Radar Structure of Earthquake-Induced, Coastal Landslides in Anchorage, Alaska
WALTER A. BARNHARDT and ROBERT E. KAYEN
U.S. Geological Survey
© 2000, AAPG/DEG
Ground-penetrating radar (GPR) was used to investigate the internal structure of two large landslides in Anchorage, Alaska that resulted from the great 1964 earthquake. The Government Hill and Turnagain Heights landslides occurred in similar stratigraphic and geographic settings, yet the style of ground deformation is different at each site. GPR data are compared with previous investigations and are shown, under certain conditions, to have utility in the identification of ancient landslides. Reflection surveys accurately reproduced the subsurface geometry of horst and graben structures and imaged finer scale features such as ground cracks and fissures. Where more complete disintegration of the bluff occurred, GPR reflections from within the slide mass are generally chaotic and include no recognizable evidence of the original stratigraphy. Common midpoint surveys estimated GPR velocity in the sediment and allowed the conversion of travel times to depths.
Key Words: earthquake hazards, ground-penetrating radar, landslides.
A subduction zone earthquake (Mw = 9.2) struck Alaska on March 27, 1964, causing extensive subsidence (locally >2 m) along the south-central coast (Plafker, 1969). The earthquake also triggered thousands of landslides, including many in Anchorage. The landslides in Anchorage were especially common along the margins of steep coastal bluffs (Hansen, 1965) and were responsible for several casualties and damage to >300 buildings in the city (Grantz et al., 1964). In 1998, we revisited the sites of landslides at Government Hill and Turnagain Heights (Figure 1) and used ground- penetrating radar (GPR) to investigate the stratigraphic record of this event. The objectives of this investigation are to (1) obtain high-resolution GPR imagery of landslides that were triggered by intense ground shaking, (2) compare the GPR-based structure with well-documented geotechnical and geological observations, and (3) develop criteria for recognition of ancient landslide deposits in the geologic record.
Anchorage is the largest city in Alaska and is located in the south central part of the state ~130 km west of the 1964 earthquake epicenter (Figure 1). It is built on a relatively flat, triangular-shaped plain that lies between the Chugach Mountains and two bifurcating arms of Cook Inlet. Downtown Anchorage and the port facilities, which suffered the most severe landslide damage in 1964, occupy the northwestern part of the plain along the coast of Knik Arm. Steep coastal bluffs and deeply incised stream valleys exhibit up to 50 m of relief in this part of the city. Water depths exceed 100 m in the center of Knik Arm, 2 km seaward of the Turnagain Heights landslide. Spring tidal range is ~10 m.
The city is underlain by up to 180 m of late Quaternary deposits that are heterogenous in texture (Miller and Dobrovolny, 1959). The stratigraphy generally consists of (1) glacial till unconformably overlying bedrock, (2) silty clay of the Bootlegger Cove Formation resting on till or bedrock, and (3) outwash sand and gravel on the surface (Miller and Dobrovolny, 1959). Till locally crops out in a large moraine complex that lies just north of the city, marking the most recent advance of late Pleistocene glaciers into the area. The finely laminated silty clays of the Bootlegger Cove Formation were deposited in front of the glaciers in a glacialmarine to glacial-lacustrine setting. These fine-grained deposits average 30-45 m thick in the vicinity of downtown Anchorage and contain discontinuous layers of sand and scattered pebbles (Seed and Wilson, 1967); water content is often at or near saturation. The surficial outwash is thickest (up to 18 m) near the moraine and thins toward the south and west, forming a wedge-shaped deposit of sand and gravel. In undisturbed sections, the contact between outwash and the underlying clay is nearly horizontal. The different properties of the sediment greatly influenced the distribution and magnitude of earthquake damage in 1964 (Hansen, 1965).
Earthquake-induced landslides in Anchorage occur in a variety of settings and exhibit different forms. The most damaging landslides primarily moved by translation rather than rotation and slid laterally along subhorizontal surfaces (Hansen, 1965). The failures developed within the Bootlegger Cove Formation, not in the overlying outwash deposits (Hansen, 1965; Seed and Wilson, 1967). Extensive geotechnical investigations of this formation by Shannon and Wilson, Inc. (1964) detected a central weak zone located between upper and lower zones of stiff, competent clay. Loss of strength in the sensitive central zone resulted from the fabric collapse of quick clay during the long duration of ground shaking. Liquefaction of sand layers and lenses within the clay deposits might have partially contributed to the failure of the bluffs (Seed, 1968). Stratigraphic sections (Figure 2) were produced shortly after the earthquake and helped elucidate the mechanics of these dramatic bluff failures (Hansen, 1965; Seed and Wilson, 1967).
We surveyed the Government Hill and Turnagain Heights landslides with a pulseEKKO 100 GPR system. This relatively new technology, pioneered by Ulriksen (1982), enables high-resolution mapping of bedrock and soil stratigraphy and is similar to seismic reflection techniques (Davis and Annan, 1989). The GPR system includes a transmitter, a receiver, a small electronics console, and a laptop computer. A constant separation of ~1 m is maintained between the transmitter and receiver as both are incrementally moved along a profile (Figure 3). The transmitter sends a short pulse, or shot, of high frequency (10-1000 MHz) electromagnetic energy into the ground and the signal reflects off an interface between electrically contrasting materials. The receiver precisely measures the two-way travel time of this reflected signal. We used 100-MHz antennas, which provide a good compromise between depth of penetration and scale of resolution. Image quality was optimized by stacking multiple shots in the field (up to 64 at each point along a profile) and subsequent processing of the digital data. Topographic corrections were surveyed with a handheld level and stadia rod. Reflection profiling with GPR provides extensive and continuous observations of subsurface structure, unlike widely spaced cores or limited outcrops.
Common midpoint (CMP) surveys also were performed at each site to determine the velocity of radar through the ground and thus better constrain the depth of subsurface features. Whereas a reflection survey bounces radar energy off a series of points along a reflector, all shots in a CMP survey are centered on a single point on that same reflector (Figure 3). The antenna separation is not constant, rather the transmitter and receiver are moved in opposite directions; the distance between them increases by a set increment with each shot. By expanding the horizontal separation symmetrically about a central point, the system measures the travel times along multiple paths through the same geologic unit. The velocity analysis assumes a constant velocity model for the Earth, where the travel times for signals along these paths varies in a hyperbolic manner. After adjustments for normal moveout (i.e., changes in path length), the CMP traces are stacked or added together along hyperbolae of many different velocities. Where the modeled velocities are too fast or too slow, the traces do not stack coherently. At the correct velocity, however, the signals do stack coherently and thus are amplified. By plotting and visually picking the highest amplitude signal, we determine the best-fit velocity for a given unit. This estimated velocity may provide insight into the type of sediment present, and when applied to reflection profiles, it allows the conversion from travel time to depth.
Government Hill is a flat-topped bluff ~35 m above sea level that is located ~1 km from the shore of Knik Arm, a branch of Cook Inlet (Figure 1). The 1964 slide occurred along a steep, south-facing slope above Ship Creek, a small coastal stream that has deeply incised the outwash plain. The landslide measured ~360 m parallel to the bluff and reached 120 m behind the original bluff line, involving almost 700,000 m3 of sediment (Hansen, 1965). Lateral slippage of the unconfined bluff face was accompanied by the retrogressive formation of several large graben, each ~30 in wide, and the formation of a large pressure ridge at the leading edge (Figures 2A and 4).
Approximately 400 m of GPR reflection profiles were collected at the Government Hill site. The profiles started on the bluff top and crossed the slide from north to south in the general direction of slippage. A strong, continuous reflection is visible in the subsurface and mimics the shape of the ground surface (Figure 5). This reflection is interpreted as the upper contact of the clay-rich Bootlegger Cove Formation, which is buried beneath outwash sand and gravel. Abrupt changes or offsets in the depth of the reflection may represent normal faults along the margins of the graben. The profile descends ~7 m down the steeply sloping scarp to the flat floor of the innermost graben. Outwash deposits in the graben are 3-4 m thinner than in the adjacent undisturbed section. Narrow, wedge-shaped packages of strong GPR reflections in the graben floor probably represent the parallel series of ground cracks or crevasses that appear in photographs from 1964 (Figure 4). We believe that natural deposition or landscaping filled the cracks with loosely packed sediment, whose bulk density differs from surrounding landslide debris and thus creates the anomalous, wedge-shaped packages of GPR reflections. A displaced, flat-topped block (horst) on the south side of the graben exhibits little stratigraphic disturbance and no ground cracks. The GPR profile crosses over the block and down into a second graben that opened on the south side (Figure 5). Note the disruption of the originally horizontal outwash/clay contact beneath the scarp and graben floor. Extensive regrading on this part of the slide has removed a second displaced block that is visible in 1964 photographs (Figure 4).
In conjunction with the reflection profiles, a CMP survey also was performed adjacent to the Government Hill landslide. In this area of the original bluff, ~15 m north of the main scarp, the stratigraphic section is relatively undisturbed and experienced no significant ground cracking during the earthquake (Hansen, 1965). Based on data from the CMP soundings (Figure 6A), we estimate an average velocity of 0.12 m/ns through the upper unit of sand and gravel (Figure 6B). Given ~200 m of two-way travel time, we calculate a depth of 12 m to the prominent reflection in Figure 5. This value is in agreement with direct measurements made by Hansen (1965) just after the earthquake (Figure 2A). No additional reflections occur deeper in the section; thus, no velocity estimates are available for the underlying clay. The best method to measure velocity in the clayey sediment is crosshole GPR, a technique in which GPR antennas are lowered down a pair of closely spaced boreholes (e.g., Barnhardt et al., 1999).
The 1964 earthquake also triggered a large landslide at Turnagain Heights, a residential area located 5 km southwest of Government Hill (Figures 1 and 7). The neighborhood is built on a flat-topped bluff that is ~21 m above sea level and overlooks Knik Arm. The bluff consists of coarse-grained outwash overlying thick clay deposits of Bootlegger Cove Formation, the same layer-cake stratigraphy present at Government Hill. Hansen (1965, p. A59) described the Turnagain Heights landslide as follows:
". . . the largest, most complex, and physiographically devastating landslide in the Anchorage area. It extended west to east along the bluff line about 8,600 feet (2606 m). Its maximum headward retrogression from the bluff was about 1,200 feet (364 m). . . . A total area of about 130 acres (0.5 km2) was completely devastated by displacements that broke the ground into countless deranged blocks, collapsed and tilted at all odd angles. The ground surface within the slide area behind the prequake bluff line was lowered an average of about 35 feet (11 m) below the old prequake level. The volume of earth within the slide was about 12.5 million cubic yards (9.6 X 106 m3)."
Approximately 600 m of GPR reflection profiles were collected on the eastern part of the Turnagain Heights landslide. Three profiles were oriented north-south, parallel to the general direction of slippage, and crossed the steep, 10-m-high scarp at the head of the slide. In areas above the main scarp, GPR images reveal high amplitude, horizontal reflections at a depth of ~5 m below the surface (Figure 8). The flat-lying reflections represent the contact between coarse-grained outwash and the underlying glacial-marine clay and are only visible beneath undisturbed parts of the bluff, adjacent to the landslide zone. No clear outwash/clay contact appears seaward of the scarp, rather the slide deposits are internally chaotic. The absence of easily recognizable stratigraphy within the slide volume is probably the result of disintegration of the sediment mass during sliding. Unlike the relatively simple horst and graben structures at Government Hill, the slide at Turnagain Heights created "hundreds of sharp crested clay ridges alternating with collapsed troughs. . . (that) ranged in height from about 10 to 15 ft (3.0-4.5 m)" (Hansen, 1965 p. A61)." The rugged topography of the slide deposits has experienced only minor modification by erosion, and a large area is preserved in the state it was in 1964. By necessity, fieldwork was confined to trails and roads due to impassable willow thickets and many water-filled depressions. The profile in Figure 8 was collected along an established access trail, where the ground surface has been modified.
A CMP survey also was performed at Turnagain Heights on top of the original bluff. Local ground cracks developed at this location during the 1964 earthquake but the stratigraphy remains relatively intact (Hansen, 1965). Based on the CMP data, we calculated a velocity of 0.10 m/nsec for the upper unit of sand and gravel. By using this velocity, the prominent reflector at 100 ns (two-way travel time) represents a depth of 5 m (Figure 8), approximately the same thickness as observed in cores and outcrops (Figure 2; Seed and Wilson, 1967). Although the velocity is slower than through comparable material at Government Hill, it is well within the range of empirically derived values for coarse-grained sediment. The lower velocity at Turnagain Heights may indicate higher water content, an important factor that may partially explain the catastrophic nature of the landslide.
At both the Government Hill and Turnagain Heights landslides, the principal cause of translatory movement was the dramatic reduction of soil-shear resistance during straining (Shannon and Wilson, Inc., 1964; Seed and Wilson, 1967; Seed, 1968). This reduction, termed "sensitivity," is expressed as the ratio of peak-to-remolded shear strength. The sensitivity of the Bootlegger Cove Formation varies between ~5-30 (Seed and Wilson, 1967). This means that in the zone of maximum sensitivity, the strength of the remolded clay (post earthquake) is as little as 1/30 of its peak (pre-earthquake) value. After the peak shear strength is reached during cyclic loading, a precipitous drop in shear resistance allows for large inertially driven displacements to occur. It is important to note that this greatly reduced shear strength persists, leaving the soils susceptible to future failure even under substantially lower stress than experienced in 1964. Peak strength is not regained in the time scale of up to 10 years; rather it can require consolidation processes acting over considerably longer periods of time (100-1000 years), especially for soils undergoing "fabric collapse (Mitchell, 1993).
At Government Hill, the inertial-displacements resulted in an observed morphology of laterally displaced glide-blocks (Figures 2A, 4, and 5). The blocks consist of outwash deposits and part of the underlying clay deposits, which were rafted as a largely intact crust atop a softer deforming zone. There is little internal deformation of the blocks themselves, which lie between and adjacent to large graben. A house on one block moved laterally ~10 m during the quake (Hansen, 1965), but suffered little damage and remains occupied today. Morphology of the landslide suggests that the residual shear strength of the deforming clay unit exceeded the gravitational shear stress induced on the slide mass. That is, it appears that after earthquake motions ceased, the blocks did not continue to accelerate downslope and disintegrate. The sliding at Government Hill was certainly dramatic and damaging but not characteristic of a disintegrative-flow morphology. Varnes (1978) classifies this relatively simple type of landslide as an earth block glide.
The morphology and magnitude of deformations at the Turnagain Heights landslide characterize it as an earth lateral spread (Varnes, 1978). At this site, the internal structure of slide deposits is chaotic, and the original stratigraphy is unrecognizable in GPR profiles (Figure 8). The bluff apparently lacked substantial buttressing to resist horizontal shear stresses (Wilson, 1967), and observers noted that lateral sliding toward Knik Arm continued long after the shaking stopped (Seed and Wilson, 1967). The ultimate residual-shear resistance apparently fell below the downslope gravitational-shear stress, such that the mass continued to deform after inertial loading ceased. These continued motions are indicative of disintegrative-flow behavior, the state in which the residual-shear resistance is less than the slope stress. Once the slide mass was mobilized, gravitational-shear stress alone exceeded the strength of the remolded clay, leading to very large deformations and even catastrophic flow-sliding. Certainly, at Turnagain Heights the resultant morphology and incoherent stratigraphy of the landslide deposits is indicative of just such an unlimited and highly destructive flow slide.
Most of the two sites are preserved as city parks today, with landscapes that remain relatively unmodified since the 1964 earthquake. Future earthquakes may cause additional movement along these sections of bluff, but few occupied structures will be at risk. Despite the potentially unstable soils, however, development on adjacent areas has continued since 1964. A mix of private housing, military facilities, and fuel storage tanks are built on or near old landslides that line the margin of Government Hill (Varnes, 1969; Updike and Carpenter, 1986). The City of Anchorage initially prohibited development at the Turnagain Heights site, but new home construction has resumed and the city has spent >$2 million to build roads, sewer, and water lines across the landslide debris (Doogan, 1998). Although inclonometer analyses from 1965-1980 detected negligible strain in the Turnagain Heights area (Updike, 1983), evaluations of slope stability indicate that further development should proceed with caution (Updike et al., 1988).
GPR clearly imaged the internal structure of coastal landslides that were caused by the 1964 Alaskan earthquake. The contrasting electrical properties of late Quaternary sediment in the landslide zones were particularly well suited for the high-resolution imaging technique. GPR signals penetrated through surficial layers of sand and gravel to depths of at least 10 m, and strong reflections were produced at the contact with underlying clay deposits. Different styles of deformation occurred at the two sites. Large blocks and graben formed as Government Hill slid onto a subaerial valley floor; the bluff at Turnagain Heights more completely disintegrated as it catastrophically slid out into deep water of Knik Arm. The stratigraphic sections derived from GPR data compare well with sections that were measured shortly after the 1964 earthquake. Electrically conductive clay deposits rapidly attenuate GPR signals, however, and GPR was unable to image the zone or surface along which the failures actually occurred. The velocity of GPR signals is also different at the two sites, indicating differences in water content, an important factor in soil strength. Combined with traditional geotechnical and geological methods, GPR has great utility in evaluating landslide zones in a rapid, nondestructive manner. GPR surveys are also capable of recognizing ancient landslide deposits, at least in the case of Government Hill where the original stratigraphic contacts are only moderately deformed.
We thank Frederic (Ric) Wilson for his support of this project, especially the fine lodging and emergency loan of a field computer. Lorin Amidon, Boyd Benson, and Nathaniel Wilson were of great assistance in the field. Diane Brenner of the Anchorage Museum of History and Art provided historic photographs. We also thank Les Davis of Sensors and Software, Inc. for his patient tutelage on GPR methodology. This manuscript benefited from reviews by G. Plafker, M. Rymer, J. Kelley, and an anonymous reviewer. Any use of trade names is for descriptive purposes only and does not necessarily indicate endorsement by the U. S. Geological Survey.
Barnhardt, W. A., Jaffe, B. E., and Kayen, R. E. (1999). Evaluation of landslide hazards with ground-penetrating radar, Lake Michigan Coast. In N. C. Krause and W. G. McDougal (Eds.), Proceedings of Coastal Sediments '99 Conference (pp. 1153-1165). Hauppage, NY: American Society of Civil Engineers.
Davis, J. L., and Annan, A. P. (1989). Ground-penetrating radar for high-resolution mapping of soil and rock stratigraphy. Geophys Prospect, 37, 531-551.
Doogan, M. (1998, April 10). Turnagain property owners want the rest of us to pay their bills. Anchorage Daily News, p. B1.
Grantz, A., Plafker, G. and Kachadoordian, R. (1964). Alaska's Good Friday earthquake, March 27, 1964, a preliminary geologic evaluation. Reston, VA: U.S. Geological Survey Circular 491.
Hansen, W. R. (1965). Effects of the earthquake of March 27, 1964, at Anchorage, Alaska. Reston, VA: U.S. Geological Survey Professional Paper 542-A.
Miller, R. D. and Dobrovolny, E. (1959). Surficial geology of Anchorage and vicinity, Alaska. Reston, VA: U.S. Geological Survey Bulletin 1093.
Mitchell, J. K. (1993). Fundamentals of soil behavior. New York: John Wiley and Sons.
Plafker, G. (1969). Tectonics of the March 27, 1964 Alaska earthquake. Reston, VA: U.S. Geological Survey Professional Paper 543-I.
Seed, H. B. (1968). Landslides during earthquakes due to soil liquefaction. J Soil Mech Found Div, ASCE, 94, 1053-1122.
Seed, H. B. and Wilson, S. D. (1967). The Turnagain Heights landslide, Anchorage, Alaska. J Soil Mech Found Div, ASCE, 93, 325-353.
Shannon and Wilson, Inc. (1964). Report on Anchorage are soil studies, Alaska, to the U.S. Army Engineer District, Anchorage, Alaska. Seattle, WA.
Ulriksen, C. P. F. (1982). Application of impulse radar to civil engineering. Unpublished Ph.D. thesis. University of Technology, Lund, Sweden.
Updike, R. G. (1983). Inclinometer strain analysis of Anchorage landslides, 1965-1980. Anchorage, AK: Alaska Division of Geological and Geophysical Surveys Professional Report 80.
Updike, R. G., and Carpenter, B. A. (1986). Engineering geology of the Government Hill area, Anchorage, Alaska. Reston, VA: U.S. Geological Survey Bulletin 1588.
Updike, R. G., Olsen, H. W., Schmoll, H. R., Kharaka, Y. F., and Stokoe, K. H., 11 (1988). Geologic and geotechnical conditions adjacent to the Turnagain Heights landslide, Anchorage, Alaska. Reston, VA: U.S. Geological Survey Bulletin 1817.
Varnes, D. J. (1969). Stability of the west slope of Government Hill Port Area of Anchorage, Alaska. Reston, VA: U.S. Geological Survey Bulletin 1258-D.
Varnes, D. J. (1978). Slope movement types and processes. In: R. L. Schuster and R. J. Krizek, (Eds.), Landslides: Analysis and control (pp. 11-33). Washington, DC: Transportation Research Board, National Academy of Sciences Special Report 176.
Wilson, S. D. (1967). Landslides in the City of Anchorage. In F. J. Wood (Ed.), The Prince William Sound, Alaska, earthquake of 1964 and aftershocks (pp. 253-297). Washington, DC: U.S. Department of Commerce, Coast and Geodetic Survey.
Robert E. Kayen
contact: Robert Kayen
last modified 2018