The Zymoetz River rock avalanche, June 2002, British Columbia, Canada

The Zymoetz River rock avalanche, June 2002, British Columbia, Canada

Engineering Geology 83 (2006) 76 – 93 The Zymoetz River rock avalanche, June 2002, British Columbia, Canada N. Boultbe...

1MB Sizes 0 Downloads 15 Views

Engineering Geology 83 (2006) 76 – 93

The Zymoetz River rock avalanche, June 2002, British Columbia, Canada N. Boultbee a, D. Stead a,*, J. Schwab b, M. Geertsema c a

Department of Earth Sciences, Simon Fraser University, Burnaby, BC, Canada b BC Ministry of Forests and Range, Smithers, BC, Canada c BC Ministry of Forests and Range, Prince George, BC, Canada Accepted 24 June 2005 Available online 23 November 2005

Abstract On June 8, 2002, the Pacific Northern Gas pipeline in the Zymoetz River valley was severed by a large debris flow. The event initiated as a rock avalanche in Glen Falls Creek, a tributary of the Zymoetz River. The rock avalanche involved 1  106 m3 of volcaniclastic bedrock, and travelled through a complex flow path, to finally deposit a large fan in the main Zymoetz River. Approximately half of the debris volume was deposited in the cirque basin at the head of the valley, with the rest deposited in the channel, and the fan. Examination of the initiation zone showed a very persistent, slightly curved, joint set that forms the main sliding surface for the failed block with a dip of 458, and dip direction of 3008. A Geographic Information System (GIS) was used to examine the event and allowed for further interpretation of field data. Preliminary dynamic analysis indicates that the event reached velocities of up to 34 m/s. Comparison of the Zymoetz River rock avalanche (ZRRA) with other similar events from the literature indicates that it exhibited similar mobility and velocities. As evidenced from the literature, these long runout events can cause significant damage, and have the potential to be a very high risk as forestry and recreation activities spread further into remote areas. D 2005 Elsevier B.V. All rights reserved. Keywords: Rockslide; Rock avalanche; Runout

1. Introduction Mass movements represent a significant hazard in the Coast Mountains of British Columbia. At approximately 1:30 am on June 8, 2002, the Pacific Northern Gas (PNG) pipeline in the Zymoetz River Valley (also known as Copper River) near Terrace, BC, was broken by a large debris flow that traveled down Glen Falls Creek (local name), a tributary of the Zymoetz River * Corresponding author. E-mail address: [email protected] (D. Stead). 0013-7952/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.enggeo.2005.06.038

(Fig. 1). The debris flow started after a rock avalanche impacted a cirque basin at the top end of the valley 4.5 km away. The Zymoetz River rock avalanche (ZRRA) is part of a growing number of large mass movements that have occurred in BC in the past 50 years. This includes four recent events occurring in the spring and summer of 2002; the Harold Price, June 23, 2002 (Schwab et al., 2003), Pink Mountain, July, 2003 (Geertsema et al., 2006—this issue-b), McAuley Creek April/May, 2003 (Evans et al., 2003), and a not yet examined event near Kitimat, BC (Geertsema et al., 2006—this issue-a) (Fig. 1). The Howson rock ava-

N. Boultbee et al. / Engineering Geology 83 (2006) 76–93


Fig. 1. Location map of the Zymoetz River Rock Avalanche, and five other large rock avalanches that have occurred in British Columbia in the past 4 years.

Fig. 2. Aerial photograph and longitudinal profile of the ZRRA. The longitudinal profile is divided into the detachment, transport, and deposition zones. The numbers on the long profile show the locations of the cross-sections through the channel (see Fig. 13). The detached mass is outlined by the dashed line in the detachment zone. The curves used for velocity calculations are marked on the aerial photo (Table 1).


N. Boultbee et al. / Engineering Geology 83 (2006) 76–93

lanche, located within 100 km of the Zymoetz River and the Harold Price avalanche, occurred in September of 1999 (Schwab et al., 2003). Failures such as these, are a major hazard, and have the potential to comprise a very high risk, if they occur in populated areas or transportation corridors. Even if failures occur in unpopulated areas, they still constitute a risk to fisheries and forestry activities. Of the most hazardous failures, debris flows and debris avalanches can be particularly dangerous due to their high velocities, and frequent long runout distances (Hungr and Evans, 2004; Crosta and Agliardi, 2003; Evans et al., 1989). This paper presents a characterization of the Zymoetz River rock avalanche (ZRRA), including descriptions of the rock and soil characteristics, a summary of analytical techniques employed, a debris flow path description, preliminary dynamic appraisal of the event and finally a brief comparison with other large landslides. The characterization of the event will follow the format used by Couture et al. (1999), and Evans et al. (1989) which divides the flow path into the detachment, transport and deposition zones (Fig. 2). The term rock avalanche is used to describe this complex event, although it displayed characteristics of a rockslide, and debris flow at different stages during failure (Cruden and Varnes, 1996). Fig. 2 illustrates the path division into the different stages. 2. Physical setting of the Zymoetz River event 2.1. Physiography and climate The ZRRA occurred near Terrace in west central British Columbia, approximately 65 km east of Prince Rupert. The town of Terrace gets its name from the Skeena River terraces upon which the city is built. Precipitation in the region is typical of Coastal BC, with mean annual precipitation on the order of 2500 mm, but reaching 3500 mm in some areas. In the Kitsumkalum–Kitimat trough (Fig. 1), within which Terrace is situated, precipitation is significantly lower, with a mean annual rate of 1000–2500 mm/year (Clague, 1984). The high precipitation in the region supports a temperate rainforest, with the main tree species being Western Hemlock, with Western Red Cedar and Sitka Spruce, typical of the Coastal Western Hemlock Biogeoclimatic zone (Meidinger and Pojar, 1991). 2.2. Geology of the study area The ZRRA is located near the western edge of the Intermontane belt, within the Stikinia Terrane. The

basement of Stikinia is composed of Devonian and Permian arc volcanics with platform carbonates. These rocks are overlain by Triassic and Lower Jurassic arc volcanics, volcaniclastics, chert and arc-derived clastics that are intruded by plutonic rocks (GSC, 1992; Mihalynuk, 1987). The main rock Group in the Terrace area, and thus the Zymoetz River Valley, is the Hazelton Group, with the largest portion being comprised of the Telkwa Formation which is Triassic in age and of nonmarine origin. The Telkwa Formation consists of reddish, maroon, purple, grey and green pyroclastic and flow rocks. Zeolite alteration is widespread in the formation with locally extensive veining and formation of zeolite cemented breccias. Other metamorphic minerals found in the formation include epidote, calcite and quartz. Rock types observed in the Glen Falls creek valley include green and purple volcaniclastics, and buffcoloured limestone. Field investigation showed that the initiation zone occurs in volcaniclastic rocks on the east side of the valley, and that the limestone only outcrops on the west side. The geology in the valley is complex and highly variable over short distances, related to a complex depositional history. Within the valley, Glen Falls Creek has taken advantage of the weakness offered by the main contact between the volcanics and the limestone, and flows along this contact in the middle of the valley. According to the British Columbia Geological Survey (BCGS) Geology Map, there are numerous faults in the Zymoetz River Valley, including thrust and normal faults, although the majority of faults are shown as unidentified (Fig. 3) (BCGS, 2000). Although not shown on the map, it is likely that the Zymoetz River follows faults, due to the distinct shape of the river (Fig. 1). The map also shows a relatively persistent fault running along the east side of Glen Falls Creek that potentially coincides with the main failure plane for the initial rockslide. 2.3. Geomorphology and quaternary geology Glen Falls Creek is a north flowing tributary of the Zymoetz River, which flows into the Skeena River, approximately 10 km north east of Terrace, BC. Terrace itself is located in the Kitsumkalum–Kitimat trough, built on terraces of the Skeena River (Fig. 1). Surficial materials in the area include till, colluvium, and exposed bedrock on hillslopes, with waterlain deposits such as glaciomarine and fluvial and glaciofluvial sediments in valleys (Clague, 1984). The glaciomarine

N. Boultbee et al. / Engineering Geology 83 (2006) 76–93


Fig. 3. BCGS hillshaded DEM showing the large fault in the Glen Falls Creek drainage (dashed box) passing through the area of the headscarp (H) of the ZRRA. Arrows outline the debris flow path. This fault is unidentified as to movement direction. Modified after BCGS (2000).

deposits in the area are prone to large earthflows (Geertsema and Schwab, 1996). Locally, other glacial sediment deposits are found, including kame terraces, and glaciolacustrine sediments (Clague, 1984). According to Clague (1984), landslides, including debris flows, are common in the area, although typically small in size.

Failures are generally located in areas of high precipitation (N2500 mm/year). Recently, two large rock avalanches have occurred within 100 km of the Zymoetz River event; the Howson avalanche on September 11, 1999, and the Harold Price avalanche on June 22 or 23, 2002 (Schwab, 2002;

Fig. 4. Two diamicts are found in Glen Falls Creek valley. The black line separates the reddish till on the top, from the green old debris flow deposit. Man in picture is standing on till that has sloughed off of the channel wall.


N. Boultbee et al. / Engineering Geology 83 (2006) 76–93

Schwab et al., 2003) (Fig. 1). The Howson event is similar to the ZRRA in that it severed the Pacific Northern Gas pipeline through the Telkwa pass. The Howson event involved 0.9  106 m3 of granodioritic rock, and initiated as a toppling failure onto a glacier. The Harold Price rock avalanche started as a failure of ice and volcanic bedrock rubble from a rock glacier, and quickly transformed into a debris flow involving a large volume of timber and forest debris, travelling 4 km (Schwab et al., 2003). Within Glen Falls Creek, there is evidence of past instability, including old levees, and diamict deposits within the channel that appear to be debris flow deposits. This diamict is green in colour (similar to the colour of the volcanic bedrock in the valley), and composed of layers of granule to small boulder sized, angular clasts (Fig. 4). This first diamict is covered by another diamict that is interpreted as a till. The till is a reddish colour and is typical of tills in the area that are rich in Telkwa volcanics. The Telkwa volcanics contain a deep red-coloured ash tuff that is very friable, and disintegrates readily, thus defining the colour of the till in the region. The till in the Glen Falls Creek Valley is up to 20 m thick in places, and covers the central part of the valley. It maintains very steep angles (808–908) due to the significant fines content (plasticity index = 18). Towards the northern end (mouth of the valley), the sediments include some lacustrine-like deposits of bedded silt and sand, and a 3-m-thick section of silt and clay. 3. Event description At approximately 1:30 am on June 8, 2002 the Pacific Northern Gas (PNG) pipeline in the Zymoetz River Valley was severed over an estimated length of several tens of meters at the mouth of Glen Falls Creek by a large debris flow (Cavers, 2003; Schwab et al., 2003) (Fig. 5). The event initiated as a rock avalanche when a 1  10 6 m3 block of bedrock detached from the eastern wall of the cirque basin at the head of the valley. The rock avalanche travelled over snow in the cirque basin for approximately 650 m and then dropped out of the basin and proceeded down Glen Falls Creek. The Zymoetz River rock avalanche (ZRRA) travelled a vertical distance of 1245 m, from 1390 m el. to 145 m el., over a horizontal distance of 3.5 km, and a path length of 4.2 km (Fig. 2). Readings from the Water Survey of Canada station (Station No. 08EF005) located 3 km downstream on the Zymoetz River indicate a large turbidity spike, and a sharp decrease in water level

Fig. 5. Glen Falls Creek channel above the canyon. Note the superelevations in the channel. The headscarp can be seen at the top, center of the photo (arrow).

associated with the event (Fig. 6). The gauge shows a smaller second event that occurred at approximately 9:30 am on June 8. This event was also witnessed by PNG personnel that were investigating the damage to the pipeline. The WSC station also collects water samples with an ISCO water sampler when the turbidity reaches a given value. The turbidity spikes were correlated to a suspended sediment load of approximately 3600 to 12 000 mg/L. The ZRRA path is complex; the event initiated as a rockslide into the cirque basin, then transformed into a debris flow/avalanche as it melted snow and incorporated water, as well as trees and other debris. The ZRRA behaved fluidly as it progressed down the valley, with superelevated curves and mud splashes observed on trees up to 60 m above the base of the creek channel (Fig. 5). The ZRRA then impacted an almost 908 corner and traveled through a constricted bedrock canyon, and finally over a 10-m-high waterfall into the main Zymoetz River valley (Fig. 7). Erisman and Abele (2001) describe similar shapes of debris flows in Europe and North America. Investigations into possible triggers for the slide event revealed that there was no seismic activity near

N. Boultbee et al. / Engineering Geology 83 (2006) 76–93


Fig. 6. Water Survey of Canada Station No. 08EF005 readings for the Zymoetz River on June 8, 2002. The two separate events can clearly be seen in the drop in water level, and large increase in turbidity (Schwab et al., 2003).

the time of failure, and no anomalous precipitation recorded. The climatic conditions for 2 weeks prior to the event included significant precipitation and colder temperatures than normal for that time of year. The days before the event saw cool temperatures with snow falling at higher elevations (Schwab et al., 2003) (Fig. 8). Failure was probably a result of progressive, long-term degradation of the tectonically deformed and altered rock mass. The influence of freeze– thaw and changes in pore water pressures cannot be discounted in the gradual reduction of the shear strength along the eventual failure surface.

3.1. Analytical techniques A variety of data collection and analysis methods were used to characterize the ZRRA. The techniques employed included a Geographic Information System (GIS) and the block size distribution program, WipFrag (WipWare Inc., 2003). Four days after the event occurred, on June 12, 2002, high-resolution air photos were flown off the valley. Using the post-failure photos along with air photos from 1975, contour and image files were created that are suitable for use in a GIS. Fieldwork was completed in the summer of 2003,

Fig. 7. The large fan deposited in the Zymoetz River by the ZRRA, and the constricted bedrock canyon that the debris flowed through. Note the dark patch of trees around the mouth of the canyon. These trees were burnt by the fire that started as a result of the gas pipeline burst.


N. Boultbee et al. / Engineering Geology 83 (2006) 76–93

Fig. 8. Graph of temperature and precipitation in the days before and after the ZRRA. The days before the event show colder than average temperatures, and significant precipitation for 2 weeks prior to the event (modified after Schwab et al., 2003).

during which time photographs were taken specifically for use in the program WIPFRAG Discontinuity surveys were also conducted for use in a stereographic analysis of the discontinuities. The GIS was created with the ESRI product ArcGIS (ESRI, 2002). ArcGIS contains a variety of tools that allow different options for data interpretation. Using ArcMap, the data are displayed in 2D, and distance and volume measurements are obtainable. A digital elevation model (DEM) can be created using the contour files. In ArcScene, the data can be displayed in 3D and rotated for different views, and this proved extremely useful. A digital image was draped over top of the DEM, providing a detailed 3D view of the area. With high quality data, and a 3D model that can be rotated, repeated observations at different view points and elevations are possible. The 3D viewing capabilities improved observation of joint sets on inaccessible parts of the failure plane, as well as allowing for calculated estimates of the volume of the failed block. Fig. 9 shows a 3D block contour model of the cirque wall before and after the ZRRA produced using the GIS. WipFrag is an image analysis software tool that allows estimation of preliminary block size distribution curves using a photograph with a user-defined scale. The program can determine the size of particles in the photograph, and hence output a block/grain size distribution curve. During fieldwork, photographs were taken with a 1-m square for scale, to be used in WipFrag (Fig. 10). This program has been utilised extensively in blasting to determine the size of fragments that remain after a blast, and thus control blast

efficiency. The limitations of photographic techniques in the estimation of absolute grain sizes have been noted by several authors including Maerz and Zhou (1998). In the ongoing investigation of this landslide, it is intended that the program will be used to determine the brelative changesQ in measurable block size and debris fragment distributions from the beginning of deposition in the basin to the fan at the end of the event. Similar studies at the Frank slide were undertaken by Couture et al. (1999) and they may allow indications of energy changes during the transport and comminution of rockslide debris (Locat et al., 2003). Results derived from the WipFrag analysis include a grain size distribution curve, and the calculated Dvalues. The D-values may then be graphed according to location within the channel, both across the channel, at selected cross-section locations, and longitudinally down the channel. Fig. 10 shows the changes in grain size with changing location across the channel at crosssection 10. The station numbers refer to the picture number taken every 10 m along the cross-section. A detailed analysis of the WipFrag data is ongoing, and the subject of a future publication. 3.2. The rockslide initiation zone The rockslide initiated as a large mass of volcaniclastic bedrock detached from the northeast wall of the cirque basin at an approximate elevation of 1390 m. The headscarp is 125 m long at the top of the rupture surface, with a maximum width of 300 m. The failure involved a main sliding surface with a rear release plane

N. Boultbee et al. / Engineering Geology 83 (2006) 76–93


Fig. 9. (a) Three-dimensional plot derived from the 2 m contour lines of the ZRRA headscarp area and basin. D is the mass that failed during the event. E is the old failure surface (see also Fig. 11). F shows the westerly trending linears that parallel the joints that form the lateral release for the headscarp. G shows the northerly trending cliff faces that parallel the main failure surface. (b) Three-dimensional plot of the post-event 2 m contours. Also seen on this plot are contoured stereonets, clockwise from i to iv: i) shows the high concentration of surfaces making up the main failure plane and backscarp; ii) shows the joint orientation in outcrops above the failure surface; iii) shows the joint orientations in the outcrop on the north side of the creek (closest to the failure surface), the orientations are similar to those in the other outcrops above the headscarp, and could be associated with a lateral release surface for the failure; iv) the joint orientations in the outcrop on the south side of the creek, reflect a different stress regime during their genesis.

(Fig. 11). The main sliding surface is a very persistent (~ 200 m long), slightly curving surface dipping at 438– 458, with a dip direction of 3008, while the backscarp is sub-vertical and appears highly sheared in places (Fig. 11). Immediately next to the fresh failure, there is an old failure scar that can be seen clearly in old airphotos, and in the 3D contour image (Fig. 9). From the BCGS Geology Map, it is seen that a large, persistent fault of unknown relative displacement is located along the east side of Glen Falls Creek

(BCGS, 2000). The location of this fault very closely coincides with the initiation zone (Fig. 3). The presence of a fault also explains the intense shearing that was observed in the backscarp of the failure plane. In addition to the linear north–northwest trending feature, additional westerly trending linears can also be identified in Fig. 9(F). These west-trending features appear to be of the same orientation as the creek that flows along the side of the headscarp (Figs. 9 and 11). Closer examination also shows northerly trending cliff faces


N. Boultbee et al. / Engineering Geology 83 (2006) 76–93

Fig. 10. Graph showing preliminary results from the WipFrag analysis. A sample picture used in the analysis with the 1 m reference frame is shown in the upper right corner of the graph.

below the failure surface that parallel the failure plane/ fault (G). Discontinuity surveys were conducted in the area of the headscarp, both along the backscarp, and a small portion of the main sliding surface, and along outcrops around the headscarp to gather data for the preliminary kinematic analysis. Fig. 9b(i) shows the strong concentration of surfaces that formed the main failure surface. Surveys of other outcrops above the headscarp show a general north–northeast dip direction which would be the orientation required for lateral release surfaces associated with the failed block (Fig. 9b(ii)). The joints measured had varied persistence (0.3 m–4 m) and spacing (b 0.01 m–1 m), and commonly had calcite or epidote on surfaces. The alteration and consequent reduction in strength of the rock mass may have been a factor in the location of the initial rockslide. There were also a significant number of slickensided surfaces. Immediately to the south of the headscarp is a westerly flowing creek. Discontinuity surveys in the creek show that the creek follows a contact between different volcaniclastic rocks, and is probably another fault, as it was observed that there are very different structural domains represented on either side of the creek (Fig. 9b(iii) and (iv)). Approximately half way down the creek, there is another contact that could be the continuation of the fault that coincides with the failure plane. It was noted that there was evidence of shearing in this area, including numerous slickensided surfaces. Access to the headscarp was by helicopter. The failure plane itself was both dangerous and difficult to

access, as the fresh surface still exhibited ravelling of loose material. A large block and considerable failure debris remain perched on the failure surface. Estimates of the volume of the failed mass of approximately 1  106 m3 were obtained using the Geographic Information System code ArcGIS. The main failure plane can be traced under the backscarp to a small depression on the surface above the headscarp. In November of 2003, examination of the area by helicopter showed evidence of seepage onto the failure plane, from under the backscarp. While the carbonate bedrock in the valley is intact (geological strength index (GSI) 70–90), the volcanic bedrock varies from blocky (GSI 80–95) to disintegrated (GSI 25–35) (Marinos and Hoek, 2004; Hoek et al., 2002). The bedrock in the vicinity of the headscarp is also highly fractured; however, there are several prominent joint sets, including the joint sets that form the main failure plane, and the backscarp of the failure plane (Fig. 11). There are also some indications of tectonic deformation near the failure plane. The rock mass forming the backscarp itself appears to be moderately sheared with an estimated GSI of 10–20. Rock samples were obtained from the backscarp of the failure plane and the slide debris for laboratory testing. Estimates of compressive strength were made according to geological hammer and point load tests on hand samples. Point load tests conducted according to the ISRM suggested methods (1985) on debris blocks indicate an average uniaxial compressive strength of approximately 150–200 MPa, or an R-value of R4. Rock samples from the backscarp

N. Boultbee et al. / Engineering Geology 83 (2006) 76–93


Fig. 11. Two views of the headscarp of the ZRRA. A is the main sliding surface. B is the backscarp. C is debris remaining on the slope. D are large blocks of still intact bedrock, that remains attached to the slope. E is the old failure surface. The hatch-marked sections indicate vegetated areas that were not involved in the failure. Lettering also applies to Fig. 9.

however were less competent (due to shearing) and disintegrated during transport to the laboratory (estimated value of R2/R3). Thin sections were prepared and within the backscarp rock micro-fractures can be seen at a spacing of 2–5 mm.

3.3. The cirque basin After detachment, the failed mass impacted the cirque basin at an elevation of approximately 900 m, and travelled across a 3-m-thick deposit of snow (Fig.


N. Boultbee et al. / Engineering Geology 83 (2006) 76–93

12). (The cirque basin varies in elevation from 940 m el. to 830 m el. over a distance of 600 m.) From the aerial photo, streak marks can be seen on the snow on the west side of the basin, where the debris dskatedT over the snow and accelerated down valley (Fig. 2). From the top of the basin, the debris travelled over a moraine and then subdivided and funnelled down both sides of a forested island. The forested island itself is formed by a resistant volcanic bedrock knob that forms a knick point in the channel. The elevation drop out of the basin is approximately 150 m over a distance of 250 m (Fig. 2). A large volume of material was deposited in the cirque basin, with measurements from the GIS giving a volume estimate of 500,000 m3, which is approximately half the volume of the failed mass. This estimate was obtained by measuring the basin from the far south end of the debris to the top of the forested island, and the width from west to east wall. The debris depth was assumed as 3 m, based on the average depth that was visible on top of the snow, and the size of some of the very large boulders found in the basin. The failure debris that was deposited in the basin consists entirely of fragmented bedrock from the detached mass. The rocks are very angular and sharp, with the main block size being boulders, with a significant number of very large blocks, some measuring up to 14 m long. Very commonly the rocks have calcite or

epidote, and possible zeolite on surfaces, or contain veins of these minerals. The larger boulders typically exhibit joint bounded surfaces with coatings of calcite and epidote, up to 0.5 cm thick. Estimates of the snow pack in the basin prior to the slide are unavailable, but as approximately 3 m of snow remained under the debris after the event, there was probably little entrainment of snow and no entrainment of debris, at least at the far south end of the basin. A contribution of the snow at the south end of the basin to the debris acceleration could have been a reduction in the surface frictional resistance. As the event moved further north in the basin, and encountered the moraine, saturated sediment could have been entrained, in addition to some snow entrainment. As the debris towards the north end of the basin is considerably thicker, estimates of any remaining snow under that portion of the basin could not be made. Snow pillow data from the closest station (Tsai Creek, 1360 m el., 54839V, 127840V) indicated a snow depth of only approximately 2 m; therefore, it is likely that the snow depth in Glen Falls Creek was not significantly greater than the 3 m that remained after the event. It is likely that some entrainment of debris did occur in the basin, as samples of the debris that were collected immediately below the island had low plasticity values, and D50 values of ~ 15 mm, indicating substantial entrainment of fines.

Fig. 12. The landslide deposit on top of approximately 3 m of snow. Person in photo for scale.

N. Boultbee et al. / Engineering Geology 83 (2006) 76–93

3.4. Main valley transport stage Upon exiting the cirque basin at 690 m el., the debris divided on either side of the forested island shown in Fig. 2. The majority of the rock avalanche debris emerged on the north side of the forested island and continued down valley to the Zymoetz River at 160 m el. The approximate travel distance between the cirque basin and the Zymoetz River was 2.6 km. Throughout the main path the debris exhibited high fluidity, with mud splash marks observed on trees up to 60 m above the base of the channel. The extreme mobility of the debris is also seen in superelevations along the length of the channel, some up to 40 m above the valley floor (Fig. 13). The width of the main transport zone is 45 m– 90 m. This is compared to a pre-ZRRA channel width of approximately 10 m, obtained from the 1975 airphotos, indicating that the landslide removed a significant volume of timber as it progressed down the valley.


It is unclear how much material was entrained during the event, as very little evidence of the red till was found in the debris. The authors suggest that minimal amounts of debris were entrained, due to the cohesive nature of the valley fill. While some till rip-up clasts were discovered in the deposit, the deposit was mostly composed of sand to cobble sized volcanic bedrock, with few boulders, and a moderate amount of wood fragments, and with moderate fines content (plasticity index = 2–8). The material that was entrained would have consisted of the soil cover that housed the vegetation in the valley, however, with the colluvial cover that now exists it is very difficult to estimate the exact amount of erosion that occurred. Similar difficulty in estimating entrainment is noted by Hungr and Evans (2004). In the field, deposition in the valley was evident as distinct lobes/zones of debris. The lobes were defined by a length and width and an estimated average depth

Fig. 13. Cross-sections in Glen Falls Creek valley, showing the superelevations of the debris. Locations of cross-sections are shown in Fig. 2.


N. Boultbee et al. / Engineering Geology 83 (2006) 76–93

Fig. 14. Aerial photo with deposition lobes, showing Glen Falls Creek from the forested island to the start of the canyon. The legend indicates the debris depth for the different shades. Flow direction is from left to right (to the north).

(Fig. 14). The deposits in the channel ranged in depth from a thin veneer to 4 m deep, with the average being about 1.5 m. Using the GIS the estimate of material deposited in the channel is approximately 200,000 m3, which agrees closely with the estimate of 190,000 m3 made from measuring debris lobes in the field. The debris deposited in the channel is generally more compact and coherent than that found in the basin, as there is a higher concentration of finer-thanboulder material. In addition, the channel deposits have been washed by Glen Falls Creek as the creek attempted to re-establish itself after the complete destruction of the channel. The main grain size in the channel deposits is cobble sized, although there are still a few large boulder size blocks (up to 6 m long) throughout the channel. 3.5. Landslide depositional zone Deposition of debris occurred along the entire length of Glen Falls Creek, from the basin to the channel, with the majority of debris deposited in the form of a large fan in the Zymoetz River (Fig. 7). According to the

estimates of minimal entrainment, the volume of the deposits should be similar to the volume of the detached mass. Prior to this event, photographic evidence shows that there was no fan in the Zymoetz River at the mouth of Glen Falls Creek; however, this section of the river has always been a sediment storage area (Fig. 15). The newly formed fan spans the river, with a width and length of approximately 250 m. The fan initially blocked the river, as evidenced by the severe drop in water level measured by the WSC station (Fig. 6), but was quickly overtopped. Water upstream of the fan has remained ponded, however, and the forest service road upstream has been elevated by up to 3 m in places. The depth of the fan is difficult to estimate, but may be up to 10 to 12 m in places. Large boulders, up to 10 m wide, were carried full length from the channel to the fan. Estimates of the dimensions from the GIS provide a volume of material deposited in the fan as approximately 625,000 m3, assuming an average depth of 10 m. Examination of the deposits on the fan show that the debris is composed mostly of fragmented volcanic bedrock, of sand, gravel and cobble size, with a few large

Fig. 15. Series of air photos showing that the mouth of Glen Falls Creek has always been a sediment storage area in the Zymoetz River, but no fan existed prior to the ZRRA.

N. Boultbee et al. / Engineering Geology 83 (2006) 76–93

boulder size blocks. The main grain size in the fan is cobbles, and the majority of grains are subangular to subrounded. There are some limestone grains (b10%), as well as a small amount of granite (b 2%). While the majority of material in the fan is cobble sized and smaller, there are a few very large boulders that were transported the entire 4.5 km by the event. The debris blocks observed in the main river are up to 10 m long, the same size as many of the large boulders found in the basin. From the estimates of the material deposited during the event, the total deposit has a volume of approximately 1.3  106 m3, while the calculated volume of the displaced mass is estimated at 1.0  106 m3. According to Hungr and Evans (2004), a bulking factor of 25% due to fragmentation can be applied to debris in rock avalanches. From that value, the bulked deposited volume of material would be 1.25  106 m3. Errors involved in the calculations are due to the inherent difficulties in the estimation of volumes of the detached mass, entrained material and the depth of the fan.

at the canyon, where the debris turns two almost right angle corners (Fig. 2). It is recognised that there may be errors associated in maximum velocity determinations associated with the derivation of the maximum vertical run-up, h (Erisman and Abele, 2001). The maximum velocity in this paper is calculated using the same technique as Evans et al. (1989) in order to allow comparison with the Pandemonium Creek event. This yielded velocities of 20 m/s to 34 m/s (Table 1). Velocity was also estimated using superelevation data and Eq. (2): v2 ¼ rc g tanh cosa


where r c is the radius of curvature, h is the transverse slope and a is the longitudinal slope (Hungr et al., 1984). Cross-sections were drawn to obtain the superelevation data (Fig. 13), and the radius of curvature was measured for curves from the GIS and printed maps. The velocities that were obtained from Eq. (2) ranged from 14 m/s to 26 m/s (Table 1). Velocity calculations were also undertaken for the run-up in the basin using Eq. (2), which yielded a velocity of 26 m/s, while Eq. (1) gave 34 m/s. Measurement error associated with Eq. (2) exists as determining the radius of a curve from an airphoto is subject to limitations. The authors suggest that although the velocity of the debris avalanche would have varied considerably within the transport zone (owing to the varying lateral confinement and funnelling of the flow), a range from 15 m/s to 25 m/s is probable.

4. Preliminary landslide velocity estimates The velocity calculations for the ZRRA were undertaken using both run-up and superelevation data and the velocity potential energy equation (Evans et al., 1989; Jordan, 1994) v2 ¼ 2gh



where h is the maximum vertical run-up and g is acceleration due to gravity (Table 1). This formula is used where debris impacts a surface that is almost perpendicular to the flow direction. For this event, Eq. (1) can be used in the basin, and in the channel

5. Discussion The Zymoetz River rock avalanche is a large landslide that resulted in significant direct and indirect damages. The direct damages associated with this

Table 1 Run-up and superelevation data used in the velocity calculations at different locations in the channel Equation


h (m)

g (m/s)

rc (m)

h (deg)


Superelevation Run up Superelevation Superelevation Superelevation Superelevation Superelevation Superelevation Run up Superelevation Run up Superelevation

Basin 1 Basin 1 C-S 2 C-S 3 C-S 7 C-S 8 C-S 9 Corner 1 Corner 1 Corner 2 Corner 2 Corner 3

60 60 25 30 46 38 44 22 22 36 36 –

9.8 9.8 9.8 9.8 9.8 9.8 9.8 9.8 9.8 9.8 9.8 9.8

204 – 156 155 150 125 130 156 – 44 – 52

19 – 8 10 18 12 18 10 – 45 – 33.7

0.34 – 0.14 0.18 0.32 0.21 0.32 0.18 – 1 – 0.67

a (deg)


V (m/s)


0.99 – 0.99 0.99 0.96 0.96 0.98 0.98 – 0.99 – 0.99

26.11 34.29 14.63 16.29 21.48 15.86 20.12 16.24 20.77 20.64 26.56 18.32

– 5 8 15 15 12 12 – 9 – 9


N. Boultbee et al. / Engineering Geology 83 (2006) 76–93

Table 2 Mobility comparison between the ZRRA and other similar events Event


Estimated volume

H (m)

L (m)


Little Tahoma Peak Huascaran

1963 1970

11 (106 m3) 50–100 (106 m3)

1890 4100



5–6 (106 m3)


Zymoetz River


1 (106 m3)


6900 16 000 76 8600 30 4500

0.27 0.26 (mean) 0.23 (mean) 0.27

Velocity (m/s)


29–42 278

(peak) (peak)

Fahnestock (1978) Plafker and Erickson (1978)



Evans et al. (1989)



This study

Modified after Evans et al. (1989).

event include the severed PNG pipeline, the Forest Service road, and the lost timber, with an estimated dollar amount of $5.9 million. The indirect costs greatly increase to an estimated $27.5 million, when the lost time and revenue from the logging road and pipeline are taken into account (Schwab et al., 2003). A number of large landslides occurred in British Columbia between 1999 and 2003, with 5 of the 6 occurring in the spring and summer of 2003. The idea has been proposed by several authors that this may be associated with retreat of glaciers in alpine areas, leading to de-buttressing of slopes, and eventually instability (Evans et al., 1989; Schwab et al., 2003). As evident there is the potential for remote alpine slope failures to develop into the more hazardous long-runout events. A review of selected large rock avalanches in the literature demonstrates two general types, [1] those that travel long distances in confined mountain valleys,

exhibiting high fluidity, and [2] those that runout into relatively open valleys, are not channelized, and travel shorter distances. The Zymoetz River event is of type [1], along with Pandemonium Creek (Evans et al., 1989), Huascaran (Plafker and Ericksen, 1978), and Little Tahoma Peak (Fahnestock, 1978), among others. These type [1] events are comparable to the A type situation discussed by Nicoletti and Sorisso-Valvo (1991), involving a channelized mass. Type [2] includes events like the Hope (Mathews and McTaggart, 1978) and Frank slides (Cruden and Hungr, 1986), and is equivalent to the types B (unobstructed spreading, or moderate energy dissipative) and C (rightangle impact against an opposite slope, or high-energydissipative) defined by Nicoletti and Sorisso-Valvo (1991). The estimated volume, velocity and fahrboschung ( F) angle of the ZRRA are compared with other morphologically similar British Columbian rock avalanche

Fig. 16. Graph showing the relationship between Fahrboschung and landslide volume. Modified after Evans et al. (1989).

N. Boultbee et al. / Engineering Geology 83 (2006) 76–93

events (Table 2). The fahrboschung angle is the angle between the headscarp and the extreme limit of the deposit of a landslide, and has been used as a measure of mobility by numerous authors, including Evans et al., 1989 (Fig. 16). In order to allow a direct comparison, preliminary estimates of the fahrboschung angle were measured in the same manner as in Evans et al. (1989); from the highest point of the detachment, to the lowest point of the deposit, for the vertical elevation difference, and along the path length, for the horizontal difference measurement. From this comparison, it can be seen that the ZRRA exhibits high mobility similar to the Pandemonium Creek rock avalanche (PCRA), and these other BC events. Another event with some similarities to the PCRA and the ZRRA is the Nomash River landslide on Vancouver Island. The Nomash River event occurred on April 26, 1999, near the west coast of northern Vancouver Island (Guthrie, 1999). The landslide ran out 2.6 km, and involved a failure volume of 300,000 m3, and travelled at speeds of up to 22 m/s. While the Nomash event involved a smaller failure volume, and shorter runout, it is interesting to compare its shape to that of PCRA and ZRRA (Fig. 17). These three BC landslides have very similar shapes, all showing failure onto open slopes, impact against the opposite valley wall, with subsequent right-angle turns and confinement into creek channels. While the PCRA and ZRRA ran out to main river channels (for PCRA the South Atnarko River, and for the ZRRA the Zymoetz River), the Nomash event does not travel

Fig. 17. Comparison of the similarities in channel shape between (A) Pandemonium Creek rock avalanche (Evans et al., 1989), (B) Zymoetz River Rock avalanche, and (C) Nomash River landslide (Favero, 2000).


as far as the junction with the Zeballos River, it remains confined to the Nomash Valley. The shape of these channels seems to be a combination of Nicoletti and Sorisso-Valvo (1991) types A (channelling of the debris mass) and C (right-angle impact against opposite slope). They found that the type A shape seemed to have the longest runout, while type C had the shortest. It is recommended that further work be undertaken to investigate the frequency of occurrence of this particular channel shape, and evaluate the runout distances relative to the other channel shapes presented in the literature. 6. Conclusions The ZRRA was a natural landslide that initiated as a result of progressive degradation of an altered rock mass, on the glacially over steepened walls of a cirque basin. As the rockslide impacted the cirque basin, it was accelerated down the valley, aided by reduced basal friction on the snow at the south end of the basin, and potential entrainment of some snow and saturated debris towards the north end of the basin. The debris avalanche exhibited spectacular mobility as it flowed down the valley, evidenced by high superelevations, and mud splashes on trees 60 m above the valley floor. A large amount of debris remained in the basin, some was deposited in the valley, replacing the thin layer of entrained soil, while the majority of the debris travelled the entire 4.5 km to the Zymoetz River, and formed a large fan. The damaged PNG pipeline had to be rerouted to the top of the bedrock canyon, so as not to be impacted again by future events initiating in Glen Falls Creek. In British Columbia, the low population density rarely necessitates extensive monitoring of isolated high mountain slopes. As population, logging and recreation industries continue to spread into remote areas, there is a consequent incremental increase in risk posed by rockslides. While it is already common practice to investigate the potential for debris flows or other small mass wasting events, in populated or environmentally sensitive locations, it may become increasingly necessary in the future to investigate the potential for long runout large landslides when considering recreational developments in remote areas or areas proposed for timber harvesting. This threat was clearly evident after the Pandemonium Creek event that devastated a fan 9 km away from the source (Evans et al., 1989). The Zymoetz River rock avalanche, which caused extensive damage to both a gas pipeline and forest road access further emphasises the


N. Boultbee et al. / Engineering Geology 83 (2006) 76–93

need to consider the risks posed by such long runout events. Acknowledgements This paper forms part of MSc. Thesis research undertaken by the first author in the Department of Earth Sciences, Simon Fraser University. Funding was provided through Forest Renewal B.C. and NSERC. The authors would like to acknowledge the B.C. Ministry of Forestry and the Geological Survey of Canada for their support. Many thanks also to D. Kinakin, and M.-A. Brideau for numerous valuable discussions, and technical support. References British Columbia Geological Survey — Geology Map (online) 2000. Cavers, D.J., 2003. Zymoetz River and Limonite Creek rock avalanches. Proceedings of the 3rd Canadian Conference on Geotechnique and Natural. Hazards, Edmonton, Alberta, ISBN: 0-920505-23-6, p. 299. June 9 and 10. Clague, J.J., 1984. Quaternary geology and geomorphology, Smithers–Terrace–Prince Rupert area, British Columbia. Geological Survey of Canada, Memoir 413. Couture, R.E., Evans, S.G., Locat, J., Hadjigeorgiou, J., Antoine, P., 1999. A proposed methodology for rock avalanche analysis. In: Yagi, N. (Ed.), Slope Stability Engineering. Proceedings of the international Symposium on Slope Stability Engineering — Shikoku’99. Brookfield, Rotterdam, pp. 1369 – 1378. Crosta, G.B., Agliardi, F., 2003. Failure forecast for large rockslides by surface displacement measurements. Can. Geotech. J. 40, 176 – 191. Cruden, D.M., Hungr, O., 1986. The debris of the Frank slide and theories of rockslide–avalanche mobility. Can. J. Earth Sci. 23, 425 – 432. Cruden, D.M., Varnes, D.J. 1996. Landslide types and processes Special Report 247: Landslides: Investigation and Mitigation (A.K. Turner and R.L. Schuster eds. ) TRB, National Research Council, Washington, D.C., 36-71. Erisman, T.H., Abele, G., 2001. Dynamics of Rockslides and Rockfalls. Springer-Verlag, Heidelberg, ISBN: 3-540-67198-6. ESRI, 2002. ArcGIS Desktop 8.2. Environmental Systems Research Institute, USA. Website: Evans, S.G, Clague, J.J., Woodsworth, G.J, Hungr, O., 1989. The Pandemonium Creek rock avalanche, British Columbia. Can. Geotech. J. 26, 427 – 446. Evans, S.G., Couture, R., Fuller, T., Turner, K., 2003. The 2002 rock avalanche at McAuley Creek, near Vernon British Columbia; implications for regional landslide hazard assessment. Proceedings of the 3rd Canadian Conference on Geotechnique and Natural Hazards, Edmonton, Alberta. ISBN: 0-920505-23-6, p. 299. June 9 and 10. Fahnestock, R.K., 1978. Little Tahoma Peak rockfalls and avalanches, Mount Raiunier, Washington, USA. In: Voight, B.Rockslides and Avalanches, vol. 1. Elsevier, Amsterdam, The Netherlands, pp. 181 – 196.

Favero, P., 2000. Sedimentological and dynamic analysis of the Nomash River landslide, Vancouver Island, British Columbia. Unpublished BSc. Honours Thesis. Simon Fraser University. Geertsema, M., Clague, J.J., Schwab, J.W., Evans, S.G., 2006 (this issue-a). An overview of recent large landslides in northern British Columbia, Canada. Eng. Geol. 83, 120–143. doi:10.1016/j.enggeo.2005.06.028. Geertsema, M., Hungr, O., Evans, S.G., Schwab, J.W., 2006 (this issue-b). A large rock slide — debris avalanche in cohesive soil at Pink Mountain, Northeastern British Columbia, Canada. Eng. Geol. 83, 64–75. doi:10.1016/j.enggeo.2005.06.025. Geertsema, M., Schwab, J., 1996. A photographic overview and record of the Mink Creek earthflow, Terrace, BC. B.C. Ministry of Forests, Research Report, vol. 08. Geological Survey of Canada, 1992. In: Gabrielse, H., Yorath, C.J. (Eds.), Geology of the Cordilleran Orogen in Canada, vol. G-2. Guthrie, R. 1999. Report on the Nomash landslide. Ministry of Environment ands and Parks, Water Management. Nanaimo. File: 76855-06/falls. Hoek, E., Carranza-Torres, C., Corkum, B., 2002. Hoek–Brown failure criterion — 2002. 5th North American Rock Mechanics Symposium and 17th Tunneling Association of Canada Conference: NARMS–TAC, 2002, pp. 267 – 271. Hungr, O., Evans, S.G., 2004. Entrainment of debris in rock avalanches: an analysis of a long run-out mechanism. GSA Bull. 116 (9/10). Hungr, O., Morgenstern, G.C., Kellerhals, R., 1984. Quantitative analysis of debris torrent hazards for design of remedial measures. Can. Geotech. J. 21, 663 – 677. I.S.R.M., 1985. Suggested method for determining point load strength. Int. J. Rock Mech. Min. sci. geomech. abstr. 22 (2), 51 – 60. Jordan, R.P., 1994. Debris flows in the southern coast mountains British Columbia: dynamic behaviour and physical properties. PhD Thesis, UBC, Vancouver, 260 pp. Locat, P., Couture, R., Locat, J., Lereoueil, S., 2003. Assessment of the fragmentation energy in rock avalanches. 3rd Canadian Conf. on Geotechnique and Natural Hazards, Edmonton, pp. 261 – 268. Maerz, N.H., Zhou, W., 1998. Optical digital fragmentation measuring systems — inherent sources of error. FRAGBLAST, The Int. J. Blasting Fragm. 2 (4), 415 – 431. Mathews, W.H., McTaggart, K.C., 1978. Hope rockslides, British Columbia, Canada. In: Voight, B. (Ed.), Rockslides and Avalanches. Elsevier, Amsterdam, pp. 259 – 275. Marinos, P., Hoek, E., 2004. GSI — a geologically friendly tool for rock mass strength estimation. Proc. GeoEng2000 Conference, Melbourne. Meidinger, D.V., Pojar, J.J., 1991. Ecosystems of British Columbia. BC Ministry of Forests. Special Report Series 6. Mihalynuk, M.G. 1987. Metamorphic, structural and stratigraphic evolution of the Telkwa Formation, Zymoetz River Area (NTS 103 I/8 and 93 L/5), near Terrace, British Columbia; M.Sc. thesis, Dept. of Geology and Geophysis, University of Calgary, Alberta. Nicoletti, P.G., Sorisso-Valvo, M., 1991. Geomorphic controls of the shape and mobility of rock avalanches. Geol. Soc. Amer. Bull. 103, 1365 – 1373. Plafker, G., Ericksen, G.E., 1978. Nevados Huascaran avalanches, Peru. In: Voight, B. (Ed.), Rockslides and Avalanches: Vol. 1. Elsevier, Amsterdam, The Netherlands, pp. 277 – 314.

N. Boultbee et al. / Engineering Geology 83 (2006) 76–93 Schwab, J.W., 2002. Catastrophic rock avalanche: Howson Range, Telkwa Pass. Ministry of Forests. Forest Sciences Extension Note #46. Schwab, J.W., Geertsema, M., Evans, S.G., 2003. Catastrophic rock avalanches, West-Central British Columbia. Proceedings of the 3rd Canadian Conference on Geotechnique and Natural Haz-


ards, Edmonton, Alberta. ISBN: 0-920505-23-6pp. 299. June 9 and 10. WipWare, 2003. WipFrag for Windows. Version 2.2. Operating manual. WipWare Inc. website: for_windows.htm.