THE SLUMGULLION LANDSLIDE (SOUTH
Transcription
THE SLUMGULLION LANDSLIDE (SOUTH
THE SLUMGULLION LANDSLIDE (SOUTH-WESTERN COLORADO, USA): INVESTIGATION AND MONITORING MARIO PARISE CNR-IRPI, National Research Council of Italy, Bari, Italy JEFFREY A. COE, WILLIAM Z. SAVAGE, DAVID J. VARNES* U.S. Geological Survey, Geologic Hazards Team, Golden, Colorado, USA *deceased ABSTRACT: This paper summarizes the research performed at the Slumgullion landslide. The landslide consists of an active slope movement currently moving on the upper-middle part of an older, larger, and inactive landslide. After the first specific study carried out in 1958, the landslide has been the object of detailed investigation and monitoring starting from 1990. These, which will be briefly described here, consisted of: high-scale and multi-temporal mapping of the features produced by the landslide movement at the ground surface; geological characterization of the materials involved in landsliding; evaluation of the hazards, with particular reference to deformation of the area ahead of the active toe; geophysical investigation; monitoring of the landslide by means of surveys, photogrammetric methods, and GPS methods; comparison of landslide movement with meteorological data, and assessment of the relationships between movement and rainfall and snow melting. Keywords: landslide, monitoring, earth flow, hazard 1 INTRODUCTION The Slumgullion landslide (Fig. 1), in the San Juan Mountains of south-western Colorado, is one of the most famous slope movements in North America. The landslide has attracted the attention of many investigators since the XIX century beginning with Endlich (1876). The name Slumgullion, already ascribed to the landslide by miners at the time of Endlich’s description, is an old term referring to a meat stew containing many different ingredients of various colors and having a variegated appearance. The term accurately describes the variegated landslide deposits, that consist of bright yellow clay-rich materials, reddish brown volcanic rocks, and local concentrations of red and purple clay-rich materials. In addition to a wide variety of colors, the landslide deposits show a strong lateral variability, with individual units typically elongated in the direction of movement which results in a striped appearance. The Slumgullion landslide is a complex feature consisting of an active landslide currently moving on the upper-middle part of an older, larger, and inactive landslide (Fig. 2). After Endlich (1876), who described it in a short note in the report of the Hayden Survey of 1874, it was cited by numerous other investigators in the early part of the XX century (Cross, 1909; Howe, 1909; Atwood, Mather, 1932; Burbank, 1947). The earliest photographs of the landslide were published at the beginning of last century by Whitman Cross (plate XXB in Howe, 1909). In 1958, Crandell and Varnes made the first measurements of movement on the Slumgullion landslide (Crandell, Varnes, 1960; 1961). Observations of changes in survey lines crossing the slide and in advances of the active toe were then made several times between 1958 and 1973. In 1990, the U.S. Geological Survey (USGS), with the assistance of scientists provided by a cooperative agreement between the USGS and the Italian National Research Council (CNR), began a new study of the landslide. The preliminary results of this new phase of study have been published in several USGS open-file reports and in a Slumgullion-dedicated USGS Bulletin (Varnes, Savage, 1996). Figure 1 - Aerial view of the Slumgullion landslide. Lake San Cristobal is at lower right. The inactive landslide extends from the lake uphill to the active toe. Active toe is prominent lightcolored bulge uphill from where Colorado State Highway 149 crosses the landslide. In 1998, Brigham Young University (BYU) and the USGS began work on a National Aeronautics and Space Administration (NASA) funded study to monitor hydrologic and meteorologic fluctuations on and in the landslide and attempt to relate these to movement measured using interferometric capable Synthetic Aperture Radar (INSAR) and Global Positioning System (GPS) surveys. Results from this study are presented in Coe et al. (2003). This paper is a summary of research that has been carried out since 1990 at Slumgullion. Even though some studies are still on-going (e.g. periodic campaigns for GPS and leveling measurements), many interesting results have been obtained so far. These results are briefly described in this paper. Observations dealing with kinematics of the movement at Slumgullion site showed that most of the movement of the active part of the Slumgullion earth flow takes place along the shear surfaces which bound the landslide (Crandell, Varnes, 1961; Guzzi, Parise, 1992; Baum, Fleming, 1996; Fleming et al. 1996, 1999); therefore, it is most correct to refer to it as an earth slide or landslide (Cruden, Varnes, 1996). However, since it has often been described as an earth flow (and sometimes also used as typical example for this category of slope movement), the term earth flow has become, and still remains, traditionally popular for the Slumgullion. In this paper, we use the term landslide to describe the Slumgullion slope movement. Meteorological data collected between October, 1980 and March, 2002 at a Snotel weather station (Western Regional Climate Center, U.S. Department of Agriculture, unpublished data, 1999) located about 4 km southeast of the head of the slide at an elevation of 3487 m, reveal the following observations about precipitation and temperature conditions in the area (Coe et al. 2000b, 2003). Annual precipitation totals ranged from 500900 mm, about two-thirds of which came in the form of snow between October and June. Extreme temperatures ranged from 33˚C on February 1, 1985 to 27˚C on July 4, 1989. Average daily temperatures ranged from -28˚C on February 1, 1985 to 16˚C on July 4, 1989. Figure 2 - The Slumgullion landslide, in southwestern Colorado, showing boundaries of the active and inactive portions. 2 GEOLOGIC AND GEOMORPHIC SETTING The Slumgullion landslide is located in the San Juan Mountains, a large Tertiary volcanic center in south-western Colorado. The landslide occupies the valley formed by Slumgullion Creek, a tributary of the Lake Fork of the Gunnison River. It formed as a result of collapse of hydrothermally altered volcanic materials in the rim of the Lake City caldera (Lipman, 1976), on the southern end of Mesa Seco. The detached materials slid and flowed downhill to block the Lake Fork of the Gunnison River, creating Lake San Cristobal. Originally the lake was 4.3 km long, with an area of 1.8 km2, but sediment deposited in deltas of Slumgullion Creek and Lake Fork have reduced its length to 3.3 km and its area to 1.34 km2 (Schuster, 1985; Parise, Moscariello, 1997); it has a maximum depth of 27 m and a volume of about 14 million m3. Except for an average of 10 m of erosional downcutting of the natural outlet channel across the landslide dam, little change has occurred in the geometry or character of the landslide dam since it was formed. Headward erosion of the downstream channel has created the 25-m high Argenta Falls at a resistent bedrock ledge. The continued stability of this natural dam is of great importance to Lake City and the surrounding area. Because the channel now appears to be stable in regard to erosion, there is no reason to expect failure of this broad, fairly flat dam (Schuster, 1996). The materials exposed in the scarp are hydrothermally altered and unaltered volcanic rocks including tuffs and intrusive and extrusive rocks of variable compositions (Diehl, Schuster, 1996). At the top of the 250-m high scarp is the Miocene Hinsdale Formation, an unaltered vesicular basalt, which is underlain by the Miocene Sunshine Peak Tuff, a welded ash flow tuff (Fig. 3). The Miocene units unconformably overlie the highly altered Oligocene Uncompahgre Peak volcanics, which consist mainly of andesite flows (Lipman, 1976). Alunite, smectite, and opal, are the main products of hydrothermal alteration, and can be found in the main scarp as well as in the landslide debris (Larsen, 1913; Diehl, Schuster, 1996). The hydrothermal brecciation and alteration, combined with the weakening of the rock mass by intersections of numerous faults, was apparently a prime factor in the origin of the Slumgullion landslide. The bedrock surrounding the landslide consists of volcanic rocks ranging in composition from rhyolite to basalt. A few glacial erratics of granite and related crystalline rocks occur in glacial deposits south of the landslide. Figure 3 - View of the main scarp. In the Slumgullion area, as well as in many other sites of the San Juan Mountains, the topography is generally controlled by the erosional resistance of different types of volcanic rocks. The steep cliffs are composed of lava flows, and gentler topography is formed on unconsolidated and altered tuffs. The pre-landslide topography of the valley of Slumgullion Creek was the object of a study by Parise and Guzzi (1992): on the basis of analysis and comparison of shape and morphology of nearby valleys in this sector of the San Juan Mountains to the valley of Slumgullion Creek, they reconstructed the original topography, and estimated an average thickness of the displaced material at about 40 m (Parise, Guzzi, 1992). In this reconstruction, the maximum thickness of landslide deposits is about 140 m; it occurs where the toe of the active landslide has piled material on top of the inactive part of the landslide. This point is about 250 m uphill from where Colorado State Highway 149 crosses the inactive landslide (Figs. 1 and 2). Based on the results of seismic reflection and refraction profiles, Williams and Pratt (1996) estimated a maximum thickness of about 95 m for the landslide, along a profile below the active toe and east of State Highway 149. In the same area, Parise and Guzzi (1992) suggested a thickness of about 120 m. The difference can partly be explained when considering that the pre-landslide topography reconstructed by Parise and Guzzi assumed a V-shape prelandslide valley, and, on the other hand, Williams and Pratt (1996) interpreted the pre-landslide valley topography to be Ushaped, so they obtained a lesser value for the thickness of the deposit. The seismic surveys were locally limited because steep topography prevented acquisition of data across the boundaries of the earth flow and along the stable slopes (Williams, Pratt, 1996). The shape of the valley underlying the inactive landslide could best be resolved by drilling a few boreholes in the area downhill from the active toe. much larger, inactive part. The active part of the landslide is 3.9 km long, covers an area of 1.46 km2, and has an estimated volume of about 20 x 106 m3 (Parise, Guzzi, 1992). Total elevation difference from the toe of the inactive part at the Lake Fork of the Gunnison River to the top of the 250-m high head scarp is about 1000 m (Table I). The length from the top of the scarp to the inactive toe at the level of the Lake Fork is about 7 km. Part of the inactive toe is concealed under the water of Lake San Cristobal. Table I – Morphometry of inactive and active parts of the Slumgullion landslide Feature Dimensions of Dimensions of the inactive landslide the active landslide Area of deposit * 4.74 km2 1.46 km2 Length 6.8 km 3.9 km Width - head 1,130 m 280 m - narrowest part 290 m 150 m - toe 530 m 430 m Relief - elevation of top 3,700 m 3,500 m - elevation of tip 2,700 m 2,960 m Average slope - deposit only 7° (12%) ---- including main scarp 8° (14%) 11° (19%) Thickness - average 40 m 13 m - average on thalweg of buried valley 90-100 m --- maximum 140 m 48 m Volume 6 3 19.5 x 106 m3 - earth flow deposit 168 x 10 m 6 3 ---- detached mass ** 142 x 10 m Length : width > 6:1 9:1 * Including the deposit identified by Chleborad (1993). ** Including a 30% bulking factor; for further details, see Parise, Guzzi (1992). Based on radiocarbon dating, differences in degree of weathering and soil formation, and morphologic observations to separate ages of deposits, several episodes of movement have been identified at the Slumgullion landslide (Crandell, Varnes, 1960, 1961; Chleborad, 1993; Madole, 1996; Fleming et al. 1999): the first episode blocked Slumgullion Creek about 10001,300 years ago; the second episode blocked the Lake Fork and caused the impoundment of Lake San Cristobal about 800-900 years ago; the third episode was collapse of the north part of the headscarp about 300 years ago; the fourth episode is the reactivation of old landslide deposits, and is represented by the current episode of movement. This sequence is based on field evidence and measurements at several single points at the landslide surface. Fleming et al. (1999), however, suggested that there is also the possibility that the above sequence might have occurred as a continuum, rather than in separate episodes. The landslide deposits are fine sand, silt, and clay derived from hydrothermally altered volcanic rocks. Locally, large assemblages of boulders are present; in plan view they have an elongated lenticular shape with the long axis parallel to the movement direction of the earth flow. Overall, these materials are elongated in the direction of sliding and have a strong lateral variability. 3 LANDSLIDE DESCRIPTION 3.1 The active landslide The Slumgullion landslide (Fig. 2; Table I) is 6.8 km long, covers an area of 4.74 km2, and has an estimated volume of about 170 x 106 m3 (Parise, Guzzi, 1992). It consists of a younger, active, upper part that moves on and over an older, Observations of the surface features produced by differential movement of the active part of the Slumgullion landslide have been used to understand the behavior of different parts of the landslide, and to estimate a zonation of landslide elements (Guzzi, Parise, 1992; Baum, Fleming, 1996; Fleming et al. 1999): among the many different types of features which can be observed at the surface of the landslide, the most diffuse and important were scarps, fractures, flank ridges, and basins. Each of these features is related to the style of deformation acting at that moment in that particular part of the landslide (Fleming, Johnson, 1989). Even though many of them are ephemeral, since weathering, erosion, and continued movement tend to erase prior traces of movement, their periodic observation may result in interesting information about the landslide evolution. At the Slumgullion landslide, additional observations were made by investigating both the presence and conditions of trees and vegetative cover. In particular, vegetation on the landslide surface, and its degree of disturbance, provided information to distinguish between active and inactive sectors, and to identify the areas more affected by movement. The continuously active sectors are generally characterized by bare ground, or by the presence of few trees; clear evidence of disturbance can be recognized on trees in the areas of recent movement. Trees have been tilted during the movement, and then grow vertically, so that the relative age of movement can be inferred by the curvature of trunks and by their position. Further information may be drawn from the many places on the landslide where trees have been split by differential displacement across a fracture: the direction of differential movement required to produce the split can be used, together with stretching or buckling of tree roots, as an indicator of local deformation (Fleming et al. 1999). The base of the main scarp is mantled with continuous belt of detrital material consisting of talus cones and rock-fall deposits. The talus cones are built by successive and discontinuous accumulations of debris falling from the near-vertical to overhanging rock walls above. Rock-fall deposits are widespread at the foot of the main scarp and form a large accumulation of clast-supported angular blocks and coarse boulders, which typically show an irregular hummocky topography. The direction of sliding at the head is toward the south and is in the same direction as the inferred collapse of the Mesa Seco. In this area, there is a mixture of compressional and extensional features: the latter are produced by normal faults and tension cracks in the active landslide, while the compressional features are related to the collapse of part of the main scarp (Fleming et al. 1999). A prominent flank ridge is present at the left boundary of the active landslide (Fig. 4); east of it, the landslide head enlarges. Figure 4 - View to southwest from main scarp region showing the upper reaches of active slide with prominent levee on left flank. Downhill from the active source area, a change in the main direction of movement of the landslide material (from a N-S to an about E-W strike) is evidenced by a bend toward the right in the active as well as the inactive landslide. This stretch, which has many normal scarps and tension cracks, is located just before the so-called neck, where width of the landslide reaches its minimum value, and displacement rate is at its maximum (Table I). Narrowing of the landslide is in fact accommodated in an area where the width decreases from the 300 m at the head, to 230 m and then to 150 m. The narrowest part of the landslide is where Crandell and Varnes (1961) measured the maximum displacement rate and established a survey line (Fig. 5). Figure 5 - View from camera station of narrow neck region, where velocities are up to 7 m/yr. The materials entering the narrowest part of the landslide are constrained laterally, and are limited by shear zones on both the sides of the landslide. The narrowing is also accompanied by an abrupt change in slope. The most striking features of the Slumgullion landslide are its flank ridges, that form on both its flanks, and adjacent to shear zones within the landslide as well (Fig. 6). Even though data on formation of the ridges are available from many studies (e.g. Fleming, Johnson, 1989; Baum, Fleming, 1991; Baum et al. 1993), a coherent explanation of the process of formation is still lacking. Many of the proposed mechanisms for formation of flank ridges have been reviewed by Keefer, Johnson (1983), Fleming, Johnson (1989), and Corominas (1995). Growth of flank ridges and the presence of relict flank ridges in both the active and inactive deposits of the slide have been documented at the Slumgullion earth flow. Flank ridges extend continuously for several hundred meters, following the landslide boundaries. Observation of the main geomorphic, pedologic, and sedimentologic characters of the several generations of flank ridges helped in reconstructing the chronology of their formation and to relate them to the landslide history of movement (Parise et al. 1997; Fleming et al. 1999). The great majority of the flank ridges at the Slumgullion landslide appear to have formed along the boundary between active and inactive landslide movement as a result of displacement on the bounding strike-slip shear surface. Flank ridges that have formed along the flanks of a landslide provide a basis to evaluate the history of movement in much the same way as moraines are useful indicators of movement history of a glacier. Analysis of flank ridges performed at the Slumgullion landslide provide indications of past boundaries and height of the landslide; in addition, the distribution of ridges shows evidence for gradual thinning and narrowing of the earth flow (Parise et al. 1997). Figure 6 - Flank ridges at the left boundary of the active Slumgullion landslide. Note the difference in height between the presently forming flank ridge and the older one. An area characterized by the presence of ponds is recognizable further downhill: pond deposits show uniform color and silty texture. About 0.6 km uphill from the active toe, there is a wide area of pond deposits, extending 80 m by 80 m. Apparently, the location of the pond has remained fixed while the landslide material has been displaced across the pond site. Parise and Guzzi (1992) suggested that the pond sediments and thrust features mark the initial position of the toe of the reactivated Slumgullion landslide. Figure 7 is a sketch showing the process of formation and preservation of the pond: after the collapse of part of the scarp, and after that loading in the head of the inactive landslide triggered reactivation, fractures propagated downslope through the inactive landslide material to the approximate position of the pond (Fig. 7A). Then, the fracture along the basal failure surface emerged to the ground surface (Fig. 7B). From this point on, the reactivated landslide advanced over the surface of the old, inactive landslide (Fig. 7C). The active toe has a rounded shape in plan view; it has steep slopes, up to 40 m high, with an average slope angle of 18°. The overall difference in elevation between active and inactive deposits is accommodated through a series of scarps, whose height ranges from 0.5 to 8 meters. The active toe is advancing across the surface of the old, inactive landslide, pushing trees over and engulfing them under the advancing landslide. The toe has an irregular trace along its northern half, which is probably a result of a velocity discontinuity (higher displacement rates have actually been observed at the southern half, south of the major right lateral shear surface of the landslide; Fleming et al. 1996). Observations of the displacement vector of the active toe relative to nonmoving ground farther downhill indicate continuing thickening of the toe (Fleming et al. 1999); therefore, displacement of the landslide tends to steepen the front of the active toe. This could lead deformation in the area ahead of the active toe (see below), as well as favoring minor slope movements at the steep and unstable active front. As a matter of fact, several debris-flow episodes occurred at this site in 1999; debris-flow material deposited ahead of the active toe, covering some measurement points of the level circuits at the inactive landslide surface. The debris-flow deposit consists of subrounded to sub-angular boulders, cobbles and pebbles in a finer matrix. Several small flank ridges, up to 0.4-m high, are also present, showing sub-angular clasts aligned at their top. On the basis of the above described observations, the active part of the Slumgullion landslide can be partitioned into three regions (Guzzi, Parise, 1992; Fleming et al. 1999), characterized by distinct styles of deformation: i) the head of the slide, with prevailing extensional features; ii) the central part, where the width reaches the smallest value (150 m) and surface features related to the strike-slip shear surfaces which bound the landslide predominate; iii) the active toe, showing a series of thrusts and internal toes, which are morphologically expressed by scarps of variable height. 3.2 The inactive landslide Figure 7 - Sketch of pond formation in relation to emergence of the toe of the reactivated landslide (after Fleming et al. 1996). See text for explanation. Downhill from the narrowest part, the landslide widens through lateral steps, where basins (pull-apart like) are formed. One of the best expressed basin is present on the north side, along the right flank of the active landslide; its formation and evolution have been described in detail by Fleming et al. (1999). Downhill from the active toe, the deposits of the inactive landslide are mostly hummocky, but the hummocks are smooth and rounded. Gentle slopes hundreds of meters across, containing closed depressions and ponds, are joined by intervening steep ramps as much as tens of meters high. Along the longitudinal profile extending from the active to the inactive toe, four separate zones can be distinguished (Parise, Moscariello, 1997): the first one is a pond-rich region, with abundance of water, and perennial and ephemeral streams flowing off and out of the active landslide. Then, there is a main bulge, containing several internal ridges; it is comparable in size to the active toe and perhaps represents an old internal toe in the landslide, that has been remodeled by erosion and weathering to Figure 8 - Map of the area near the active toe of the Slumgullion landslide, showing vertical movements at the leveling stations (after Varnes et al. 1996). All movements are downward unless marked “+”. its present rounded shape. Further downhill, an area of steps is present. The surface of the steps consists of rounded, curved downslope, hummocks a few meters high. Finally, the inactive toe occupies the valley of the Lake Fork of the Gunnison River, and its materials extend both upstream and downstream from the point where the landslide deposits entered the valley. The materials in the inactive toe are different from the rest of the landslide: most of the deposit is fine-grained materials, and boulders are rare to absent. 3.3 Hydrology Very few quantitative hydrologic data exist so far for the Slumgullion landslide. It appears, however, that the active landslide, and even the inactive landslide debris, are largely water saturated. Many ephemeral streams flow down the landslide and perennial streams flow down each flank (e.g. the Slumgullion Creek at the left flank of the active landslide). Perennial and ephemeral springs, with discharge of several litres per second, exist at several places on the surface of the landslide. Changes in location and discharge of springs, and diversions in pattern of drainage ways, are controlled by the overall movement of the landslide, and by secondary, shallow slope movements at its surface. Water flowing from the springs can thus flow downhill hundreds of meters before it either soaks into the ground, flows into a sink, or joins a stream. Abundance of surface water on the Slumgullion landslide, and complex relationships between topography, structure, and hydrology of the landslide are yet to be completely understood. In 1998, as part of the NASA-funded study, the USGS began to conduct systematic observations of surface water, landslide movement, and pore pressure data from pressure transducers installed in a shallow (~2.5 m deep) auguer hole. These observations and measurements indicated a close correspondence between available surface water, landslide movement, and pore pressure (Coe et al. 2003). The work also indicated that the pattern of continual, but seasonally variable movement observed, closely fits the bathtub model for landslide movement (Baum, Reid, 2000). In this model, the landslide is isolated both mechanically and hydrologically from adjacent materials by low permeable clays. These clays cause the landslide to retain water and thus be responsive to seasonal changes in infiltration of surface water. (Varnes et al. 1996). The most general deformation is depression, in the range 5-20 mm/yr, which seems to decrease away from the active front. However, some points close to the toe rose significantly between 1991 and 1993 (20-80 mm in 2 years). Continuing movement of the active part of the landslide, which might reach Colorado State Highway 149, was further testified by debris-flow activity in 1999 at the active toe and in the area just ahead of it. Moreover, it must be remembered that several additional hazardous areas are present at Slumgullion. Other than the possibility of a repeat of a major collapse of a significant part of the main scarp, the principal concerns in terms of hazards are rock-fall and talus accumulation processes at the head scarp, debris-flow and flooding processes in Slumgullion Creek, deformation of the inactive landslide deposit downhill from the active toe, and secondary slope movements within the main landslide body. In particular, at the mouth of Slumgullion Creek, an alluvial fan consisting of several debris-flow and flood deposits is identifiable. Due to the presence of several resort facilities close to Slumgullion Creek and within the boundaries of the depositional area of debris-flow materials, the alluvial fan at the mouth of the creek poses one of the most significant hazards related to the landslide (Parise, Moscariello, 1997). 3.4 Geological hazards 4 MONITORING Several geological hazards are present at Slumgullion (Parise, Moscariello, 1997). The older, inactive part of the landslide is crossed by Colorado State Highway 149 which connects Lake City and Creede, Colorado. The inactive landslide is increasingly occupied at its lower reach by condominiums and other recreational structures near Lake San Cristobal. State Highway 149, in particular, is only 250 m downhill from the actively deforming area and is in possible jeopardy from a continuation of the current movement pattern of the active toe. To determine if loading exerted by the currently moving material causes deformation of the inactive deposits, a party led by David J. Varnes began surveying this part of the landslide in 1991 (Fig. 8). The surveys carried out to date indicate that the inactive part of the Slumgullion landslide is responding to the load caused by the advancing toe, and is not as stable as previously assumed Crandell and Varnes (1960, 1961) began the first investigation of the rates and history of movement of the Slumgullion landslide in 1958. Observations of changes in survey lines crossing the slide and in advances of the active toe were made several times between 1958 and 1973. Aerial photographs (1:12,000 scale) were obtained in 1985 by the Colorado Geological Survey and in 1990 and 2000 by the USGS (1:6,000 scale). A photogrammetric study of surface displacements performed on the active part of the landslide identified and measured over 300 natural points (trees and bushes) on the 1985 and 1990 photographs (Smith, 1993, 1996); displacements from 1985 to 1990 ranged from less than 0.25 m to more than 25 m. Figure 9 - Active part of the Slumgullion landslide with structural elements, contours of mean annual movement, GPS base stations, control points, and monitoring points. An alternative study was performed by using Geographic Information System to determine the movements of visually identifiable objects, such as trees and rocks, on the surface of the active landslide (Powers, Chiarle, 1996). More than 800 horizontal displacement vectors were measured directly from ortho-images that were produced by digital photogrammetric techniques using the same sets of aerial photographs. Overall, the results obtained are consistent with those derived from the standard photogrammetric study. The comprehensive study begun in 1990 by USGS included geophysical investigations, precise surveying and leveling, history of movement, displacement rates, landslide kinematics, material properties, and landslide dams. Since June 1993, GPS has also been used to measure movements on the landslide (Jackson et al. 1996). In 1993, the GPS method was tested with two objectives: to determine if satellite geodesy could be used to map the short-term spatial and temporal distribution of velocity on an active slide; and to determine possible very slow movements of the inactive part of the landslide. Beginning in 1998, GPS observations have been made at 11 control points and 19 monitoring points (MP1 to MP19; Figs. 9 and 10) installed around and on the active landslide, respectively. Control points were distributed in nonmoving areas around the periphery of the landslide, while the monitoring points were evenly distributed on the active landslide, but also placed strategically in relation to the main structural features of the landslide, and with a clear view to the sky for reliable acquisition of the GPS signal (Coe et al. 2003). Displacement data collected during the past 40 years by surveying, photogrammetric methods, and GPS methods reveal that surface velocities are, in general, inversely proportional to the width of the active earth flow (Fig. 9). Maximum surface velocities of 6-7 m/yr occur in the narrowest (150 m-wide) region of the flow, and surface velocities of 1 m/yr occur in the head and toe regions of the active flow (300- to 450-m wide). The rate of movement appears to be fairly constant from year to year and season to season, but generally corresponds with annual precipitation. In a 3.5-year monitoring period (from July 1998 to March 2002) twenty 1- to 3-day field campaigns of GPS surveys were performed, with a mean time between campaigns of 69 days (Coe et al. 2000a, 2003). This monitoring (also done with extensometers) showed that the Slumgullion landslide moved throughout the period, but that daily velocities varied on a seasonal basis (Fig. 11). Landslide velocity increased in response to snowmelt and rainfall and decreased during dry periods. Annual movements and average daily velocities were smallest at the head and toe of the landslide and largest in the central, narrowest part of the landslide (Figs. 12 and 13); movements and velocities deviated from this distribution in areas where they were affected by major structural elements within the landslide (also see Gomberg et al. 1995; Baum, Fleming, 1996). Figure 10 - Station consisting of radar corner reflector and GPS monitoring antenna, providing ground truth for INSAR measurements. Figure 11 - Diagram showing average daily velocities of GPS monitoring points. Dates shown are the mid-points between GPS observations. Figure 12 - Bar graph showing the ratio of minimum velocity to maximum velocity at each monitoring point. See Figure 9 for location of monitoring points. Figure 13 - Bar graph showing the annual horizontal movement of GPS monitoring points. See Figure 9 for location of monitoring points. Velocities were slowest in mid-winter when air and soil temperatures were coldest and precipitation was generally low and/or in the form of snow with a low water content. The seasonal variability in velocities has been interpreted (Coe et al. 2003) as due to ground-water levels and corresponding pore pressures that decrease when surface water is unavailable or cannot infiltrate frozen landslide material, and increase when surface water from melting snow or rainfall infiltrates unfrozen landslide material (Fig. 14). This hypothesis is supported by data showing seasonal changes in soil-water content and by a correspondence between velocity variability and air- and soil-temperature data from a meteorological station on the slide. In addition, patches of bouldery debris and fractures (created by continuous movement of the landslide) have been considered likely conduits through which surface water can infiltrate, regardless of the frozen or unfrozen state of the landslide matrix material. Such high permeability zones exist at Slumgullion as evidenced by sinks and perennial springs on the landslide (Guzzi, Parise, 1992; Fleming et al. 1999). Therefore, the availability of surface water is more important than landslide temperature in controlling the rate of landslide movement. This hypothesis is supported by field instrumentation data that show (1) landslide velocities coinciding with precipitation amounts regardless of the depth of freezing of landslide material, (2) spring and annual landslide velocities that were greatest when the depth of freezing was also the greatest, and (3) a rapid (several weeks or less) velocity and pore pressure response to rainfall. Figure 14 - Diagram showing landslide velocity, soil temperature, and pore pressure recorded at station IS2 and daily precipitation recorded at IS1 (after Coe et al. 2003). Pore pressure is shown as pressure head in meters of water above the piezometer, which is 2.0 m below the ground surface. In the summer and fall of 2001, five sets of INSAR data were collected by BYU over the active part of the landslide. These data are currently being processed and interpreted by BYU and the USGS. 5 CONCLUSIONS AND FUTURE PERSPECTIVES Nearly constant movement makes the Slumgullion landslide an excellent, large-scale natural laboratory. The active part of the Slumgullion landslide has been moving continuously for the past 300 years, transporting material hundreds of meters and repeatedly creating and destroying surface features. Mapping and interpretation of these features along the active margins and within the body of the active landslide show that the Slumgullion landslide can be partitioned into three separate regions based on the style of deformation. The main scarp and the head of the slide exhibit listric normal faulting in a region of extending flow. Deformation in the narrow, central part of the landslide is characterized by nearly vertical, strike-slip shear surfaces that bound discrete blocks; in this part of the slide, individual blocks move downslope as rigid bodies (region of plug flow). The toe of the slide is characterized by listric thrusts developed in a region of compressive flow. The Slumgullion landslide fits a Coulomb plastic model for landslide flow with extending, plug, and compressive flow in the upper, middle, and lower reaches respectively (Savage, Smith, 1986). The pattern of the overall movement on the landslide surface is consistent with the above model; it shows in fact that the highest surface-flow velocity, averaging 6-7 m/yr, occurs halfway down the active portion where the earth flow is confined to a width of 150 m. Flow velocities in the upper, wider part of the flow are estimated to be 2 m/yr. Flow velocities in the active toe (about 430-m wide) are estimated to be 1.3 m/yr. Surveys indicate that the inactive part of the Slumgullion landslide is responding to changing loads caused by the advancing toe and is not as stable as previously assumed. In general, ground is depressed in front of the advancing toe, the amount decreasing with distance from the toe. Current evaluations of hazards related to the landslide include potential damage to Colorado State Highway 149 and to residences in recreational areas in the vicinity of Lake San Cristobal. The continual but seasonally variable movement observed at Slumgullion closely fits the bathtub model for landslide movement proposed by Baum and Reid (2000). In this model, the presence of low permeable clays impede drainage of water from the landslide, keeping the shear zones perennially saturated and intermittently allowing pore pressures to increase sufficiently to destabilize the slide (Baum, Reid, 2000). Eventually, the most recently collected data (Coe et al. 2003) have shown (i) that movement was measured on landslide deposits near the head of the landslide that were previously identified as inactive, thus implying that the active landslide is larger and more complex than previously recognized; and (ii) that the driving force responsible for moving the upper part of the landslide may be less than it has been in the past, due to slower annual movements measured on the upper part of the landslide. REFERENCES Atwood, W.W., Mather, K.F. 1932. Physiography and Quaternary geology of the San Juan Mountains, Colorado. U.S. Geological Survey Professional Paper 166: 176 p. Burbank, W.S. 1947. Lake City area, Hinsdale County. In: Mineral Resources of Colorado, Colorado Mineral Resources Board: 439443. Baum, R.L., Fleming, R.W. 1991. Use of longitudinal strain in identifying driving and resisting elements of landslides. Geological Society of America Bulletin, 103: 1121-1132. Baum, R.L., Fleming, R.W. 1996. Kinematic studies of the Slumgullion landslide, Hinsdale County, Colorado. In Varnes, D.J., Savage, W.Z., eds., The Slumgullion earth flow: a large-scale natural laboratory. U.S. Geological Survey Bulletin 2130: 9-12. Baum, R.L., Reid, M.E. 2000. Ground water isolation by lowpermeability clays in landslide shear zones. Proceedings 8th International Symposium on Landslides, Cardiff 1: 139-144. Baum, R.L., Fleming, R.W., Johnson, A.M. 1993. Kinematics of the Aspen Grove landslide, Ephraim Canyon, central Utah. U.S. Geological Survey Bulletin 1842: F1-F34. Chleborad, A.F. 1993. Description, origin, and implications of a newly identified Slumgullion landslide deposit, San Juan Mountains, southwestern Colorado. USGS Open-File Rep. 93-548: 17 p. Coe, J.A., Godt, J.W., Ellis, W.L., Savage, W.Z., Savage, J.E., Powers, P.S., Varnes, D.J., Tachker, P. 2000a. Seasonal movement of the Slumgullion landslide as determined from GPS observations, July 1998 – July 1999. U.S. Geol. Survey Open-File Report 00-101: 47 p. Coe, J.A., Godt, J.W., Ellis, W.L., Savage, W.Z., Savage, J.E., Powers, P.S., Varnes, D.J., Tachker, P. 2000b. Preliminary interpretation of seasonal movement of the Slumgullion landslide as determined from GPS observations, July 1998 – July 1999. U.S. Geological Survey Open-File Report 00-102: 25 p. Coe, J.A., Ellis, W.L., Godt, J.W., Savage, W.Z., Savage, J.E., Michael, J.A., Kibler, J.D., Powers, P.S., Lidke, D.J., Debray, S. 2003. Seasonal movement of the Slumgullion landslide determined from Global Positioning System surveys and field instrumentation, July 1998 – March 2002. Engineering Geology, 68: 67-101. Corominas, J. 1995. Evidence of basal erosion and shearing as mechanism contributing to the development of lateral ridges in mudslides, flow-slides, and other flow-like gravitational movements. Engineering Geology, 39: 45-70. Crandell, D.R., Varnes, D.J. 1960. Slumgullion earthflow and earthslide near Lake City, Colorado [abs.]. Geological Society of America Bulletin, 71 (12-2): p. 1846. Crandell, D.R., Varnes, D.J. 1961. Movement of the Slumgullion earthflow near Lake City, Colorado. In: Short Papers in the Geologic and Hydrologic Sciences, U.S. Geological Survey Professional Paper 424-B: B136-B139. Cross, W. 1909. The Slumgullion mud flow. Science, 30 (760): 126127. Cruden, D.M., Varnes, D.J. 1996. Landslide types and processes. In Turner, A.K., Schuster, R.L., eds., Landslides. Investigation and Mitigation. Transportation Research Board, Special Report 247, National Academy Press, Washington, D.C.: 36-75. Diehl, S.F., Schuster, R.L. 1996. Preliminary geologic map and alteration mineralogy of the main scarp of the Slumgullion landslide. In Varnes, D.J., Savage, W.Z. eds., The Slumgullion earth flow: a large-scale natural laboratory. U.S. Geol. Survey Bulletin 2130: 13-19. Endlich, F.M. 1876. Report of F.M. Endlich. In: U.S. Geol. and Geograph. Survey (Hayden) of the Territories Ann. Rep. 1874: 203 p. Fleming, R.W., Johnson, A.M. 1989. Structures associated with strike-slip faults that bound landslide elements. Eng. Geology, 27: 39-114. Fleming, R.W., Baum, R.L. Savage, W.Z. 1996. The Slumgullion landslide, Hinsdale County, Colorado. In Thompson, R.A., Hudson, M.R., Pillmore, C.L., eds., Geologic Excursions to the Rocky Mountains and Beyond, Geological Society of America Special Publication 44, Denver, Colorado, October 28-31: 21 p. Fleming, R.W., Baum, R.L. Giardino, M. 1999. Map and description of the active part of the Slumgullion landslide, Hinsdale County, Colorado. U.S. Geological Survey Geologic Investigation Series I2672, scale 1:1,000. Gomberg, J.S., Bodin, P.W., Savage, W.Z., Jackson, M.E. 1995. Landslide faults and tectonic faults, analogs? The Slumgullion earthflow, Colorado. Geology, 23 (1): 41-44. Guzzi, R., Parise, M. 1992. Surface features and kinematics of the Slumgullion landslide near Lake City, Colorado. U.S. Geological Survey Open-File Report 92-252: 45 p. Howe, E. 1909. Landslides in the San Juan Mountains, Colorado. U.S. Geological Survey Professional Paper 67: 45 p. Keefer, D.K., Johnson, A.M. 1983. Earth flows: morphology, mobilization, and movement. U.S. Geological Survey Professional Paper 1264: 56 p. Larsen, E.E. 1913. Alunite in the San Cristobal quadrangle, Colorado. U.S. Geological Survey Bulletin 530-F: 179-183. Lipman, P.W. 1976. Geologic map of the Lake City caldera area, western San Juan Mountains, south-western Colorado. U.S. Geol. Survey Misc. Investigations Series Map I-962, scale 1:48,000. Madole, R.F. 1996. Preliminary chronology of the Slumgullion landslide, Hinsdale County, Colorado. In Varnes, D.J., Savage, W.Z., eds., The Slumgullion earth flow: a large-scale natural laboratory. U.S. Geological Survey Bulletin 2130: 5-7. Parise, M., Guzzi, R. 1992. Volume and shape of the active and inactive parts of the Slumgullion landslide, Hinsdale County, Colorado. U.S. Geological Survey Open-File Report 92-216: 29 p. Parise, M., Moscariello, A. 1997. Geologic hazards and relative slope stability of the inactive part of the Slumgullion landslide, southwestern Colorado. U.S. Geological Survey Open-File Report 97-706: 16 p. Parise, M., Moscariello, A., Fleming, R.W. 1997. Evidence from flank ridges for long-term diminishing movements of the Slumgullion landslide, Hinsdale County, Colorado. U.S. Geological Survey Open-File Report 97-517: 14 p. Powers, P.S., Chiarle, M. 1996. A digital photogrammetric method to measure horizontal surficial movements on the Slumgullion landslide, Hinsdale County, Colorado. In Varnes, D.J., Savage, W.Z., eds., The Slumgullion earth flow: a large-scale natural laboratory. U.S. Geological Survey Bulletin 2130: 51-55. Savage, W.Z., Smith, W.K. 1986. A model for the plastic flow of landslides. U.S. Geological Survey Professional Paper 138-S: 32 p. Schuster, R.L. 1985. Landslide dams in the Western United States. Proceedings IVth International Conference and Field Workshop on Landslides, Tokyo: 411-418. Schuster, R.L. 1996. Slumgullion landslide dam and its effects on the Lake Fork. In Varnes, D.J., Savage, W.Z., eds., The Slumgullion earth flow: a large-scale natural laboratory. U.S. Geological Survey Bulletin 2130: 35-41. Smith, R.L. 1993. Photogrammetric determination of movement on the Slumgullion slide, Hinsdale County, Colorado, 1985-1990. U.S. Geological Survey Open-File Report 93-597. Smith, R.L. 1996. Photogrammetric determination of slope movements on the Slumgullion landslide. In Varnes, D.J., Savage, W.Z., eds., The Slumgullion earth flow: a large-scale natural laboratory. U.S. Geological Survey Bulletin 2130: 57-60. Varnes, D.J., Savage, W.Z. eds. 1996. The Slumgullion earth flow: a large-scale natural laboratory. U.S. Geol. Survey Bulletin 2130: 95 p. Varnes, D.J., Smith, W.K., Savage, W.Z., Powers, P.S. 1996. Deformation and control surveys, Slumgullion landslide. In Varnes, D.J., Savage, W.Z., eds., The Slumgullion earth flow: a large-scale natural laboratory. U.S. Geological Survey Bulletin 2130: 43-49. Williams, R.A., Pratt, T.L. 1996. Detection of the base of Slumgullion landslide, Colorado, by seismic reflection and refraction methods. In Varnes, D.J., Savage, W.Z., eds., The Slumgullion earth flow: a large-scale natural laboratory. U.S. Geological Survey Bulletin 2130: 77-83.