USGS Bulletin 2130 -- CHAPTER 13

DETECTION OF THE BASE OF SLUMGULLION LANDSLIDE, COLORADO, BY SEISMIC REFLECTION AND REFRACTION METHODS

Detailed studies of the surface features of the Slumgullion landslide, located in the San Juan Mountains near Lake City, Colo., have been conducted since 1991. These studies have attempted to explain many of the characteristics of the slide based on surface samples and measurements. However, because no borehole data exists, seismic reflection and refraction data can potentially provide subsurface data over a wide area of the slide. This would complement the surface studies and provide information critical to determining landslide volume, pre-slide valley morphology, internal slide stratigraphy, geotechnical properties, and potential locations for future drilling.

To test the usefulness of seismic methods, north-south-oriented P-wave refraction and reflection profiles were acquired over 400 m of the slide in June 1993. The coincident profiles were sited on an older portion of the slide, below the active toe and east of State Highway 149 (fig. 1). The profile location was selected for easy vehicle access, relatively flat surface topography, and its position within a well-known survey grid. Despite the relatively good access, locally steep topography and seismic cabling limitations prevented acquisition of seismic data on the northernmost and southernmost portions of the slide. Data quality was also degraded by poor geophone coupling at the northern 50 m of the seismic profiles where the seismic line crossed a marshy beaver pond. Generally, data quality is good over the length of the profile, and the base of the slide was easily detected in both the reflection and refraction data so that the general shape of the pre-landslide valley could be interpreted.

Conventional techniques were used to acquire and process the data (table 1). For the reflection data we used standard common-depth-point (CDP) acquisition and processing methods (Mayne, 1962; Yilmaz, 1987). A 24-channel recording system, a shotgun source, and 28-Hz geophones were used to produce the 12-fold, 1-second CDP profile. Receiver spacing was 4 m, source spacing was 8 m, with two shot records acquired at each of the receiver stations.

For refraction data acquisition, we used two receiver arrays of 24 channels spread over 230 m. For each array we fired shots in the middle, at the ends, and, where possible, at a distance off the end of the array. The first refracted arrivals were visually picked from a display of the shot records (fig. 2) and analyzed using a ray-trace modeling program. The program starts by generating an approximate depth model based on the delay-time method described by Pakiser and Black (1957), then it uses this depth model as input for iterative ray tracing. Model adjustments are then made by the computer to minimize the discrepancies between the picked travel times and those determined by ray tracing (Scott and others, 1972). Some key assumptions made in the refraction analysis are: (1) layers are continuous and extend from one end of the refraction line to the other, and (2) layer velocities increase with layer depth.

The 90-Hz dominant frequency of the shotgun reflection data, combined with stacking velocities of 1.7 km/s, permits a minimum vertical resolution of about 4-5 m at 70 m depth. These resolution limits are based on the 1/4 wavelength criteria of Widess (1973). Clear hyperbolic reflection curves like the curve labeled "slide-base reflection" in figure 3 produce stacking velocities with an estimated uncertainty of 10 percent for the slide-base reflection. This results in depth estimates having errors of±5.0 m for a bed at 70 m depth.

An additional depth uncertainty of as much as 12 m is introduced by the assumption that the reflection data is minimum phase. This means that one assumes that most of the energy in the seismic wavelet produced by an explosive source is concentrated in the front of the wavelet, and, hence, the front end of the seismic wavelet gives the travel time of the wavelet from the source down to the reflector and up to the receiver. This minimum-phase assumption seems accurate, however, because the slide-base refractor depths derived from the refraction data (also minimum phase) correspond to the depths derived from interpreting the onset of the reflection as the top of the reflector. If this phase assumption is not correct, then a large uncertainty exists because the top and bottom phases of the doublet reflection wavelet are as much as 12 m above and below the central negative phase. The dominant frequency of the first-arrival refraction data is about 25 Hz, so the reflection data, with shorter geophone intervals and higher frequency seismic waves, does give better resolution of the slide-base interface.

Acquisition parameter	Refraction data	Reflection data
Source:	1/3 lb. Kinestik	12-ga. shotgun, 1-oz. slug
Shot hole depth:	1 m	0.3 m
Shots per station:	10 for total profile	two, on alternate stations
Source point interval:	about 50 m	8 m
Geophone array:	12, 28-Hz, bunched	same
Geophone interval:	10 m	4 m
Recording geometry:	24 channels, reversed	24 chan., in-line, off-end
Near trace offset:	5-210 m	8-16 m
Recording pass band:	0-500 Hz	20-500 Hz
Recording system:	I/O DHR-2400	I/O DHR-2400
Sampling interval:	0.001 sec	0.001 sec
Record length:	1.0 sec	1.0 sec

7. Normal-moveout correction (from 1.45 km/s-south end, to 1.9 km/s-north end at 0.08 s)

9. Predictive deconvolution filter (prediction distance 0.01 s, operator length 0.1 s)

13. Time-to-depth conversion (using velocities derived from refraction interpretation)

Migration of the shotgun reflection data degraded data quality and is not shown here.

A large, positive acoustic contrast is interpreted to represent the base of Slumgullion slide. We infer the base of the slide from a clear deflection in refraction first-arrival phases to a high-seismic-velocity unit on all refraction shots (fig. 2) and as a prominent reflection at 0.09 - 0.12 s on many of the reflection shots (fig. 3). The refraction data indicate that the seismic velocity jumps from 2.3 km/s within the slide to 5.4 km/s for the material underlying the slide. This 5.4 km/s value is characteristic of hard crystalline rock, such as granite or volcanic rock (Dobrin, 1976). Layered, densely welded tuffs were mapped by Lipman (1976) adjacent to the slide near the seismic profiles, and their shallow dip (5^o) suggests that they underlie the slide. The seismic velocity of the rock underlying the slide is close to the value obtained by Gomberg and others (this volume) at a location about 1.5 km east of the seismic refraction/reflection site.

Previous studies of the pre-landslide topography of the Slumgullion landslide valley were conducted by projecting the slope of the existing valley walls (outside the boundaries of the slide) to a point of intersection under the flow (Parise and Guzzi, 1992). These projections, which assumed a V-shape pre-landslide topography, were made about 200 m west and 300 m east of the seismic profile locations and suggested a maximum depth of about 120 m.

From the refraction and reflection data, we interpret the pre-landslide valley topography to be generally U-shaped, with a 25^o slope on the north end of the profile (position 28 - 44, figs. 4 and 5), a flat-to-undulating middle section (position 13 - 25), and 10^o - 20^o slope on the south end of the profile (position 0 - 13). Projecting the interpreted slide base north of the seismic profiles suggests that bedrock reaches the surface 20 to 30 m north of position 44 (the rock is covered by surficial deposits in the field). The interpretation of the slide base south of position 13 is uncertain because the reflection data quality degrades and we are left to depend solely on the refraction data. Topographic constraints beyond the south end of the profile prevented acquiring longer offset refraction shots that would have reduced interpretation uncertainty. For the middle section of the slide, we rely on reflection data to show greater detail.

Beneath position 28, the reflection data show a V-shaped trough about 15 m deep and 100 m wide (fig. 5). South of the trough, the reflection and refraction data results diverge and, because of uncertainties in the refraction data, we present alternate interpretations. Also, out-of-plane reflections may complicate the reflection data and contribute to the mismatch of the reflection and refraction data at any point on the profile. We favor the refraction interpretation that coincides with the strong reflection between stations 14 and 20 (fig. 5). The reflection between positions 10 and 20 shows the base of the slide to be relatively flat-lying at 60- to 65-m depth over a distance of 100 m. South of position 10, the basal reflection weakens, but we infer from the refraction data that the base of the slide slopes up at 10^o-20^o (figs. 4 and 5). Projecting this slope south suggests that bedrock crops out in the vicinity of a small lake (beaver pond, fig. 1), which is in an area mapped as glacial moraine deposits (Lipman, 1976). Bedrock (rhyolite) is exposed off of the slide in the 22^o slope near the lake (Lipman, 1976). An alternate, less certain refraction model shows a small hump in the slide base interface south of position 10 (fig. 4). We interpret no reflections below 110 m because the data are considered to be predominantly incoherent seismic noise below this depth (fig. 5).

The refraction data indicate three distinct P-wave seismic velocity layers within the slide: (1) a 1- to 3-m-thick surficial layer with a velocity of 350 m/s, 2) underlain by a 2- to 30-m-thick unit with a 1,500 m/s velocity, and (3) the basal 20- to 60-m-thick slide layer with a velocity of 2300 m/s overlying bedrock (fig. 4). The seismic velocity and thickness of layer 1 suggest it is a loose, aerated (weathered) soil (Clark, 1966); the ease of planting geophones and digging shotholes in this layer confirmed its relatively loose texture. The seismic velocity of layer 2 suggests that it is a saturated, unconsolidated unit (Clark, 1966) whose top may represent the water-table surface. The velocity of layer 3 is consistent with that of a firm clay (Clark, 1966), and layer 3 could be a more consolidated form of layer 2 or a separate, older slide deposit. These internal layers were not detected by the reflection method.

We estimated S-wave (shear-wave) velocities, in a fashion similar to Nazarian and Stokoe (1984), from surface waves recorded on the P-wave seismic reflection records. Generally, the penetration depth of surface waves is about 1 or 2 wavelengths (Ballard, 1964). For the surface waves recorded in the refraction and reflection surveys, the penetration depth is as much as 25 m. The surface waves were identified on the P-wave common-shot records by their high-amplitude, dispersed wave train, and low speed (fig. 3). In this study, we have assumed the shear-wave velocity to be 10 percent greater than the surface-wave group velocity; empirical studies show that the surface wave velocity is generally 10 percent slower (Ballard, 1964; R. Williams, unpub. data, 1994). By this approach, we found two distinct S-wave velocity layers: from 0-5 m depth, surface waves recorded on the shotgun reflection data with a period of 0.03 s correspond to a S-wave velocity of 200 m/s, whereas surface waves recorded on the refraction data with a 0.05 s period correspond to a depth of 5-25 m with a S-wave velocity of 400 m/s.

The principal finding of this study is that the contact between slide and non-slide material is easily detectable by seismic reflection or refraction methods. The 5.4 km/s seismic velocity and the local geology suggest that the rock underlying the slide is a densely welded tuff. From the reflection and refraction data, the pre-landslide valley appears to be broadly U-shaped with a maximum depth of about 95 m. Three seismic velocity layers were detected within the slide by seismic refraction methods: a thin weathered layer at the surface, underlain by saturated unconsolidated material, and a more consolidated basal layer that makes up the majority of the slide. Shear-wave velocities calculated from recorded surface waves range from 200-400 m/s in the upper 25 m of the slide. Finally, the subsurface data described in this paper show how geotechnical sampling and testing of slide material at the surface would not be representative of the majority of the slide material.

William Z. Savage and William K. Smith surveyed the seismic station locations and reduced that data. Philip S. Powers provided us with the map of our seismic line (used for fig. 1) on the base topography.

Accessibility FOIA Privacy Policies and Notices

U.S. Department of the Interior | U.S. Geological Survey
URL: http://pubsdata.usgs.gov/pubs/bul/b2130/Chapter13.html
Page Contact Information: GS Pubs Web Contact
Page Last Modified: Friday, 23-Mar-2012 14:21:29 EDT