Pore pressure response during failure in soils
EDWIN L. HARP U.S. Geological Survey, 345 Middlefield Road, M.S. 998, Menlo Park, California 94025 WADE G. WELLS II U.S. Forest Service, Forest Fire Laboratory, 4955 Canyon Crest Drive, Riverside, California 92507 JOHN G. SARMIENTO Wahler Associates, P.O. Box 10023, Palo Alto, California 94303
111 45' ABSTRACT
Three experiments were performed on natural slopes to investi- gate variations of soil pore-water pressure during induced slope fail- ure. Two sites in the Wasatch Range, Utah, and one site in the San Dimas Experimental Forest of southern California were forced to fail by artificial subsurface irrigation. The sites were instrumented with electronic piezometers and displacement meters to record induced pore pressures and movements of the slopes during failure. Piezome- ter records show a consistent trend of increasing pressure during the early stages of infiltration and abrupt decreases in pressure from 5 to 50 minutes before failure. Displacement meters failed to register the amount of movement, due to location and ineffectual coupling of meter pins to soil. Observations during the experiments indicate that fractures and macropores controlled the flow of water through the slope and that both water-flow paths and permeability within the slopes were not constant in space or time but changed continually during the course of the experiments.
INTRODUCTION
The mechanism most generally ascribed (for example, Campbell, 1975, p. 18-20) to account for rainfall-induced failure in soils indicates that an increase in the pore-water pressure within the soil mantle results in a reduction of the normal effective stresses within the material. Failure Figure 1. Location of experiment sites 1 and 2, north of Salt Lake occurs when pore pressure increases to the point that the frictional resis- City, Utah.
Ricks Creek'
Figure 2. Oblique aerial photo- graph showing location of sites 1 and 2 and local features. View to east.
Geological Society of America Bulletin, v. 102, p. 428-438, 12 figs., 1 table, April 1990.
428
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021 PORE PRESSURE RESPONSE DURING FAILURE IN SOILS 429
tance to downslope movement drops below the downslope component of Accordingly, typical characteristics of pore-pressure behavior during fail- weight. ure could be delineated, and any premonitory responses could be discov- Iverson and Major (1986), however, favor the use of seepage forces ered and evaluated. The experiments were carried out at three sites, two in to describe slope stability. Their analysis concluded that minimum slope north-central Utah and one in southern California. Landslide nomencla- stability occurs when the angle between the seepage force vector and the ture used in this report is after Varnes (1978). normal to the slope, A, is 9QP-
"TTTT- Failure scarp, hachures on hanging wall
Figure 3. Location of in- struments with respect to soil failure at site 1 near Ricks Creek, Utah.
(68.6)
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021 430 HARP AND OTHERS
UTAH SITE2
o s 111 49'W
CENTER TRENCH — . _ OF • 40°53'N -
Pz4 (30.5) Pz6 (35.5) Pz8 (35.5) Pz3 (61.0) Pz5 (61.0) Pz7 (61.0) 2.5m
Pz9 (55.8) V 2.0m Figure 4. Location of in- struments with respect to soil SCARP failure at site 2 near Ricks Creek, Utah. Pz1 O (53.3) 1,5m
Pz11 (50.8)
1.0m
VERTICAL CUT
0.5m 0 1 meters
Om
EXPLANATION
Pz4 (3 0.5) Piezometer, location number to side, depth in cm in parentheses
0.5m Surface contours in m, elevations relative to arbitrary control points
DM Displacement meter line
Failure scarp, hachures on hanging wall, dashed where fracture only
SITE DESCRIPTIONS AND INSTRUMENTATION had been active during the period of intense debris-flow formation in May-June, 1983. The head of the slide complex contained several incip- Ricks Creek, Wasatch Range, Utah ient soil slides and slumps that had begun to undergo failure in 1983. We decided to attempt to initiate further sliding of this material and also of Two sites were chosen near Bountiful, Utah, in the Wasatch Range adjacent unfailed material near the head of the slide complex (sites 1 and about 20 km north of Salt Lake City (Fig. 1). The sites are located in the 2; Fig. 2). headwaters of Ricks Creek at about 2,500 m elevation. They are located at The headwall scarp of the slump/debris-flow complex is located just the head of a slump/debris-flow complex developed in residual soils that downslope from a nearly horizontal topographic bench within a cirque-
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021 PORE PRESSURE RESPONSE DURING FAILURE IN SOILS 431
11 8° 00' of the slope and to observe the effects of the seepage of water from the upper trench through the slope. At site 2, a single trench 2 m long and about 0.6 m deep (Fig. 4) was dug along the slope strike at the head of a 43° planar slope. The flow of water from the trenches into the slope was intended to simulate the flow of water through the soil that occurs when the water table (perched or annual) is extremely high and near the surface. We had previously at- tempted experiments to trigger slope failures by the use of sprinkler systems. Even at sprinkling rates many times higher than normal rainfall and duration rates, all attempts were unsuccessful. Despite the fact that water infiltration from a trench may not exactly reproduce natural water- flow conditions in soils, it does introduce a high water table, elevate pore-water pressures, and initiate failures. The objective of the experiments was to study the pore-water pressure variation and slope deformation during failure rather than to exactly reproduce natural ground-water con- ditions. During the course of the experiments, the trenches were kept as full as possible to saturate the slopes. Data from the piezometers and displacement meters were recorded digitally and stored on cassette tapes.
Monroe Canyon Site, California: San Dimas Experimental Forest
A site on the west side of Monroe Canyon near the summit of a Dimas Experimental Forest, southern California. prominent peak in the San Dimas Experimental Forest near Glendora, California (Fig. 5), was chosen to initiate a soil failure because of the numerous debris flows that had occurred there in previous years. The slopes at this site are grass covered; they had been converted from chapar- shaped valley that contains several abandoned beaver ponds. The beaver ponds are fed by springs that produce water at a combined rate of -175 1/min, which is enough to infiltrate the slopes at the experiment sites and cause small soil failures. The ponds are about 100 m from the experiment sites. Water was pumped from the ponds into a 220-1 holding bag just SAN DIMAS SITE above the sites, to be dispensed into trenches dug within the slopes to be failed. The locations of the sites and ponds with respect to each other are 3340 shown in Figure 2. Bedrock at the sites consists of Archean and early Proterozoic gneiss and schist of the Farmington Canyon Complex. The soils vary considera- bly but are of three basic types. Soil at site 1 at the head of the slide complex is a gravelly, silty sand. Site 2 (Fig. 2) has a surficial soil consist- ing of permeable silty sand about 0.6 m thick underlain by a relatively impermeable bluish-gray silty sand (silt fraction 22%-25%). Site 1 was located at the head of the slide complex on a 6-m2 soil slide that had become detached from soil above the main slide scarp but had not mobilized into a debris flow (Fig. 2). Nine electronic piezometers were installed within the incipient slide mass at depths varying from 30.5 to 172.7 cm (Fig. 3). Each piezometer is constructed of a differential pressure transducer housed in a cylindrical nylon chamber having a coni- cal tip (see Harp and others, 1984). The pressure resolution of the pie- zometers is ~0.6 mm head, and the accuracy is ±1.3 mm head. Two displacement meters were also installed on the slide; these consisted of light-weight braided wire cable attached to 10-turn potentiometers. The EXPLANATION failure surface of the soil slide was —70 cm deep. Site 2 was located ~ 15 m • 6(68.6) Piezometer, location number to 'A north of the first experiment site. Here an unfailed, essentially uniform side, depth in cm in parentheses slope of 43° was instrumented with nine electronic piezometers and one 333S Surface contours in m
displacement meter (Fig. 4). | pm-i > Displacement meter line In each of the experiments, water was introduced into the subsurface —TTT-t- Failure scarp, hachures on hanging through trenches. At site 1, a trench was dug at the head of a small soil wall mass that had undergone previous sliding but was still intact. The trench was dug along the strike of the slope face, 1.6 m long and 0.6 m deep. A Figure 6. Location of instruments with respect to soil failure at similar trench was dug 1.3 m downslope in the midsection of the soil mass Monroe Canyon experiment site, San Dimas Experimental Forest, (Fig. 3). This was done to provide greater access of water to the lower part southern California.
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021 432 HARP AND OTHERS
DESCRIPTION OF EXPERIMENTS, DATA, AND OBSERVATIONS
Ricks Creek Site, Utah
The experiment at Ricks Creek was begun in July 1984, while the last remnants of the preceding winter snowpack were still in evidence. At depths above 0.6 m throughout most of the area, the soil was moist but not saturated. Water levels in the trenches at both sites were kept relatively constant for an average of about 8 hr per day. At 5:38 p.m., July 23, on the fourth day of irrigating the slope, a soil slide occurred after introducing 4,8351 of water into the two trenches at site 1. The slide (0.5 m3) occurred immediately downslope from piezometer 3 (Fig. 3), and it slid into the lower trench (Fig. 7). For about 30 min prior to the failure, water was seeping rapidly from the upslope face of the lower trench; the greatest volume appeared to flow from small fractures in the soil. Small pieces of soil began to fall into the lower trench every few minutes. About 15 sec before failure, water flow from cracks in the lower trench face increased. At 1-2 sec before failure, cracks appeared to widen and link with each other to form a basal failure surface. At this time, water poured from this surface, and the mass of soil above these fractures slid into the lower trench. The slide movement took about 1.5 sec. The slide that occurred was ~35 cm in thickness and represented only a small portion of the slope affected by infiltration from the trenches; its shape and position relative to the instruments and the trenches are shown in Figure 3. Neither of the displacement-meter anchors was posi- tioned on the area of the slope that failed, and so no displacement data were gathered during this failure.
TABLE 1 TIME SEQUENCE OP PORE-PRESSURE BEHAVIOR AND OBSERVATIONS BEFORE AND DURING FAILURE, MONROE CANYON SITE
Elapsed time Observations and pore pressure (minutes) Figure 7. Soil slide at experiment site 1 near Ricks Creek, Utah. 0 Beginning of slope irrigation on final day of Soil has slid into lower trench. Note scarp of landslide complex (pale experiment.
slope) across top of photograph. 20 Pore-pressure decrease of 0.8-cm head at piezometer 2.
24 Piezometer 2 exposed by small slump in lower trench wall.
28 Piezometer 2 buried by other small failures in lower trench wall; continued recording of ral about 25 years ago. An extensive network of debris-flow scars extends pore pressures. 33 Water begins seeping from vertical cut at below the experiment site, and the depths of failures range between 0.15 m bottom of experiment slope.
and 1.8 m. The headward parts of many of the scars contain soil slides that 39 Many seeps appear in vertical cut. Several had not yet mobilized into debris flows. miniature debris flows (2-4 cm in width) form from these seeps. The soil mantle at the site varies from 0.15 m to about 1.0 m in depth 41 0.5-m^ upslope section of lower trench wall and was developed on an ultramafic dike within surrounding granitic slumps into trench. Piezometers 4,7,9, 10, 11, and 12 show moderate rises before and gneiss. Because the bedrock is highly weathered, it is not clear to exactly during slumping. Piezometers 2 and 7 show 5- what depth the term "soil" should be applied. The surface rock is ex- and 6-cm head increases; increase of piezometer 2 probably due to burial by small failure tremely soft due to its weathering profile, and it is like soil to depths of masses. more than 2.0 m in places. The soil varies from coarse- to medium-grained 44 Piezometers 10 and 12 (those nearest the vertical cut) show pressure decrease of 6- and sand and contains as much as 20% clay. 12-cm head, respectively.
Twelve electronic piezometers and two displacement meters were 46 Piezometer 2 shows decrease of 1 -cm head. Piezometers 4 and 5 show 0.3-cm head decreases installed within a 4.8-m by 4.8-m plot (Fig. 6). The depths of the piezome- followed by 0.8- and 1.5-cm head increases, ters ranged from 59.7 cm to 177.8 cm. As at the Utah experiment sites, respectively, in the following 5 minutes. water was introduced into the slope from trenches. 48 First of many small stumps occur from vertical cut. In following 7 min, 6 to 8 small slumps A 3.24-m-long, 1.0-m-deep trench was dug along the strike of the and 2 larger slumps (each about 2 m^) occur slope near the crest of Monroe peak (informal name). A 2.0-m-long trench from vertical cut. 50 Abrupt pore pressure drops shown by piezometers (0.8 m deep), was also dug 1.2 m downslope from the upper trench. Water 2, 4, 5, 7, 9, 10,11, and 12 (2- to 37-cm had to be transported to the site by truck and stored in a collapsable head). holding tank having a 6,593-1 capacity. Water was supplied to the trenches 55 Translational soil failure of 14-m^-volume slides onto bench below vertical cut. by pumping at a rate of ~ 100 l/min.
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021 PORE PRESSURE RESPONSE DURING FAILURE IN SOILS 433
1 4 1 1 T" Pz 3 —(91.0 cm) 1 2 Pz 4—(71.1 cm) 10 Pz 5 —(55.8 cm)
8
6 4 J 2
O
-2
-4
1 4 1 1 1 1 1 1 1 1 ! 1 1 1 T> Pz 6 —(66.0 cm) •••"• 0 1 2 Pz 7 — (1 72.7 cm) / „
2
3 m 0 m CD -2 -4 1 1 1 1 1 1 1 1 1 1 1 1 Q_ 14 0) 1 1 1 1 1 [ 1 1 1 1 1 1 12 Pz 9 -(137.2 cm) 0 Pz 1 O —(30.5 cm) Q_ 10 Pz 11 -(68.6 cm)
8 .."" 'r-. 1 • 1 6 v ' v 4 : ,-r -K. I ' " '• —^———\ * ~~ 2 - ...... erf 0 ! -N ' 11 \ 1 -2 1 , , - 1 1 \ 1 -4 1 1 J \ 1 p.m . -6 p.m . 'I > 1 0 1 4 : -8 1 5:3 8
-10 _ 1 1 ! 1 200 400 600 800 1000 1 200
Time (minutes)
All of the piezometer records show a gradual rise of 1.0- to 5.0-cm 6:00 p.m. on July 31,1984. A total of 5,2641 of water was used during the head from 60 to 420 min and an abrupt rise of 1.0- to 4.0-cm head at experiment on the last day. Except for a rise in head applied at the trench, about 420 min (-11:00 a.m.; Fig. 8). Records of piezometers 4, 7, and 9 which is discussed below, water level was kept approximately constant show abrupt rises between 600 and 660 min, whereas the rest show throughout the experiment. The flow rate to accomplish this was in the sporadic declines between 550 and 700 min. Two piezometer records, 7 range of 1.0-5.01/min. and 9, show rapid changes shortly before or at the time of failure. They Measurements of slope displacement during this experiment, as at site both show a rapid decrease in pressure about 20 min prior to failure. None 1, were also unsuccessful because the displacement meter anchors were of the piezometers at this site was located within the failure mass, and all outside the boundary of the failure mass. but piezometer 10 were deeper than the failure surface. The abrupt post- For this slope in permeable, sandy soil, the piezometers show rapid failure rise in pressure of piezometer 3 is due to instrument disturbance. responses to fluctuations in water levels in the trench. As the trench was At site 2, piezometers were installed at shallower depths (30.5-61.0 filled at 10:40 a.m., pore pressures indicated by piezometers 3,5, and 6 at cm) to insure their positions above a potential slide surface. Here, the the top of the slope rose immediately 6.0-15.0 cm (Fig. 9). Pore pressures experiment was conducted in a manner similar to that at site 1 except that at piezometers 9,10, and 11, farther down the slope, began to rise rapidly only one trench was used, and the slope itself had not undergone recent from 20 to 40 min after the trench-filling began. About 30 min after failure. After 2 days of infiltration for about 6 hr per day, the experiment infiltration began, a wet spot appeared above piezometer 10; subsequently, on the final day was conducted over a period of 8 hr, from 10:00 a.m. until a miniature debris flow developed, about 30 cm in length. This wet spot
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021 45 Pz 3 -(61.0 cm) 40 Pz 4 -(30.5 cm) \ ' Pz 5 -(61.0 cm) I 35 I \ _ 30 M A / 25 ^"VJ / y - y 20 / - 15
10
5
0 Figure 9. Plots of pore pressure versus time -5 during soil failure at site 2 near Ricks Creek, Utah. 40 Vertical lines denote time Pz 6 -(35.5 cm) of specified events. Pz 7 -(61.0 cm) 30 Pz 8 -(35.5 cm)
\ *>/ 20
\t
10 I- A s I y Wj 0 / \
-10
40
Pz 9 -(55.8 cm) Pz 10 —(53.3 cm) 30 Pz 11 -(50.8 cm) - - _ ^ v
20 .. v 4- ^ Hi s / / ; • 10 '/: —^ " ii .. / : rv- 0 » JC JIf \» • j y V -10 0 200 400
Time (minutes)
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021 Figure 11. Plots of pore pressure versus time during soil failure at Mon- roe Canyon site, southern California. Vertical lines across records denote times of specified events.
20 40 60 80 100 Time (minutes)
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021 PORE PRESSURE RESPONSE DURING FAILURE IN SOILS 435
Figure 10. Arcuate scarp above soil slide at experiment site 2 near Ricks Creek, Utah.
continued to enlarge and seep water for about an hour; then the seepage piezometers and their depths in the slope are shown in Figure 6. A hori- stopped, and the wet spot ceased to enlarge. At this time, another wet spot zontal bench with a 1.7-m-high vertical cut was constructed at the bottom formed and produced seepage to the right and downslope, above piezome- of the monitored part of the slope to simulate the natural scarps present on ter 11. the hillside and to increase the chances of producing a slope failure by At 12:48 p.m., after 1,8111 of water had been added to the slope that decreasing the stability of the slope. It was reasoned that dynamic hydro- day, deformation was well underway, as indicated by prominent cracks logic processes that would cause failure of a slope above a vertical cut developing above piezometer 11. These cracks continued to widen until a would be similar to processes causing retrogressive failure above natural prominent asymmetric scarp (Figs. 4 and 10) extended from a position scarps on nearby slopes. Two trenches were constructed, but water was below and to the right of piezometer 11 to a position to the right and introduced into the slope from the upper trench only. above piezometer 10 (Fig. 4). The right side of the scarp opened up to 40 Because the permeability of the slope at the Monroe Canyon site was cm, but the left side developed only hairline fractures. Piezometer 11, extremely high, the trench could not be kept full at a stable water level within the failure mass, recorded a consistent decrease in pore pressure, with an inflow rate of less than 102 1/min. The soil profile here, as starting about 60 min before the first cracks of this scarp were noticed. revealed by the vertical cut, was 0.4 to 0.6 m of sandy topsoil containing as Other piezometers showed no significant decrease in pore pressure until much as 20% clay and underlain by highly weathered and fractured dia- much later in the experiment. base. After water began issuing from the vertical cut during the experi- At 2:10 p.m., we began to build up the soil embankment around the ment, it became obvious that most of the water was flowing from trench to raise the water level and increase the hydrostatic head to the macropores and fractures rather than from the soil matrix. slope. At 2:50 p.m., we completed this effort and succeeded in raising the After 4 days of irrigation during the daytime of about 7 hr/day and water level in the trench by almost 10 cm. We found this to be the drainage with no irrigation at night, failure occurred 55 min after trench maximum level that we could sustain without causing overflow and sur- irrigation began on the fifth day of the experiment, at which time a total of face erosion. Records of the piezometers closest to the trench (numbers 3, -48,000 1 of water had been applied. The rate of irrigation was constant 5, and 6) show the effect of water-level rise the most. Piezometer 4 as well during the periods of infiltration. It had been 19 hr since the previous day's as piezometer 9, farther from the trench, also recorded increases in head. experiment, during which steady irrigation was maintained for 7 hr, and At 4:16 p.m., because further movement in the lower-right side of the seepage from the cut-slope face and high-pressure heads were observed. failed slope had apparently ceased, a 1-m-deep vertical cut was made at On the fifth day, piezometers 9 and 11 were the first to show pore- the bottom of the slope to further destabilize the slope and stimulate the pressure rises (Fig. 11), about 12 min after irrigation began. All piezome- failure process. At 6:00 p.m., the right half of the slope collapsed and ters except 1, 3, 6, and 8 had shown significant pore-pressure rises within breached the trench. 37 min. Piezometers 1, 3, and 6 (Fig. 6) were rather shallow and near the Approximately 30 min before the collapse of the slope and breach of top of the slope, probably above the perched ground-water surface for the trench, all of the piezometers except number 4 exhibited a sharp drop most of the experiment. Piezometer 8 was a control instrument placed in pore pressure. This decrease ranged from 15 cm (piezometer 11) to 6 outside of the experiment area to measure any possible temperature effects cm (piezometer 8). on the piezometers. From the record of piezometer 8, it appears that there were no significant temperature effects, as it shows an essentially flat trace Monroe Canyon Site, California: San Dimas Experimental Forest throughout the experiment. The time sequence of pore-pressure behavior and detailed observations throughout the experiment on the final day are The experiment on a slope of Monroe Canyon was conducted in a tabulated and presented in chronological order in Table 1. similar fashion to that of the experiments in Utah. The distribution of the At 50 min (5 min before the final failure of most of the experiment
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021 10 HARP AND OTHERS
fluctuation of piezometer 3 after failure is an artifact of instrument distur- movement before and during slope failure. Pore-pressure records from the bance. Although the pressure fluctuations of the piezometers at site 1 may experiment slopes show a consistent trend of pore-pressure increases dur- be related to the slope failure, the patterns are not easy to interpret and are ing the early stages of infiltration and of abrupt decreases in pore pressure not uniquely correlated to the failure. The fact that the piezometer records from 5 to 50 min before failure and large-scale movement. The precursory at site 1 do not show well-correlated abrupt prefailure pressure decreases decrease in pore pressure may be useful in forecasting the time of failure of similar to those of the other experiments suggests that the piezometers at slopes prone to soil slides and debris flows. Electronic piezometer arrays site 1 were probably buried too deeply to accurately reflect the pressure may be useful as elements in warning systems if future research shows that changes taking place within the soil mass that failed. the pattern of pore-pressure decreases before slide movement seen in these Observations of large spatial and temporal fluctuations of water flow experiments is repeatable and consistent. within the slopes at site 2 in Utah and at the Monroe Canyon site support Observations of water-flow rates from cut-slope faces in the slopes the assertion that flow paths within the slopes were continually changing. during the experiments indicate that fracture and macropore permeability Silt- and clay-size fractions of the soil may have been deposited, eroded, was predominant and that water-flow paths and permeability were con- and redeposited within the flow network in the slope, altering the permea- tinually changing. Accordingly, there may be serious problems with hydro- bility of portions of the slope. These observed piping phenomena should be logic models of slopes that assume isotropic, homogeneous, or constant considered in the use of ground-water flow models that assume isotropy temporal permeability. The experiment results also suggest that the dra- and constant permeability. It is possible that an assumption of constant matic drops in pore-water pressure prior to failure, as well as dilatation permeability within many slopes may be valid; however, our experience in and failure itself, may be due to the piping of fine-grained soil particles. observing flow and measuring pore-water pressure within the three slopes discussed suggests that permeability within some shallow soils is not iso- ACKNOWLEDGMENTS tropic or homogeneous and that the flow of water within a slope can redistribute and/or remove fine particles so as to change permeability with We acknowledge the help and advice of Charles Colver, Manager, time. San Dimas Experimental Forest, U.S. Forest Service (retired) and his The redistribution and removal of fine particles qualitatively ob- permission to conduct experiments and use the facilities of the experimen- served in our experiments are consistent with results from experiments tal forest. We also thank the Angeles National Forest for the loan of conducted in the laboratory by Wright and Foss (1968) that showed that water-storage equipment. We thank Richard Kline, District Ranger, Salt even coarse silt (0.02 to 0.05 mm) was readily moved through columns of Lake District of the Wasatch-Cache National Forest and Arthur N. Car- medium sand (0.25 to 0.5 mm) 33 cm in length so that 93% to 95% of the roll, Forest Supervisor of the Wasatch-Cache National Forest for permis- silt was translocated, and 50% was completely removed from the columns sion to conduct experiments in the headwaters of Ricks Creek. We also (Wright and Foss, 1968, p. 447). acknowledge the helpful suggestions of Bruce Vandre, geologist, Inter- Piping within a slope may ultimately result in the loss of enough mountain Region of the Wasatch-Cache National Forest, regarding the material along fractures to reduce pore pressure in the slope, and the artificial saturation of slopes and the skilled assistance of U.S. Forest destruction of enough contact points along a potential slide surface, so that Service pilot "Monty" Montgomery and helitach crew for airlifting our failure occurs due to the loss of shear strength. The formula for the expres- equipment in and out of the experiment site at Ricks Creek. We thank sion for shear strength is Alan Campbell and Peter Wohlgemuth of the Riverside Fire Laboratory- U.S. Forest Service and Randall Jibson of the U.S. Geological Survey for assistance with site preparation, instrument installation, and experiment r = c + (CT - ywh) tan
pore-water pressure, ywh. As grain contact points are removed across a REFERENCES CITED potential failure surface, c is reduced, and the shear strength can decrease Campbell, R. H., 1975, Soil slips, debris flows, and rainstorms in the Santa Monica Mountains and vicinity, southern even though pore-water pressures are dropping prior to failure and tending California: U.S. Geological Survey Professional Paper 851,51 p. Harp, E. L., Sarmiento, John, and Cranswick, Edward, 1984, Seismic-induced pore-water pressure records from the to increase the effective normal stress, a-ywh, or a'. Thus, the common Mammoth Lakes, California, earthquake sequence of 25 to 27 May 1980: Seismological Society of America model of rising pore pressures causing failure by decreasing the effective Bulletin, v. 74, no. 4, p. 1381-1393. Iverson, R. M., and Major. J. J., 1986, Groundwater seepage vectors and the potential for hillslope failure and debris flow normal stress, a', does not fully explain the failure of the slopes in our mobilization: American Geophysical Union, Water Resources Research, v. 22, no. 11, p. 1543-1548. Keefer, D. K., and Johnson, A. M., 1983, Earthflows: Morphology, mobilization, and movement: U.S. Geological Survey experiments. Professional Paper 1264,56 p. Pilgrim, D. H., and Huff, D. D., 1983, Suspended sediment in rapid subsurface stormflow on a large field plot: Earth Surfaces and Landforms, v. 8, p. 451-463. SUMMARY Sidle, R. C., and Swanston, D. N., 1982, Analysis of a small debris slide in coastal Alaska: Canadian Geotechnical Journal, v. 19, p. 167-174. Varnes, D. J., 1978, Slope movement types and processes, in Schuster, R. L., and Krizek, R. J., eds., Landslides analysis and control: Washington, D.C., National Research Council, Transportation Research Board, Special Report 176, Three experiments were conducted to initiate soil failures in slopes p. 11-33. prone to debris flows. Two failures were induced at a site in the headwa- Wieczorek, G. F., 1987, EfTect of rainfall intensity and duration on debris flows in central Santa Cruz Mountains, California: Geological Society of America Reviews in Engineering Geology, v. 7, p. 93-104. ters of Ricks Creek, Utah, and one at a site in Monroe Canyon in the San Wright, W. R., and Foss, J. E., 1968, Movement of silt-sized particles in sand columns: Soil Science Society of America Dimas Experimental Forest in southern California. The slopes in which Proceedings, v. 32, p. 446-448. these failures were induced were instrumented with electronic piezometers MANUSCRIPT RECEIVED BY THE SOCIETY MARCH 27,1989 REVISED MANUSCRIPT RECEIVED JULY 19,1989 and displacement meters to record pore-water pressures and landslide MANUSCRIPT ACCEPTED AUGUST 3, 1989
Printed in U.S.A.
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021 PORE PRESSURE RESPONSE DURING FAILURE IN SOILS 437
Figure 12. Soil slide and scarp at Monroe Canyon experiment site, southern California. Upper trench is at left margin of photograph. Debris flow at right margin started about S min after slope failure.
slope), piezometers 2,4,5,7,9,10,11, and 12 all showed abrupt drops in both exhibited smaller failures prior to the main slope failure and showed a pore pressure, ranging from 2 cm to 37 cm. At 55 min, a purely transla- widening of fractures 1-2 sec before failure. At the same time, the flow of tional failure of 1.3-m depth and 14-m3 volume slid onto the bench in 1-2 muddy water from fractures, especially those along the failure surface, sec (Fig. 12). Water poured rapidly from the failure scarp and ponded greatly increased. As the slopes failed, water poured out of the slope along behind the slide debris on the bench. In about 5 min, a small debris flow the failure surface. mobilized from the mixing of the ponded water and part of the slide debris. The combined results of the experiments in Utah and southern Cali- Attempts to measure displacements were largely unsuccessful. Both fornia suggest that elevated pore-water pressures and increased water flow displacement-meter anchors were located within the mass that failed (Fig. through slopes may lead to piping and may eventually induce dilatation 6); however, the stakes used as anchors simply pulled out of the soil as the prior to failure. Noticeable drops in pore-water pressure occur well before mass began to move. Thus, the displacement records indicate the time of visible cracks are produced in a slope surface. This phenomenon suggests failure but record none of the movement. that electronic piezometer arrays may be useful to predict landslide failure During observation of the vertical cut face, we noted that the greatest and to provide warning systems in certain areas of known unstable slopes. flow of water was from fractures and large openings (macropores: animal At site 2 in Utah, the first significant drops in pore pressure of burrows, root casts, and so on). Similar observations were made by Pil- piezometer 11 (Fig. 9) occurred more than 50 min before the first notice- grim and Huff (1983) from studies of suspended sediment in subsurface able cracks developed. After a vertical cut was made in the slope at this stormflow. They observed that most flow took place through macropores site, sharp drops in pore pressure were observed in 8 piezometers about 30 within the soil. min before the slope failed. At the California site, abrupt drops in pore We also noted that the rate of water flow from different parts of the pressure occurred 5 min before failure of the slope, and 2 piezometers cut slope varied with time. Water would flow rapidly from certain areas of showed sharp pore-pressure decreases 11 min prior to the failure. the face and then subside or stop altogether. Then other areas would have Several examples of pore-pressure rises were also recorded prior to increased flow. It appeared that the direction and rate of flow of water was and during failures. Pressure rises during failures, however, were not nearly constantly changing within portions of the slope and was not constant in so consistent and abrupt as were decreases. Generally pressure rises were space or time. recorded by only a few piezometers during failures and cannot be related with certainty to the failures. For instance, at the Monroe Canyon site DISCUSSION AND INTERPRETATION during slumping into the lower trench (41 min), piezometers 2,4,7,9,10, 11, and 12 showed moderate pressure increases; however, the pressures in The pattern of pore-pressure data, when compared to natural land- most of these piezometers had been rising for more than 15 min prior to slide movements, establishes a reasonably consistent trend of behavior the failure and continued to rise afterward. Piezometers 1 and 3 were precursory to slope failure and rapid movement. The pattern of pore- closer to the slump but showed little effect. pressure decrease prior to landslide movement, especially rapid movement, At site 1 in Utah, piezometers 7 and 9 showed large pressure rises 120 is interpreted to be due to the effects of subsurface removal of fine-grained to 180 min prior to failure. Piezometer 9 pressure then began to drop material (piping) and dilatation of the landslide mass in its preliminary sporadically until 20 min before failure, when it dropped abruptly. Pie- stages of failure. This interpretation is supported by observations of our cut zometer 7 pressure began dropping about 20 min prior to failure but more slopes just before and during failure. The lower trench face at site 1 in Utah gradually. Other piezometers showed only gradual decreases in pressure and the cut-slope face at the Monroe Canyon site in southern California beginning 120 to 150 min before failure. The abrupt pressure rise and
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021 ABCDE FGHI JKL
Figure 1. Landsat Thematic Mapper (TM) mosaic of the study dark blue) is the rift floor, Quaternary peralkaline silicic centers along area. The image is a mosaic of scene 51027-06594 (path 158/row 55, the active Wonji fault belt rift axis are Gadamsa (F-3), Bora (F-4), acquired December 23, 1986) and the southern half of scene 50675- Aluto (D-8) Shala caldera (C-10 to C-ll), nested Awasa-Corbetti 07075 (path 158/row 54, acquired January 5, 1986). The scenes de- calderas (B-13 to B-14), and Duguna (A-15). The eastern rift escarp- pict areas 185 km in width and more than 200 km long. The mosaic is ment is indicated by dark linear features (shadow) east of the lakes. a linearly scaled three-band composite, with TM Band 7 (short- The eastern rift flank shield volcanoes (3,700-4,200 m elevation) are wavelength infrared radiance, 2.08-2.35 jum) displayed in red, Band 4 Chilalo (H-6), Galama volcanic range (1-5 to 1-9), Hunkuolo (1-10), (near infrared radiance, 0.76-0.90 /xm) in green, and Band 3 (red Kaka (H-10 to H-U), and Chike (G-10). The Wabi Shebele River radiance, 0.63-0.69 ¿um) in blue. Edges have been enhanced by con- section (J-10 to J-l 1) runs parallel to the rift and to the volcanic range. volving each band with a 3x3 filter, using 16 as the center matrix value Vegetation appears green, open grasslands and farmlands are pink and -1 in the other matrix positions. In the north-south-oriented and light blue, volcanic rocks are brown, and clouds are white. The image, the prominent geologic landmarks are Guraghe western rift image was prepared by David Harding as a part of the East African margin (A-3 to A-7), block of pre-Tertiary rocks (0.5 cm north of B-4 Rift project at the Laboratory for Terrestrial Physics, NASA Goddard on the image), northeast-trending pinkish area with lakes (black to Space Flight Center.
Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/102/4/428/3380785/i0016-7606-102-4-428.pdf by guest on 02 October 2021