CHAPTER 6

SLUMGULLION LANDSLIDE DAM AND ITS EFFECTS ON THE LAKE FORK

by Robert L. Schuster

Introduction

The Slumgullion earth flow (fig. 1) dammed the Lake Fork of the Gunnison River approximately 700 years ago, impounding Lake San Cristobal, Colorado's second largest natural lake (fig. 2). The original impoundment was 4.3 km long and had a surface area of 1.8 km². During the past 700 years, sediment entering the lake from the headwaters of the Lake Fork of the Gunnison River and from Slumgullion Creek have formed large deltas (figs. 3 and 4) in the lake; the Lake Fork delta at the head of the lake has reduced the length of the lake to 3.3 km and the two deltas together have reduced lake surface area to 1.34 km² . The maximum depth of the lake is 27 m and its estimated volume is 14 million m³.

Landslide dams are common natural phenomena (Schuster and Costa, 1986; Costa and Schuster, 1988). However, unlike the Slumgullion blockage, most landslide dams fail fairly soon after formation by overtopping and surface erosion or by piping (internal erosion). Some landslide dams become long-lasting geologic features (Schuster and Costa, 1986; Costa and Schuster, 1988). However, a few landslide dams have failed after many years of stability. For example, in 1966 the rock- and debris-fall dam that formed Lake Yashinkul on the Tegermach River in the Kirghiz Republic, U.S.S.R., failed by piping after having been stable for 131 years (Pushkarenko and Nikitin, 1988). This poses the question: Even though the Slumgullion landslide dam is at least 700 years old, is there a possibility of it failing?

Geometry of the Dam and Its Natural Spillway

In damming the Lake Fork of the Gunnison River, the landslide buried the original channel of the Lake Fork for a straight-line length of about 2.8 km (fig. 5), of which about 1.8 km (line AB) currently is buried under the subaerial part of the landslide and about 1.0 km (line BC) lies under the part of the landslide that has been submerged by Lake San Cristobal. The average gradient of this pre-landslide stretch of the Lake Fork was 0.9 percent (based on a straight-line projection of the stream as shown by line ABC in fig. 6). Today's natural lake outlet channel around the toe of the landslide (fig. 5) has a "straight-line" length of about 1.6 km and a true length (including bends and curves) of 1.9 km. Because of the height of the dam, the straight-line and true gradients for this stream channel are much steeper than the original stream gradient; the gradient for the straight-line projection is about 3.5 percent; for the actual stream length, it is 2.7 percent. Both are on the order of three times as steep as the pre-landslide stream gradient. If it is assumed that the original stream had roughly the same deviation from a straight line as today's outlet channel has, its buried length can be estimated at 3.3 km, of which 2.1 km is under the exposed part of the landslide and 1.2 km lies buried beneath the submerged part of the slide. The original average gradient for this stretch was then about 0.8 percent, which compares closely with today's average gradient between Lake City and the upper end of Lake San Cristobal of 1.06 percent.

As shown in figure 6, the crest of the dam is about 70 m high directly above the inferred location of the old stream bed. At this point, the original stream bed lay about 200 m east of today's outlet from Lake San Cristobal (fig. 5). However, this was not the low point of the crest of the dam, which lay against the west valley wall near the location of today's Lake Fork. By extending the present-day topography near the dam crest downslope (i.e., westward) from the section in figure 6 to the toe of the slide, I find that the minimum elevation of the crest of the dam was about 2,750 m and; thus, the height of the dam at that point before spillway erosion occurred was about 50 m above the original stream elevation.

Changes in Geometry of the Dam

In referring to the toe of the landslide, Rickard (1903) noted that, "It is said, by those living on the lake shore, to be still in motion and to be extending further across the valley." With the exception of this vague statement, there has been no evidence of historic movement of the toe of the landslide. The main change in geometry of the dam since it was formed has been deepening of the lake outlet channel caused by stream erosion. Most of this erosion has been in the toe of the landslide or in colluvium or rock-fall material from the west valley wall. Four surface projections made at right angles to line AB have shown that the average amount of downward erosion in the outlet channel since overtopping has been about 10 m and that the total amount of landslide material removed by channel erosion has been about 1 million m³. At Argenta Falls (fig. 7) the stream has eroded through the landslide material and colluvium into more resistant Tertiary volcanic bedrock (Lipman, 1976), forming a falls that is approximately 25 m high.

Another change in the dam portion of the landslide is the large delta on the southeastern edge of the dam. This delta, which is continuing to grow into Lake San Cristobal at the mouth of Slumgullion Creek (figs. 4 and 5), is a fan composed of material eroded from the surface of the landslide and washed into the lake by Slumgullion Creek. Most of the delta probably formed soon after the landslide occurred and before vegetation obtained a hold on the landslide. Comparison of the 1992 delta in figure 4 with a similar photo taken by Larsen in 1910 shows little difference in size or character of the delta during that 82-yr period. However, considerable small-scale deposition did occur on the delta surface in 1985 both onshore and in the lake as a result of debris flows, triggered by meltwater from an unusually heavy snowpack that flowed down Slumgullion Creek. This delta-building activity will probably continue spasmodically during periods of extraordinarily heavy precipitation, causing unwanted deposition both in the lake and on the shore; however, it will have no detrimental effect on the stability of the dam.

Stability of the Blockage

The dam has a soil matrix (mainly clay) derived from hydrothermally altered volcanic rocks. This material has low porosity; thus, it forms a relatively water-tight dam, and, apparently, there never has been much chance of a piping failure. However, most such landslide dams have failed by surface erosion after overtopping. This one did not because the low point of the crest formed at the contact between the toe of the earth flow and the opposite bedrock and coarse-colluvium valley wall. When the rising lake overtopped the crest at its low point, it incised an average of about 10 m into the toe of the slide and the underlying colluvium/bedrock, forming an erosion-resistant stream channel across the crest about 200 m west of and 28 m higher than the pre-landslide stream. Where volcanic bedrock (Lipman, 1976) is exposed at Argenta Falls (fig. 7), the stream channel is essentially secure against erosion. The remainder of the channel has become armored by rock fragments from the colluvium that remained after removal of finer particles from the landslide material and colluvium by erosion. This spillway channel and the remaining dam have survived for 700 years; thus, they should remain stable as the lake continues to fill with sediment from the Lake Fork and Slumgullion Creek. Lake level also will remain stable at its present elevation of 2,742 m, which is maintained by a concrete lip at the head of the spillway.

Conclusions

About 700 yrs ago, the Slumgullion earth flow dammed the Lake Fork of the Gunnison River impounding Lake San Cristobal, which today is 3.3 km long, 27 m deep, and has a volume of about 14 million m³. Except for an average of 10 m of erosional downcutting of the natural outlet channel across the dam, little change has occurred in the geometry or character of the dam since it was formed. 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.

The Slumgullion landslide dam has been considered as a possible location for construction of an embankment dam across the valley near the outlet from Lake San Cristobal (Crandell, 1958). However, such a dam does not appear to be practical because of the hazard posed by the active toe of the earth flow 1.5 km upslope and possible reactivation of the toe of the main landslide due to increase in pore pressures resulting from a higher lake level behind a man-made dam.

The two large deltas that enter Lake San Cristobal from the Lake Fork and Slumgullion Creek will not affect the stability of the natural dam. However, the Slumgullion Creek delta will continue to hinder development along the shore of the landslide dam at the northeast corner of the lake, and the two deltas will continue to slowly fill the lake basin. If rates of sedimentation from the two streams continue at the same average rate as for the past 700 yr, it can be postulated that the lake will be filled with sediment in about another 2,500 yr.

References Cited

Costa, J.E., and Schuster, R.L., 1988, The formation and failure of natural dams: Geological Society of America Bulletin, v. 100, p. 1054-1068.

Crandell, D.R., 1958, The Slumgullion mudflow and its suitability as a dam site: U.S. Geological Survey Administrative Report, October, 8 p.

Lipman, P.W., 1976, Geologic map of the Lake City caldera area, western San Juan Mountains, southwestern Colorado: U.S. Geological Survey Miscellaneous Investigations Series Map I-962, scale 1:48,000.

Pushkarenko, V.P., and Nikitin, A.M., 1988, Experience in the regional investigation of the state of mountain lake dams in central Asia and the character of breach mudflow formation, in Kozlovskii, E.A., ed., Landslides and Mudflows: Moscow, UNESCO/UNEP, v. 2, p. 1359-1362.

Rickard, T.A., 1903, Across the San Juan Mountains: Engineering and Mining Journal, v. 76, p. 346.

Schuster, R.L., and Costa, J.E., 1986, A perspective on landslide dams, in Schuster, R.L., ed., Landslide Dams: Processes, Risk, and Mitigation: American Society of Civil Engineers Geotechnical Special Publication No. 3, p. 1-20.