Landslides (2012) 9:75–91DOI 10.1007/s10346-011-0286-4Received: 4 March 2010Accepted: 12 July 2011Published online: 28 July 2011© Springer-Verlag 2011

Marc-André Brideau I Matthieu Sturzenegger I Doug Stead I Michel Jaboyedoff I MartinLawrence I Nicholas Roberts I Brent Ward I Thomas Millard I John Clague

Stability analysis of the 2007 Chehalis lake landslidebased on long-range terrestrial photogrammetryand airborne LiDAR data

Abstract On December 4th 2007, a 3-Mm3 landslide occurredalong the northwestern shore of Chehalis Lake. The initiationzone is located at the intersection of the main valley slope and thenorthern sidewall of a prominent gully. The slope failure caused adisplacement wave that ran up to 38m on the opposite shore ofthe lake. The landslide is temporally associated with a rain-on-snow meteorological event which is thought to have triggered it.This paper describes the Chehalis Lake landslide and presents acomparison of discontinuity orientation datasets obtained usingthree techniques: field measurements, terrestrial photogrammetric3D models and an airborne LiDAR digital elevation model todescribe the orientation and characteristics of the five disconti-nuity sets present. The discontinuity orientation data are used toperform kinematic, surface wedge limit equilibrium and three-dimensional distinct element analyses. The kinematic and surfacewedge analyses suggest that the location of the slope failure(intersection of the valley slope and a gully wall) has facilitatedthe development of the unstable rock mass which initiated as aplanar sliding failure. Results from the three-dimensional distinctelement analyses suggest that the presence, orientation and highpersistence of a discontinuity set dipping obliquely to the slopewere critical to the development of the landslide and led to afailure mechanism dominated by planar sliding. The three-dimensional distinct element modelling also suggests that thepresence of a steeply dipping discontinuity set striking perpen-dicular to the slope and associated with a fault exerted asignificant control on the volume and extent of the failed rockmass but not on the overall stability of the slope.

Keywords Terrestrial photogrammetry . LiDAR . Distinctelement . Limit equilibrium

IntroductionThe Chehalis Lake landslide occurred on December 4, 2007 on thenorthwest shore of Chehalis Lake in southwest British Columbia(Fig. 1). The rock slope failure and debris avalanche occurred onthe southeast-facing slope of Mount Orrock in the Cohoe Creekdrainage (Fig. 2). The headscarp is located at an elevation of∼840 m asl (above sea level) while the lake shoreline is at ∼240 masl. The headscarp is 210 m in width and the landslide travelled ahorizontal distance of 950 m before entering the lake. Thelocations and cross-sections pre- and post-landslide are presentedin Figs. 3 and 4, respectively. The slope failure was associated withan intense rain-on-snow event as illustrated by the meteorologicaldata collected at the Agassiz weather station located approx-imately 30 km southeast of Chehalis Lake (Fig. 5). A cold periodbetween November 30 and December 2, 2007 with precipitation

falling as snow was followed by warmer temperature and rainfallon December 3 and 4, 2007. Rain-on-snow events increase the“effective intensity” of a precipitation occurrence by combiningthe current rainfall with the water temporarily stored as snow onthe slope. Runoff and infiltration are thought to have increasedpore water pressures in the fractured rock mass and reduced theeffective friction angle along discontinuity surfaces. The failedmass, with a volume in the order of 3 Mm3, entered Chehalis Lake,resulting in a displacement wave that destroyed three localcampgrounds and removed the vegetation on the opposite shoreof the lake to a maximum height of 38 m (Fig. 6). A preliminaryassessment of the landslide conducted by BGC Engineering Inc.(2008), at the request of the BC Provincial Emergency Program,highlighted that the failure surface was controlled by pervasiveplanar structures which in places consisted of a fault with a few-centimetres-thick gouge. Fresh tension cracks were observed upto 21 m behind the headscarp (BGC Engineering Inc. 2008). Thedamage caused by the landslide-generated displacement waveassociated with the Chehalis Lake landslide has previously beenbriefly described by Geertsema et al. (2009) and Stephenson andRabinovich (2009). The main objective of the research presentedin this paper was to characterise the geology, rock mass quality,tectonic structures and the discontinuity sets present at theChehalis Lake landslide. This information was then used as inputin slope stability analyses to evaluate potential failure mecha-nisms for the event. A secondary objective was to integrate theinformation of discontinuities in the initiation zone based onseveral data collection techniques and to discuss the applicabilityand limitations of each technique.

Relevant previous landslide studiesMeteorological conditions are regarded as one of the mostcommon trigger for landslides (Sidle and Ochiai 2004; Highlandand Bobrowsky 2008). Rain-on-snow event is a particularmeteorological condition that has been discussed by Jakob andWeatherly (2003) as a contributing or triggering factor for debrisflows in southwestern British Columbia. Similar patterns ofpronounced cold and warm temperatures cycles coupled withprecipitation prior to rock slope failure initiation have been notedin British Columbia (e.g. rockfall along transportation corridors:Peckover and Kerr 1977; Hope Slide: Mathews and McTaggart1978, East Gate Landslide: Brideau et al. 2006; ClanwilliamLandslide: Brideau et al. 2008) and elsewhere around the world(e.g. Bjerrum and Jorstad 1968; Dunlop and Hutchinson 2009).

There have been two other documented instances oflandslides entering lakes and generating displacement waves

Landslides 9 & (2012) 75

Original Paper

in British Columbia. In 1946, a rock slope failure on MountColonel Foster on Vancouver Island was triggered by anearthquake and entered the lake at its base (Evans 1989). Themaximum run-up height of the displacement wave wasmeasured at 51 m and was still evident in the mid-1980s. Thesecond event occurred in 1998 at Troitsa Lake, in centralBritish Columbia, where the surface material associated with alarge fan delta failed into the lake, generating a 20-m-high

displacement wave that unsettled the lake for a period of 4 h(Schwab 1999). Landslides entering fiords and creating dis-placement waves have also been investigated in BritishColumbia (e.g. Bornhold and Harper 1998; van Zeyl 2009).Similar landslide-generated waves in lakes or reservoirs havebeen described elsewhere in the world (e.g. Plafker andEyzaguirre 1978; Wang et al. 2004; Genevois and Ghirotti2005; Grimstad 2005; Panizzo, et al. 2005).

1

2

Rockslide

Debrisavalanche

N

Valleyslope

Northgully wall840m.a.s.l.

440m.a.s.l.

Southgully wall

240m.a.s.l.

Figure 7a

CohoeCreek

250m

Fig. 2 Overview of the Chehalis Lakelandslide showing geomorphicfeatures referred to in slope stabilityanalyses. Slope angle along runoutsegment 1 was between 36° and 40°and between 30° and 32° alongsegment 2

ChehalisLake

Landslide

StudyArea

1km

LegendRoadCampgroundRiverLakeGeologyQuartz dioriteAndesitic volcanicsFault

NBritishColumbia

N49.40o

W122.00o

Fig. 1 Location and regionalgeology map of the Chehalis Lakelandslide (geology from Massey et al.2005; faults from Thompson 1975)

Original Paper

Landslides 9 & (2012)76

Geology and rock mass descriptionChehalis Lake is located along the contact of late Jurassic islandarc rocks of the Harrison Lake Formation and middle Jurassic toearly Cretaceous rocks of the Coast Plutonic Complex (Mahoneyet al. 1995). The area east of Chehalis Lake was mapped byMonger (1970) and Arthur (1986) while the areas north and southof the Chehalis Lake landslide were mapped by Thompson (1975)and Mahoney et al. (1995), respectively. Monger (1970) and Arthur(1986) observed that most of the Harrison Lake Formationconsists of intermediate to felsic volcanic flows and pyroclasticrocks, while Thompson (1975) did not observe pyroclastic rocks inthe southern portion of Chehalis Lake. Arthur (1986) noted thatfaults were difficult to distinguish in the field (due to the lack ofbedding structures and the degree of hydrothermal alteration) butthat several lineaments trending northwest–southeast were visible

on aerial photographs. Thompson (1975) and Mahoney et al.(1995) included northwest–southeast trending faults in theirgeological maps. Thompson (1975) also noted the presence of anorth-northeast (010o) trending fault (Fig. 1).

Fieldwork conducted as part of this project found that thefailure is located near an irregular intrusive contact betweenandesite volcanics and quartz diorite units (Fig. 1). The failedmaterial is predominantly composed of quartz diorite. Xen-oliths of volcanic material varying in size from a fewcentimetres to several metres occur in the quartz diorite onthe northern edge of the landslide (Fig. 7a). Rock massdescriptions and discontinuity orientation measurements wereobtained at six stations (Fig. 3). The rock mass quality wasassessed using the Geological Strength Index (GSI) chartpresented in Marinos and Hoek (2000) which is based on field

300 0 300 sreteM051

Legend

Field geological mapping

Road

N

cross-section1

cros

s-se

ction

2

A

A’

B

B’

Field measurements ofdiscontinuitiesLocation from which terrestrialphotogrammetry acquired

Area from which discontinuity dataextracted using phogrammetry

Area from which discontinuity dataextracted from airborne-LiDARusing COLTOP-3D

Topographic contour(10m interval)

W122.00o

N49.46o

F ig . 3 Ove r v i ew map w i t horthophoto showing the location ofcross-sections, field stations and areacovered by the terrestrialphotogrammetry and COLTOP-3Danalyses

Landslides 9 & (2012) 77

observations of the rock mass structure and the condition ofthe discontinuity surfaces. The application and limitations ofthe GSI to rock slope characterization have been discussed inMarinos et al. (2005) and Brown (2008). They discussed theimportance of the scale of observation, the need for a good three-dimensional outcrop and how excavation techniques can reduce the

surface rock mass quality. The average GSI range obtained for therock mass at the Chehalis Lake landslide was 50–60. Thiscorresponds to a very blocky (four discontinuity sets or more) rockmass with good (rough) surface conditions (Fig. 7a). Field estimatesof the intact rock strength of both the quartz diorite and andesitesuggested a strong to very strong or R4 to R5 range (∼100 MPa)according to the strength classes defined in Hoek and Brown (1997).The rock mass had a fresh (WI to WII) weathering state with somelocalized zones of slight to moderate weathering (WII to WIII)conditions (ISRM 1978).

Data collectionDiscontinuity orientation data were collected at the Chehalis Lakelandslide using three different techniques. Upon increasing thescale of observation, they are: traditional field data collection near

Cross-section 1

Cross-section 2

Ele

vatio

n (m

.a.s

.l.)

Ele

vatio

n (m

.a.s

.l.)

Legend

a

A’

b

B’

0 100 200 300 400 500m

0 200 400 600 800 1000m

1000

800

600

400

200

760

720

680

600

560

640

Pre-failure topography(10m contour TRIM)

Post-failure topography(1m LiDAR DEM)

Fig. 4 Longitudinal and transverse cross-sections of the pre- and post-failuretopography (see Fig. 3 for the location of cross-sections). Note that the pre-failuretopography has a lower resolution than the post-failure topography

-8

-6

-4

-2

0

2

4

6

8

10

12

14

0

10

20

30

40

50

60

70

Maximum temperature

Precipitation (water equivalent)

Minimum temperatureT

otal

wat

er e

quiv

alen

t pre

cipi

tatio

n (m

m)

Date of the Chehalis Lakelandslide

Tem

pera

ture

(o C

)

28/11/07 29/11/07 30/11/07 01/12/07 02/12/07 03/12/07 04/12/07 05/12/07 06/12/07

Fig. 5 Meteorological conditionsleading up to the rock slope failureas recorded at the Agassiz weatherstation (located approximately 30 kmfrom Chehalis Lake). Data fromEnvironment Canada (2009)

Fig. 6 Damage to trees caused by the displacement wave generated by thelandslide entering Chehalis Lake. Photograph taken at the location of the formercampground at the northern end of the lake (view toward north and landslide 1km to the left of the photograph)

Original Paper

Landslides 9 & (2012)78

the landslide, terrestrial photogrammetry of the headscarp areaand colour-shaded relief maps from a 1-m resolution airborneLight Detection and Ranging (LiDAR) of the area surrounding thelandslide. The location and extent over which each of thetechniques was applied are summarized in Fig. 3. The applicationof remote sensing techniques was particularly suitable to this casestudy due to the steep and inaccessible rock face at the headscarpand because the slope is still considered to be unstable.

Field assessmentDetailed engineering and structural geology mapping was under-taken in the lower part of the Chehalis Lake landslide (Fig. 3). The

rock mass at each station was described according to the categoriesand terminology described in ISRM (1978). At each station, theauthors evaluated the rock mass for the presence of discontinuitysets with similar orientations. Once a discontinuity set wasidentified, several measurements of its dip and dip direction werecollected using a Clar-type compass. A similar approach is describedin Mathis (1988) for cell mapping. The advantage of the fieldassessment method is that, in addition to getting information aboutdiscontinuity orientation, it can easily provide information aboutdiscontinuity infill and roughness and about the rock lithology andweathering grade. In turn, the disadvantages include limited accessto outcrops and potential danger to the person collecting the datadue to the unstable nature of the slope.

DS2

DS1

DS3

DS4

DS5

DS4

DS5

a

b

DS1

DS2

DS3

50 m

1 m

Northgullywall

Valleyslope

Fig. 7 Discontinuity sets in the rockmass at the Chehalis Lake landslide.a Bedrock exposed below theinitiation zone, b the initiation zone(view toward northwest)

Landslides 9 & (2012) 79

Terrestrial photogrammetrySturzenegger and Stead (2009a, b) and Sturzenegger et al. (2009)have demonstrated the applicability of long-range terrestrialphotogrammetry to measure the orientation of discontinuities inlarge natural rock slopes and open pits. Their methodology wasapplied to the Chehalis Lake landslide by acquiring digitalphotographs using a Canon EOS 30D digital camera with 200-mm focal length lens and processing the images obtained using

the 3DM CalibCam/Analyst software (Adam Technology 2009).Three-dimensional low-resolution models were created for thewhole slope and higher-resolution ones for areas of interest in theheadscarp area (Fig. 3). Registration of the photogrammetricmodels was achieved using the airborne LiDAR data to providecontrol points. The ground resolution of the three-dimensionalmodels was calculated to be 38 cm (see Sturzenegger and Stead(2009a) for details). The discontinuity orientations were obtained

N

EW

SDS1 DS2

DS5

DS4

DS3

Lake

Road

N

100 m

Fig. 9 Discontinuities identified inthe area surrounding the landslidebased on the airborne LiDAR DEMcoloured according to orientationusing COLTOP-3D

DS1

DS2

DS3DS4

DS5

50m

Fig. 8 Detail of discontinuitiesidentified in the initiation zone of thelandslide based on the three-dimensional model created fromterrestrial photogrammetry (viewtoward west)

Original Paper

Landslides 9 & (2012)80

by fitting circles to the discontinuity surfaces identified in the 3Dmodels of the headscarp (Fig. 8). The main advantages of theterrestrial photogrammetry technique are that information aboutthe discontinuity orientation and persistence can be acquired withno need to directly access the slope and it is possible to obtaininformation from outcrops that cannot be physically accessedwhile its limitation include occlusion and image resolution bias(Sturzenegger and Stead 2009a, b).

COLTOP-3DCOLTOP-3D is a software package that uses high-resolution DEMsto create colour-shaded relief maps (Jaboyedoff et al. 2004). Thepole of each DEM cell is assigned a unique colour based on its dipand dip direction orientation. This facilitates the visual determi-nation of the orientation of planar features, such as longpersistence discontinuities, fault scarps or landslide scarps(Fig. 9). This approach has previously been applied successfullyto stability assessments of large rock slopes (e.g. Derron et al.2005; Jaboyedoff et al. 2007; Pedrazzini et al. 2008; Jaboyedoff etal. 2009; Metzger et al. 2009; Oppikofer 2009; Oppikofer et al.2009). The advantages and limitations of this technique aresimilar to those of the terrestrial photogrammetry listed in theprevious section.

Data interpretation

Field measurementsThe orientation details of the discontinuity sets identified fromthe field measurements are summarized in Table 1 and Fig. 10a.

Five discontinuity sets were identified (Fig 7). Based on itsorientation and pervasiveness, discontinuity set (DS) 1 representssurfaces that could have formed as sheeting joints (e.g. Hencher etal. 2011), and together with DS2 they form the main failure surfaceof the Chehalis Lake landslide. Based on its long persistence andorientation approximately parallel to the north gully wall, DS3appears to be associated with a fault that has the same orientationas the lineaments observed by Arthur (1986) in aerial photo-graphs. The surface characteristics of each discontinuity set arepresented in Table 2. The trace length of the DS5 was estimated tobe very low (<1 m); this is attributed to its dip into the slope,creating overhanging surfaces that prevent the formation fromextensive exposures.

Terrestrial photogrammetryThe stereonet of the poles to the fitted circles extracted from thef=200 mm lens models is presented in Fig. 10b. The spatialdistribution of the discontinuities identified in the 3D modelscreated from the terrestrial photogrammetry images is verysimilar to the discontinuities identified based on field measure-ments. The same five discontinuity sets can be identified (Table 1;Fig. 10b). Discontinuity sets 1 and 2 were extracted from the largeconcentration of discontinuities identified in the photogrammetrymodels based on the field observations. The biggest difference inthe orientation of the discontinuity sets is in the dip of DS1 whichis steeper in the photogrammetry models when compared withthe field measurements (Table 1). The merging of DS1 and DS2 inthe photogrammetry models is potentially due to three factors.Firstly, because the scale of observation in the photogrammetry is

Table 1 Summary of discontinuitysets mean orientation for each ofthe datasets acquired at the Che-halis Lake landslide

Dip/dip direction (o) N=number of measurements K=Fisher dispersion coefficient

Discontinuity set Field measurements Terrestrial photogrammetry COLTOP-3D

DS1 28/150 43/149 38/198

N=17 N=68 N=8 surfaces

K=48 K=17 N=285 poles to DEM cell

DS2 59/135 61/146 67/149

N=13 N=37 N=8 surfaces

K=21 K=55 N=165 poles to DEM cell

DS3 67/062 77/074 51/052

N=12 N=26 N=8 surfaces

K=20 K=24 N=210 poles to DEM cell

DS4 74/200 76/222 64/246

N=16 N=19 N=8 surfaces

K=21 K=44 N=286 poles to DEM cell

DS5 43/292 55/307 12/297

N=19 N=15 N=8 surfaces

K=23 K=20 N=83 poles to DEM cell

DS6 NA NA 20/101

N=1 surface

N=226 poles to DEM cell

Landslides 9 & (2012) 81

smaller (i.e. covers a larger area) than the field observations, DS1and DS2 cannot be resolved as separate discontinuity sets. Thiseffect was highlighted in Sturzenegger and Stead (2009a).Secondly, because the line of sight of the camera (trend/plungeof ∼30/140°) when acquiring the digital photographs was similarto the orientation of DS1 (mean dip/dip direction of ∼28/150° infield measurements), it could have resulted in an orientation bias(occlusion), as discussed in Sturzenegger and Stead (2009b) andLato et al. (2010), that would “miss” the shallower members ofDS1 (Fig. 10b). A third potential reason for the merging of DS1 andDS2 is related to the contouring routine and the increased samplesize (field measurements: N=102; photogrammetry; f=200 mm;N=202). This effect of the sample size has been discussed byStauffer (1966).

COTOP-3DThe airborne LiDAR data for the vicinity of the Chehalis Lakelandslide was used to create a DEM with a resolution of 1 m. ThisDEM was used to identify 41 surfaces using COLTOP-3D whichrepresented a total of 1,389 DEM cells (i.e. each individual surfaceis composed of multiple DEM cells; Fig. 10c). The spatialdistribution of the surfaces compares well with the discontinuitydata obtained from the field measurements and the photo-grammetry models (Table 1). The dip direction of the DS1 fromCOLTOP-3D is slightly more southward than was measured fromthe other datasets. Discontinuity set five also has a noticeablyshallower dip in the COLTOP-3D results. This is attributed to adirectional bias as DS5 dips into the slope and is under-represented in the airborne LiDAR data of the slope face onwhich the failure occurred. A sixth discontinuity set was observedin the stereonet from the COLTOP-3D results but not in the fieldmeasurements or photogrammetric datasets. Using the ortho-photo of the area to assess the origin of this surface, it wastentatively attributed to the presence of small colluvial cones onthe slope and was disregarded in further slope stability analyses.

Slope stability analyses

Kinematic analysisA kinematic analysis is a rock slope stability test for identifyingsimple structurally controlled failure modes such as planarsliding, wedge failure and toppling. It takes into considerationdiscontinuity orientation, slope orientation and the friction anglealong the discontinuity surfaces. The stereographic techniques forthe kinematic analysis of these simple failure modes are describedin Richards et al. (1978). To account for the variation of thetopography in the immediate vicinity of the Chehalis Lakelandslide, two slope orientations were considered in the kinematicanalysis (valley slope at 35/115°, dip/dip direction; north gully wallat 40/185°). The friction angle along all discontinuities wasassumed to be 30° based on the planar/undulating and roughdiscontinuity surface conditions. Table 3 and Fig. 11 summarizethe results of the kinematic analyses performed using the threediscontinuity datasets. Overall, the north gully wall slopeorientation appears to be less stable than the valley slope. Slidingalong DS1 out of the north gully wall is feasible in all datasets andmarginally feasible out of the valley slope (Fig. 11a). Taking intoconsideration the natural variability of the discontinuity setorientations, wedge failures should be considered to be margin-

COLTOP-3D

Field measurements

DS2

DS2

DS2

DS2

DS4DS4

DS1DS1

DS6DS6

DS5

DS5

DS5

DS5

N

S

EW

DS2

DS2

DS1

DS1

DS3

DS3

DS5

DS5

DS4

DS4

DS6

DS6

N

S

EW

DS1

DS1

DS2

DS2DS3

DS3

DS4

DS4

DS5

DS5

N

S

EW

DS5

DS5 DS4

DS4

DS3

DS3

DS1

DS1

DS2

DS2

N=102 polesto discontinuities

Terrestrialphotogrammetry

N=1389 polesto DEM cells

Photogrammetryline of sight

N = 202 polesto fitted circles

a

b

c

Fig. 10 Stereonets of the poles to discontinuity surfaces identified in a fieldmeasurements, b terrestrial photogrammetry and c COLTOP-3D. Lowerhemisphere equal angle stereonets

Original Paper

Landslides 9 & (2012)82

ally feasible out of the north gully wall but less so out of the valleyslope (Fig. 11b). Only random discontinuities were observed to fallwithin the toppling envelope along both the north gully wall andthe valley slope (Fig. 11c).

Surface wedge analysisSwedge is a limit equilibrium code (Rocscience 2006) that wasused to evaluate the stability of surface wedges in the rock slopefaces at the Chehalis Lake landslide. The combination analysis inSwedge uses a list of discontinuities specified by the user (e.g.discontinuity survey data) to calculate the factor of safety for eachvalid wedge intersection for a given slope face. To account for thetopography in the immediate surroundings of the Chehalis Lakelandslide, the same two slope orientations (valley slope at 35/115°dip/dip direction; north gully wall at 40/185°) considered in thekinematic analyses were used in the combination analyses. Theanalysis was performed using all three discontinuity orientationdatasets for both slope orientations. The results, summarized inTable 4 and Fig. 12, strongly suggest that the north gully wallorientation is more susceptible to wedge failures for all disconti-nuity orientation datasets. With the exception of the analysisresults for the north gully wall based on field measurements, alldatasets have consistent percentages of valid wedge failures(Table 4). The discrepancy with the field measurements might

be related to the smaller number of discontinuity orientations inthat particular dataset compared to the other datasets.

Overall, the results from the surface wedge analysis areconsistent with those of the kinematic. Both methods suggest thatthe north gully wall is more prone to failure. The advantage of thesurface wedge combination analysis is that it is capable of usingthe full dataset to account for the natural variability indiscontinuity orientations at the site.

Three-dimensional distinct element modellingThe three-dimensional distinct element code 3DEC (Itasca 2008)was used to model the interaction between the jointed rock massand the topography at the Chehalis Lake landslide. The 3DEC coderepresents the rock mass as a collection of three-dimensional blocksunder static or dynamic loading. The strength of thematerial makingup the blocks and bounding discontinuities are specified by the user.Large displacements and rotation along the discontinuities boundingthe blocks are permitted. More information about 3DEC can befound in Cundall (1988) and Hart et al. (1988).

In this paper, 3DEC was used to evaluate the influence of:(1) the mean discontinuity set orientations derived from thevarious techniques, (2) individual discontinuities and (3)variations in assumed friction angle along the discontinuities.Rigid block conditions were assumed for the analyses pre-

Table 3 Summary of the kinematic analyses performed on the discontinuity datasets acquired at the Chehalis Lake landslide

Failure mechanism Field measurements Terrestrial photogrammetry COLTOP-3D

Valley slope Sliding Feasible on DS1 Marginal on DS1 No

Wedge Marginal on DS1/DS4 No No

Toppling Marginal on randomdiscontinuities

Marginal on randomdiscontinuities

Marginal on DS4

North gully wall Sliding Feasible on DS1 Feasible on DS1 Feasible on DS1

Wedge Marginal along DS1/DS3 Marginal along DS1/DS3and DS1/DS4

Feasible along DS1/DS4and marginal along DS1/DS2

Toppling Marginal on randomdiscontinuities

Marginal on random discontinuities No

Table 2 Summary of the disconti-nuity surface characteristicsobserved during the fieldinvestigation

Discontinuity set Large (metre scale)roughness

Small (centimetre scale) roughness Tracelength (m)

DS1 Planar Rough some smooth in quartz diorite 1–3

Rough some slickensided in andesite

DS2 Planar and undulating Rough some smooth in quartz diorite 3–10

Rough some slickensided in andesite

DS3 Planar Rough some smooth in quartz diorite 1–3 and 3–10

Rough some slickensided in andesite

DS4 Planar Rough some smooth in quartz diorite 1–3 and 3–10

Rough some slickensided in andesite

DS5 Undulating Rough some smooth in quartz diorite <1

Rough some slickensided in andesite

Landslides 9 & (2012) 83

sented (i.e. the Chehalis Lake landslide is assumed to be astructurally controlled rock slope failure). The rock density and

discontinuity properties used in the 3DEC models are pre-sented in Table 5 while a summary of the results obtained in

toppling

wedge

sliding

Valley

N

S

EW

DS2DS3

DS4

DS5

DS1

Valley

N

S

EW

Valley

N

S

EW

Valley

N

S

EW

DS2

DS4

DS1

DS3

DS6

DS5

N

S

EW

Valley

N

S

EW

toppling

wedge

sliding

Frictioncone

Daylightenvelopes

Frictioncone

Daylightenvelopes

Topplingenvelope

Topplingenvelope

Topplingenvelope

Topplingenvelope

DS5DS3

DS1DS2 DS4

DS1DS2

DS3

DS4

DS5

DS6

Sliplimit

Sliplimit

Sliplimit

Sliplimit

Valley

N

S

EW

Northgully

DS1DS2

DS4

DS3DS5

N

S

EW

N

S

EW

sliding

wedge

toppling

Northgully

Valley

Northgully

Valley

Sliplimit

Sliplimit

DS1

DS2

DS3

DS4

DS5

Topplingenvelope

Topplingenvelope

Daylightenvelope

Valley

a

b

c

Northgully

Northgully

Northgully

Northgully

Northgully

Northgully

Legendpole to discontinuity

intersection of twodiscontinuity sets

Field measurementsN = 102

TerrestrialphotogrammetryN=202

COLTOP-3DN=1389 poles toDEM cells

30o friction

30o frictioncircle

cone

Fig. 11 Kinematic analysis conducted on the discontinuities a sliding, b wedge failure and c toppling failure mechanism for the field measurements, terrestrialphotogrammetry model and COLTOP-3D. Lower hemisphere equal angle stereonets

Original Paper

Landslides 9 & (2012)84

the various models considered in this study is presented inTable 6. A simplified geometry consisting of three planes(valley slope at 35/115° dip/dip direction, north gully wall at 40/185° and south gully wall at 40/080°), labelled in Figs. 2 and13a, was used to approximate the average orientation of eachslope.

Discontinuity set orientationsFigure 13 compares the calculated total displacement, contouredin metre, obtained after 30,000 numerical calculation steps inmodels with the a mean discontinuity set orientation based on thedatasets acquired at the Chehalis Lake landslide. The three-dimensional distinct element model associated with the fieldmeasurement data resulted in a large slope failure on the valleyslope and the intersection between the valley slope and the northgully wall (Fig. 13a). The volume of the unstable mass (arbitrarilydefined as the cumulative volume of blocks having a totaldisplacement >1 m; Table 6) obtained in this scenario isapproximately four times larger (11 Mm3) than the calculatedvolume based on aerial photograph interpretation (3 Mm3). The3DEC model using the mean discontinuity set orientationsobtained from terrestrial photogrammetry resulted in a smallvolume (0.17 Mm3) slope failure at the intersection of the valleyslope and north gully wall (Table 6; Fig. 13b). The COLTOP-3Ddataset did not result in a significant slope failure (Fig.13c); onlyvery small blocks displayed any significant calculated displace-ment (Table 6).

Influence of individual discontinuity setThe influence of each discontinuity set on the slope stabilityconditions was assessed by creating five three-dimensionaldistinct element models that included only four discontinuitysets, with each of the five discontinuity sets being omitted in turn(Fig. 14). This series of models was built using the meandiscontinuity set orientations obtained from the field measure-ments dataset since it is the one which resulted in the largestsimulated slope instability. Where DS1 was omitted (i.e. the modelassumed that only DS2, DS3, DS4 and DS5 were present), all threeslopes were stable with only a few very small blocks moving morethan 1 m (Fig. 14a; Table 6). This result highlights the critical roleplayed by DS1 in the failure of the Chehalis Lake landslide. Themodels that did not include DS2 (Fig. 14b), DS4 (Fig. 14d) and DS5

Table 4 Summary of the surface wedge combination analyses performed on the discontinuity measurements obtained with the various data collection methods usedthe Chehalis Lake landslide

Field measurements Terrestrial photogrammetry COLTOP-3D

Valley slope Number of combinations 5,151 20,301 963,966

Number of valid wedges 224 576 31,378

Number of stable wedges 206 (92%) 564 (98%) 2,9579 (94%)

Number of failed wedges 18 (8%) 12 (2%) 1,799 (6%)

North gully wall Number of combinations 5.151 20,301 963,966

Number of valid wedges 195 1,442 64,161

Number of stable wedges 166 (85%) 875 (61%) 41,279 (64%)

Number of failed wedges 29 (15%) 567 (39%) 22,882 (36%)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Photogrammetry

Discontinuity orientation dataset

UnstableStable

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Photogrammetry

Discontinuity orientation dataset

UnstableStable

a Analysis for valley slope

b Analysis for north gully wall

Field COLTOP-3D

Field COLTOP-3D

Fig. 12 Cumulative bar graphs summarizing the percentage of stable andunstable wedges obtained for each dataset along the a valley slope and b northgully wall

Landslides 9 & (2012) 85

(Fig. 14e) all resulted in similar maximum calculated totaldisplacements and unstable volume masses (53 to 71 m and 7.3to 8.0 Mm3; Table 6). Visually, the unstable areas in these threemodels (Fig. 14b, d, e) are similar to the equivalent model withfive discontinuity sets (unstable volume of 11 Mm3; Fig. 13a),but the models with only four discontinuity sets had smallervolumes with an average of ∼7.5 Mm3. Finally, the model thatdid not include DS3 (Fig. 14c) resulted in a 1.7-Mm3 rock slopefailure along the intersection of the valley slope and the northgully wall (Table 6). DS3 therefore plays an important role inthe development of an unstable rock mass at the Chehalis Lakelandslide.

Assumed friction angleThe previous 3DEC analyses were performed assuming a frictionangle of 25°. The sensitivity of the slope stability conditions to thefriction angle value was assessed by investigating a model with allfive discontinuity sets using the mean orientation from the fieldmeasurements dataset, 40 m discontinuity spacing and anassumed 30° friction angle. The model with the higher frictionangle value led to stable slope conditions with only small blocksmoving more than 1 m (Table 6).

Discussion

Discontinuity orientation datasetsThe three discontinuity orientation datasets acquired at theChehalis Lake landslide all revealed similar distribution patterns(Fig. 10), although the mean orientation of the specific disconti-nuity sets varied (Table 1). This variation in orientation led tosome differences in the basic slope stability analyses (kinematicand surface wedge), but the overall expected rock slope behaviourbased on the various discontinuity datasets was similar. In thethree-dimensional simulations, the orientation variations have alarger effect. Three-dimensional distinct element models based onthe field measurements resulted in a large slope failure, while theone using the terrestrial photogrammetry dataset resulted insmaller slope failures. The model using the COLTOP-3D datasetsresulted in stable slope conditions. These analyses highlight thefact that the stability of the blocks created by repeatingdiscontinuity sets and the topography is sensitive to the evalua-tion of the mean discontinuity set orientation.

The reasons for the variation in discontinuity set orientationare threefold. Firstly, the various techniques sampled differentregions of the landslide and the adjacent area (i.e. geostatisticalvariation of sample; Fig. 3). The field measurements werecollected at a few locations on the lakeshore and in the lowerpart of the failure zone due to access restrictions. The terrestrialphotogrammetry was used to primarily map the headscarp(initiation zone) of the Chehalis Lake landslide. The discontinuityorientation data generated using COLTOP-3D were collected inthe area surrounding the landslide. As a result, the differenttechniques sampled different (photogrammetry vs. field measure-ments) or several (COLTOP-3D) structural domains. Figure 15conceptually illustrates that the dip angle of DS1 changes in thevertical direction within the slope. In reality, the variation of DS1is three-dimensional as both the dip and dip direction change(stereonet in Fig. 15). To further demonstrate the spatial influenceof the dataset, COLTOP-3D was used to extract discontinuityorientation from the point cloud created in the terrestrialphotogrammetry three-dimensional model. The results show that

Table 5 Rock density and discontinuity properties used in the three-dimensionaldistinct element models

Material property

Density (kg/m3) 2,700

Discontinuity propertiesShear stiffness (GPa/m) 1

Normal stiffness (GPa/m) 5

Friction angle (o) 25 or 30

Cohesion (MPa) 0

Tensile strength (MPa) 0

Table 6 Summary of maximumtotal displacements and cumulativevolumes of unstable blocks (blockswith total displacement greaterthan 1 m) obtained in three-dimensional distinct elementmodels

Dataset Frictionangle (o)

DS notincluded

Calculatedmaximumdisplacement(m)

Cumulative volumewith displacement >1 m(Mm3)

Field 25 All included 57 11

Field 30 All included 13 0.00005

Field 25 DS1 3.9 0.000003

Field 25 DS2 53 8.0

Field 25 DS3 64 1.7

Field 25 DS4 57 7.4

Field 25 DS5 71 7.3

Photogrammetry 25 All included 3.3 0.17

COLTOP-3D 25 All included 7.9 0.00072

Original Paper

Landslides 9 & (2012)86

discontinuity sets DS1, DS2 and DS3 had a very similar orientationof the mean values obtained using the 3DM Analyst (AdamTechnology 2009) software, while the dip direction of DS4 and thedip angle of DS5 differ by approximately 20° (Table 7). The resultsobtained when applying COLTOP-3D to the three-dimensional

model created in the terrestrial photogrammetry are moreconsistent with the terrestrial photogrammetry dataset using3DM Analyst (all of them sampling the headscarp area) than thediscontinuity extracted from the airborne LiDAR DEM (samplingthe area surrounding the landslide; Figs. 10 and 16; Tables 1 and7).To fully assess the location/extent of structural domains, asystematic field survey in the various sections of the landslidesand its surroundings would need to be undertaken.

The second factor that may account for the variability indiscontinuity set orientation is the relative accuracy of each of thedifferent data collection methods. Sturzenegger and Stead (2009b)reported that variations in the order of ±4° for the dip and ±8° forthe dip direction can be expected between field measurementsand terrestrial photogrammetry discontinuity survey of the samearea. The third factor influencing the mean discontinuity setorientation includes the biases associated with the different scalesand viewing angles at which the data were collected. For example,the terrestrial photogrammetry was subject to occlusion (shadowzone) due to camera (or laser scanner) perspective (Sturzeneggerand Stead 2009b; Lato et al. 2010) in the lower part of the failureplane orientation (DS1), while the COLTOP-3D data were based onthe airborne LiDAR DEM which is based on data collectedvertically and cannot resolve discontinuities dipping into theslope (e.g. DS5 at the study area). These three factors representsources of uncertainty associated with the data collectionmethodology/processing and the structural model, all of whichtransfer into the slope stability analyses (Read 2009).

Three-dimensional distinct element modelsBased on the limited direct field assessment, the discontinuitysurfaces in the quartz diorite (Fig. 7a) were found to be planar/undulating and rough while the discontinuities in the andesiticrocks were smoother. These general surface conditions wereassumed to correspond to a friction angle of 30°, but the temporalassociation of the failure with the rain-on-snow meteorologicalevent could reduce the mobilized effective friction angle. The3DEC models demonstrated that a friction angle of 30° resulted ingenerally stable slope conditions, whereas a friction angle of 25°resulted in relatively large slope failures for a series of geometries(Table 6). This friction angle threshold (25° to 30°) is consistentwith the dominant planar sliding failure mechanism along DS1which has a mean dip of 28° based on field measurements. Anincrease of pore water pressure associated with rain-on-snowevent could have lowered the effective friction angle below acritical threshold.

The 3DEC model without discontinuity set DS3 (Fig. 14c)led to a slope failure whose location and volume most closelyresemble the actual Chehalis Lake landslide, but since DS3appears prominently in all the stereonet for all discontinuityorientation datasets (Fig. 10), it cannot simply be discounted. Areview of outcrop-scale photographs at the northern edge ofthe landslide (Fig. 7a) showed DS3 surfaces with limitedpersistence, whereas photographs of the southern edge of theslope failure showed DS3 surfaces with a larger persistence(Fig. 8). These observations combined with the coincidence ofvery-high-persistence DS3 with gullies (COLTOP-3D Fig. 9)could represent the location of faults with limited persistence

a Field measurements dataset, 40 m spacing

b Terrestrial photogrammetry, 40 m spacing

c COLTOP-3D, 40 m spacing

051015202530354045505557

00.250.500.751.001.251.501.752.002.252.502.753.003.253.29

00.51.01.52.02.53.03.54.04.55.05.56.06.57.07.57.9

(m)

(m)

(m)

North gullywall

Southgully wall

Valleyslope

Fig. 13 Calculated total displacements, contoured in metre, for the three-dimensional distinct element models using the mean orientation of thediscontinuity sets based on a field measurements, b terrestrial photogrammetryand c COLTOP-3D datasets. A spacing of 40 m and friction angle of 25° wasassumed on all discontinuity sets

Landslides 9 & (2012) 87

DS3 in between them. The northwest strike of DS3 correspondswith the northwest–southeast lineaments previously reportedby Arthur (1986) and the faults mapped by Thompson (1975)

and Mahoney et al. (1995). A more detailed field investigationshould be combined with a comprehensive analysis of thephotogrammetric datasets to confirm this interpretation and to

a DS1 not included b DS2 not included

c DS3 not included d DS4 not included

e DS5 not included

00.250.500.751.001.251.501.752.002.252.502.753.003.253.253.753.86

05101520253035404550556064

05102025303540455053

051015202530354045505557

(m)

(m)

(m)

(m) (m)

051015202530354045505560657071

Fig. 14 Three-dimensional distinct element models investigating the influence of the individual discontinuity sets on the slope stability conditions. Models without aDS1, b DS2, c DS3, d DS4, e DS5. Plots are of calculated total displacement contoured in metre. A spacing of 40 m and friction angle of 25° were assumed for alldiscontinuity sets

Original Paper

Landslides 9 & (2012)88

characterise the variation of the persistence of the discontinuitysets over the landslide area.

Discontinuity set DS1 was shown to be critical to thestability conditions of the modelled slopes. A variation of ±10°in the dip direction and ±15° in the dip between the fieldmeasurements and the terrestrial photogrammetry datasetsresults in a variation of one order of magnitude in calculatedmaximum total displacement and four orders of magnitude incumulative unstable volume in the 3DEC models (Fig. 13;Table 6).

ConclusionsThis paper presented the first detailed assessment of the failuremechanism for the Chehalis Lake landslide. Field observationsindicate that the failure occurred at the intersection of thevalley slope and a prominent north gully wall. The rock massappears to have initially slid into the gully and was thensubsequently redirected along the gully toward the lake. TheChehalis Lake landslide is also temporally associated with arain-on-snow meteorological event wherein high pore water

pressures could have provided the trigger for the rock slopefailure. Discontinuity orientations were collected at the base ofthe landslide, in the headscarp and in the area next to thelandslide using field and remote sensing techniques. The threediscontinuity orientation datasets obtained all produced asimilar fracture pattern with a minor variation in the specificorientation of the individual discontinuity sets. This variabilitywas attributed to the different sampling locations, the accuracyof the different sampling techniques, the observation scale andview orientation biases and data interpretation limitationswhen processing different sample sizes. This highlighted theimportance of combining field and terrestrial photogrammetrydata to avoid sampling biases when defining the meanorientation of discontinuity sets. Kinematic analyses using thethree datasets suggested that planar or wedge failures involvingDS1 were the likely failure mechanisms. Surface wedge analysessuggested that the north gully wall was less stable than thevalley slope. Three-dimensional distinct element modellingsuggested that the Chehalis Lake landslide occurred due tospecific interactions between the discontinuity sets and thetopography. Based on these results, it appears that: (1) planarsliding was the dominant failure mechanism; (2) the mobilizedeffective friction angle along the discontinuities was less than30° for the landslide to initiate, (3) the presence andorientation of DS1 is critical for the development of the slopefailure and (4) DS3 has an important control on the volumeand extent of the unstable rock mass but less so on its overallstability.

AcknowledgementsThe authors would like to thank E. Fea and T. Sivak for theirassistance in the field. Funding for the fieldwork was providedby BC Hydro and BC Ministry of Forests and Range. Theairborne LiDAR data were commissioned and provided by BCHydro. The authors would also like to acknowledge the editorand reviewers for their constructive comments which improvedthe paper.

Table 7 Comparison between the mean orientation of the discontinuity set ob-tained from the terrestrial photogrammetry dataset using 3DM Analyst and CO-LTOP-3D software

Discontinuity set Terrestrialphotogrammetryusing 3DM analyst(dip/dip direction)

Terrestrialphotogrammetry pointcloud using COLTOP-3D(dip/dip direction)

DS1 43/149 46/147

DS2 61/146 60/152

DS3 77/074 70/080

DS4 76/222 72/248

DS5 55/307 74/314

N

S

EW

DATASETField measurements

Terrestrial photogrammetry

Region sampled using terrestrial photogrammetry

Region sampledby field measurements

Fig. 15 Conceptual model illustratinghow the field measurements and thephotogrammetry datasets couldrepresent orientation data collected indifferent structural domains

Landslides 9 & (2012) 89

References

Adam Technology (2009) 3DM CalibCam and 3DM Analyst, version 2.3. Perth, AustraliaArthur AJ (1986) Stratigraphy along the west side of Harrison Lake, southwestern British

Columbia. Geol Survey Can Curr Res 86-1B:715–720BGC Engineering Inc. (2008) Cohoe Creek landslide and landslide induced wave (seiche)

at Chehalis Lake. Report to BC Provincial Emergency Program, January 3, 2008. 31 ppBjerrum L, Jorstad F (1968) Stability of rock slopes in Norway. Norwegian Geotechnical

Institute, Report 79, 12 ppBornhold BD, Harper JR (1998) Engineering geology of the coastal and nearshore

Canadian Cordillera. In: DP Moore, O Hungr (eds), 8th International IAEG Congress.Balkema, Rotterdam, pp 63–75

Brideau M-A, Stead D, Couture R (2006) Structural and engineering geology of the EastGate landslide, Purcell Mountains, British Columbia, Canada. Eng Geol 84(3–4):183–206

Brideau M-A, Stead D, Couture R (2008) The 1999 Clanwilliam landslide: a preliminaryanalysis of potential failure mechanisms. 4th Canadian Conference on Geohazards,Quebec, pp 263–270

Brown ET (2008) Estimating the mechanical properties of rock masses. In: Potvin Y,Carter J, Dyskin AV, Jeffrey R (eds) Southern Hemisphere International RockMechanics Symposium. Perth, Australia, pp 3–22

Cundall PA (1988) Formulation of a three-dimensional distinct element model—part I:a scheme to detect and represent contacts in a system composed of manypolyhedral blocks. Int J Rock Mech Min Sci Geomech Abstr 25(3):107–116

Derron M-H, Jaboyedoff M, Blikra LH (2005) Preliminary assessment of rockslide androckfall hazards using a DEM (Oppstahornet, Norway). Nat Hazards Earth Syst Sci5:285–292

Dunlop S, Hutchinson DJ (2009) Rockslides in a changing climate: establishingmeteorological trigger thresholds in south western Norway. 62nd CanadianGeotechnical Conference, Halifax, Canada, pp 1043–1050

Environment Canada (2009) National Climate Data and Information Archive. http://www.climate.weatheroffice.gc.ca/; accessed September 10, 2009

Evans SG (1989) The 1946 Mount Colonel Foster rock avalanche and associateddisplacement wave, Vancouver Island, British Columbia. Can Geotech J 26:447–452

Geertsema M, Highland L, Vaugeouis L (2009) Environmental impact of landslides. In: KSassa, P Canuti (eds) Landslides—disaster risk reduction. p 589–607

Genevois R, Ghirotti M (2005) The 1963 Vaiont landslide. Giornale de Geologia Applicata1:41–52

Grimstad E (2005) The Loen rock slide—an analysis of the stability. In: Senneset K,Flaate K, Larsen JO (eds) 11th international conference and fieldtrip on landslides.Taylor and Francis, Norway, pp 129–135

Hart RD, Cundall PA, Lemos JV (1988) Formulation of a three-dimensional distinctelement model—part II: mechanical calculations for motion and interaction of asystem composed of many polyhedral blocks. Int J Rock Mech Min Sci GeomechAbstr 25(3):117–125

Hencher SR, Lee SG, Carter TG, Richards LR (2011) Sheeting joints: characterisation,shear strength and engineering. Rock Mech Rock Eng 44(1):1–22

Highland LM, Bobrowsky P (2008) The landslide handbook—a guide to understandinglandslides. UGSG Circular 1325, 129 pp

Hoek E, Brown ET (1997) Practical estimates of rock mass strength. Int J Rock Mech MinSci 34(8):1165–1186

ISRM (1978) Suggested methods for the quantitative description of discontinuities inrock masses. Int J Rock Mech Min Sci Geomech Abstr 15:319–368

Itasca (2008) 3DEC v. 4.1. Itasca Consulting Group, MinneapolisJaboyedoff M, Baillifard F, Couture R, Locat J, Locat P (2004) Toward preliminary hazard

assessment using DEM topographic analysis and simple mechanic modeling. In:Lacerda WA, Ehrlich M, Fontoura AB, Sayo A (eds) Landslides evaluation andstabilization. Balkema, Rotterdam, pp 191–197

Jaboyedoff M, Metzger R, Oppikofer T, Couture R, Derron M-H, Locat J, Turmel D (2007)New insight techniques to analyze rock-slope relief using DEM and 3D-imaging cloudpoints: COLTOP-3D. In: Eberhardt E, Stead D, Morrison T (eds) 1st Canada–US RockMechanics Symposium. pp 61–68

Jaboyedoff M, Couture R, Locat P (2009) Structural analysis of Turtle Mountain (Alberta)using digital elevation model: toward a progressive failure. Geomorphology 103(1):5–16

Jakob M, Weatherly H (2003) A hydroclimatic threshold for landslide initiation on theNorth Shore Mountains of Vancouver, British Columbia. Geomorphology 54:137–156

Lato MJ, Diederichs MS, Hutchinson DJ (2010) Bias correction for view-limited LiDARscanning of rock outcrops for structural characterization. Rock Mech Rock Eng 43(5):615–628

Mahoney JB, Friedman RM, McKinley SD (1995) Evolution of a Middle Jurassic volcanicarc: stratigraphic, isotopic, and geochemical characteristics of the Harrison LakeFormation, southwestern British Columbia. Can J Earth Sci 32:1759–1776

Marinos P, Hoek E (2000) GSI: a geologically friendly tool for rock mass strengthestimation, GEOENG 2000. Technomic, Melbourn, pp 1422–1440

Marinos V, Marinos P, Hoek E (2005) The Geological Strength Index: applications andlimitations. Bull Eng Geol Environ 64:55–65

Massey NWD, MacIntyre DG, Desjardins PJ, Cooney RT (2005) Digital geology map ofBritish Columbia: tile NN10 Southwest British Columbia, BC Ministry of Energy andMines, Geofile 2005–3

Mathews WH, McTaggart KC (1978) Hope rockslides, British Columbia, Canada. In: BVoight (ed) Rockslides and Avalanches. Elsevier, New York, pp 259–275

N

S

EW

DS5

DS5 DS4

DS4

DS3

DS3

DS1

DS1

DS2

DS2

N

S

EW

DS5

DS5 DS4

DS3

DS1

DS1

DS2

DS4

DS2

DS3

Terrestrial photogrammetryusing 3DM Analyst

Terrestrial photogrammetrycloudpoint using COLTOP-3D

N = 202 polesto fitted circlesPhotogrammetryline of sight

N = 8727 polesto DEM cellPhotogrammetryline of sight

a

b

Fig. 16 Stereonets comparing the results obtained from the terrestrialphotogrammetry dataset using the a 3DM-Analyst and b COLTOP-3D software

Original Paper

Landslides 9 & (2012)90

Mathis JI (1988) Development and verification of a three dimensional rock joint model.Ph.D. thesis, University of Lulea, Sweden, 144 pp

Metzger R, Jaboyedoff M, Oppikofer T, Viero A, Galgaro A (2009) Coltop3D: a newsoftware for structural analysis with high resolution 3D point clouds and DEM,frontiers + innovation. Canadian Society of Petroleum Geologists, Canadian Societyof Exploration Geophysicists and Canadian Well Logging Society Joint Convention,Calgary, Alberta, pp 278–281

Monger JWH (1970) Hope map-area, West half (92 H W1/2), British Columbia, Paper69–47. Geological Survey of Canada, Ottawa, Ontario, 75 p

Oppikofer T (2009) Detection, analysis and monitoring of slope movements byhigh-resolution digital elevation models. PhD thesis, Université de Lausanne,193p

Oppikofer T, Jaboyedoff M, Blikra L, Derron M-H, Metzger R (2009) Characterization andmonitoring of the Aknes rockslide using terrestrial laser scanning. Nat Hazards EarthSyst Sci 9:1003–1019

Panizzo A, de Girolamo P, di Risio M, Maistri A, Petaccia A (2005) Great landslide eventsin Italian artificial reservoirs. Nat Hazards Earth Syst Sci 5:733–740

Peckover FL, Kerr JWG (1977) Treatment and maintenance of rock slopes ontransportation routes. Can Geotech J 14(4):487–507

Pedrazzini A, Jaboyedoff M, Froese C, Langenberg W, Moreno F (2008) Structures andfailure mechanisms analysis of Turtle Mountain. 4th Canadian Conference onGeohazards, Quebec, pp 349–356

Plafker G, Eyzaguirre VR (1978) Rock avalanche and wave at Chungar, Peru. In: B Voight(ed) Rockslides and avalanches, vol. 2 Engineering Sites, pp 269–279

Read JRL (2009) Data uncertainty. In: JRL Read, PF Stacey (eds) Guideline for Open PitSlope Design. CRC, pp 213–220

Richards LR, Leg GMM, Whittle RA (1978) Appraisal of stability conditions in rock slopes.In: Bell FG (ed) Foundation engineering in difficult ground. Newnes-Butterworths,London, pp 449–512

Rocscience (2006) Swedge v5.0. Rocscience, TorontoSchwab JW (1999) Tsunamis on Troitsa Lake British Columbia. Forest Service British

Columbia Extension Note #35, 4 ppSidle RC, Ochiai H (2004) Landslides: processes, prediction and land use. AGU Water

Resour Monogr 18:312Stauffer MR (1966) An empirical–statistical study of three-dimensional fabric diagrams

as used in structural analysis. Can J Earth Sci 3:473–498Stephenson FE, Rabinovich AB (2009) Tsunamis on the Pacific Coast of Canada recorded

in 1994–2007. Pure Appl Geophys 166:1777–210Sturzenegger M, Stead D (2009a) Quantifying discontinuity orientation and persistence

on high mountain rock slopes and large landslides using terrestrial remote sensingtechniques. Nat Hazards Earth Syst Sci 9:267–287

Sturzenegger M, Stead D (2009b) Close-range terrestrial digital photogrammetry andterrestrial laser scanning for discontinuity characterization on rock cuts. Eng Geol106:163–182

Sturzenegger M, Stead D, Beveridge A, Lee S, Van As A (2009) Long-range terrestrialdigital photogrammetry for discontinuity characterization at Palabora open-pit mine.In: Diederichs MS, Grasselli G (eds) Third Canada–US Rock Mechanics Symposium.Paper 3984, Toronto, 10 pp

Thompson RW (1975) Petrologic study of lower amphibolite metamorphism of igneousrock, Chehalis Lake area near Harrison Hot Springs, British Columbia. B.Sc. thesis,University of British Columbia, Vancouver, 44 pp

van Zeyl D (2009) Evaluation of subaerial landslide hazards in Knight Inlet and HoweSound, British Columbia. M.Sc. Thesis, Simon Fraser University, 184 pp

Wang FW, Zhang YM, Huo ZT, Matsumoto T, Huang BL (2004) The July 14, 2003Qianjiangping landslide, Three Gorges Reservoir, China. Landslides 1(2):157–162

M.-A. Brideau : M. Sturzenegger : D. Stead : N. J. Roberts : B. C. Ward :J. J. ClagueDepartment of Earth Sciences,Simon Fraser University,Burnaby, Canada

M. JaboyedoffInstitute of Geomatics and Risk Analysis,Université de Lausanne,Lausanne, Switzerland

M. LawrenceBC Hydro,Burnaby, Canada

T. H. MillardBritish Columbia Ministry of Forests and Range,Nanaimo, Canada

M.-A. Brideau ())School of Environment,University of Auckland,Auckland, New Zealande-mail: m.brideau@auckland.ac.nz

Landslides 9 & (2012) 91