potential for obtaining ground displacements accurately. Problems arise, however, when estimating continuous displacement fields from TLS data because reflecting points from sequential scans of moving ground are nonunique, thus repeat TLS surveys typically do not track individual reflectors. In this dissertation, the cross-correlation-based Particle Image Velocimetry (PIV) method is implemented to derive 3D surface deformation fields using TLS point-cloud data. Associated errors are estimated and the method’s performance is tested with synthetic displacements applied to a TLS dataset. The method is applied to the toe of the episodically active Cleveland Corral landslide in northern California using six different TLS scans acquired between June 2005 and April 2012. Estimated displacements agree well with independent measurements at better than 9% root mean squared (RMS) error and permit further analysis to infer the subsurface deformation characteristics of the landslide. The hypothesis that the depth and orientation of the buried slip surface and the subsurface slip rate can be estimated using the surface displacement field is tested. To estimate slip depth and slip rate of the slide, a 2D balanced cross-section (BC) method commonly applied to landslides and an elastic dislocation (ED) model widely applied to study geologic faults are performed. The BC method provides slip-surface depth; the ED model determines the slipsurface depth as well as orientation and slip magnitude. The estimated slip-surface depths using both methods agree with direct measurements of depth. This indicates that these two approaches may offer more efficient and less costly remote means of inferring landslide geometry and slip behavior. The PIV method is also compared with the iterative closest point method and the efficacy of using these two methods to estimate 3D displacement fields using TLS data are discussed. The estimated surface displacement and the inferred subsurface deformation enable assessment of the hazards posed by large, slow-moving landslides.
TABLE OF CONTENTSList of Tables List of Figures Preface CHAPTER ONE
DISPLACEMENT FIELDS FROM POINT CLOUD DATA: APPLICATION OF PARTICLEIMAGING VELOCIMETRY TO LANDSLIDE GEODESY
1.2 Terrestrial Laser Scanning
1.3 Particle Imaging Velocimetry
1.4 Synthetic Examples
1.5 Application: Cleveland Corral Landslide
1.5.1 June 2005 - January 2007
1.5.2 January - May 2010
LANDSLIDE SUBSURFACE SLIP CHARACTER INFERRED FROM SURFACEDISPLACEMENT FIELDS
2.2 Methods for Inferring Subsurface Slip
2.2.1 Balanced cross-section (BC)
2.2.2 Dislocation in an Elastic Half-Space
2.3 The Cleveland Corral landslide
2.4 3D Displacement Field
2.5 Subsurface Inference Results
2.5.1 Balanced Cross-section
2.5.2 Elastic Dislocation
DETERMINING GROUND DISPLACEMENT FIELDS OF SMALL SPATIAL EXTENT
USING TERRESTRIAL LASER SCANNER DATA: A COMPARISON OF 3D METHODSAPPLIED TO LANDSLIDE MONITORING
3.2 Displacement Estimation Methods
3.2.1 Particle Image Velocimetry
3.2.2 Iterative Closest Point
3.2.3 Synthetic Tests
3.3 Results from the Cleveland Corral Landslide
3.3.1 Displacement time series
3.3.2 Pattern of surface deformation
List of Figures
Illustration of velocity estimation based on cross-correlation.
An illustration of the PIV method applied to synthetically produced point cloud data.
a) Location of the Cleveland Corral Landslide.
TLS point cloud data and area of interest for PIV analysis.
PIV estimation of synthetic displacement applied to the June 2005 point cloud data.
PIV estimated total displacement field of CCL between June 2005 and January 2007.
PIV estimation of a synthetic signal applied to January 2010 point cloud data.
PIV estimated total displacement field of CCL between January 2010 and May 2010.
Comparison of PIV-computed displacement (magnitude) of CCL with GPS measurement.
Sketch of the two models used to infer landslide subsurface slip geometry.
Displacement fields for an active part of the Cleveland Corral landslide obtained for two time periods using repeat TLS scans and PIV.
Estimated slip-surface depth using the balanced cross-section (BC) method.
Marginal probability distribution for three dislocation slip parameters.
CHAPTER III Figure 3.1.
Movement rate and spatial extent of the most common geologic features.
TLS point cloud data of a stationary building from two temporally different acquisitions.
Location of the study area and data acquisition.
Conceptual sketch showing components of landslide displacement at surface of the sliding block.
Sketch showing the point-to-plane distance matching in ICP.
Synthetic signals and residuals using PIV.
RMSE of ICP estimation of the synthetic signal applied to a TLS data using different window sizes.
Synthetic signals and residuals using ICP.
ICP and PIV estimated displacements (Jan-May-Jun 2010).
ICP and PIV estimated displacements (2010-2014).
Comparison of PIV and ICP displacements with observations.
List of Tables Table 3.1 Average misfit of the PIV and ICP estimation with different window sizes for a synthetic signal applied to Jan 2010 TLS data and Jan-Jun 2010 TLS data.
PREFACEThis dissertation has three chapters, each of which focuses on a separate aspect of estimating landslide surface deformation using terrestrial laser scanner (TLS) data and its applications.
Landslides are ubiquitous and cause thousands of deaths and injuries each year. Understanding of landslide stability and governing processes requires good knowledge of ground surface displacements but acquiring this is challenging. The first chapter of this dissertation presents a method to estimate landslide surface displacement using TLS data from the slow-moving Cleveland Corral landslide (CCL) in California. A version of the chapter one has been published as Aryal, A., Brooks, B.A., Reid, M.E., Bawden, G.W., and Pawlak, G.R., 2012, Displacement
fields from point cloud data: Application of particle imaging velocimetry to landslide geodesy:
Journal of Geophysical Research-Earth Surface, v. 117, p. 15. This paper was also featured in the American Geophysical Union research spotlight selected by the editors. Ben Brooks, Mark Reid and Gerald Bawden collected TLS data and Geno Pawlak suggested the PIV method to use for this paper. Ben Brooks and Mark Reid aslo edited the text significantly before it was submitted for the publication.
Chapter two focuses on application of the displacement fields estimated in chapter one to infer the subsurface deformation character of CCL. This chapter demostrates the use of two different models to locate the landslide slip surface. Landslide subsurface inference using surface displacements is in a nascent stage but this study demostrates a good agreement between the estimated slip depth with the observed depths. A version of this chapter is a manuscript prepared as ‘Aryal, A., Brooks, B.A., and Reid, M.E., Landslide subsurface slip character inferred from surface displacement fields’and the manuscript has been internally reviewed by USGS scientists Ole Kaven and Jonathan Stock.
Finally, chapter three compares two competing methods to estimate 3D displacement fields using TLS data. Currently, there is no accepted best method of using TLS data to estimate 3D displacement field automatically. This chapter compares results from the particle image velocimetry (PIV) and iterative closest point (ICP) methods and discusses the results. The estimated dense displacement fields are also used to analyze the pattern of surface deformation of the landslide.
DISPLACEMENT FIELDS FROM POINT CLOUD DATA: APPLICATION OF PARTICLE
IMAGING VELOCIMETRY TO LANDSLIDE GEODESYPublished in its present form as Aryal, A., Brooks, B.A., Reid, M.E., Bawden, G.W., and Pawlak, G.R., 2012, Displacement
fields from point cloud data: Application of particle imaging velocimetry to landslide geodesy:
Journal of Geophysical Research-Earth Surface, v. 117, p. 15.