Case Study: Roaring Fork sinkhole
2005-01-17 | Dr. John Hopkins
[ED: This report from January 17, 2005 was written by Jon White, (Senior Engineering Geologist, Emeritus). Lightly edited for dated references it highlights a hazardous geological regime in the central Colorado Rockies around the Roaring Fork River Corridor.]
Last week, while on the Western Slope, I was informed by a number of people that a sinkhole had opened at an on-going development in the Roaring Fork Valley, somewhere between Glenwood Springs and Carbondale. I made some inquiries with the local geotechnical firms and found that the sinkhole had opened Sunday, January 9, 2005 on the Ironbridge Development off of County Road 109, across the river from Highway 82, at NE ¼, SW ¼, Section 1, T7S, R88W. I visited the site with the Ironbridge Construction Manager on Thursday, January 13, 2005. There was some damage to golf course structures and loss of equipment. There were no injuries or residential loss. This development was previously called Rose Ranch during the Garfield County planning approval process and I reviewed the main submittal and several resubmittals as the county reviewed the application. My main concerns were hydrocompactive soils, debris flow hazards and mitigation, and potential for evaporite karst and formation of sinkholes on the property.
The location is centered in the Cattle Creek Anticline in the Roaring Fork River Valley near the northwest limit of a swarm of sinkholes that extends to Carbondale. The site is about equidistant from Glenwood Springs and Carbondale. Earlier geotechnical investigations of the Rose Ranch Development revealed several existing sinkholes on the property. The site is located right of center in the block diagram (where the arrow points to the large alluvial/debris fan) of the CGS publication MS-34 Collapsible Soils and Evaporite Karst Hazards Map of the Roaring Fork River Corridor. The underlying bedrock is Eagle Valley Evaporite: on the same property, about 1 mile south, an oil well (Shannon Oil Co. Rose no. 1) was drilled that encountered 60 feet (18 m) of gravel, and 2,065 (630 m) feet of gypsum, anhydrite, and siltstone before drilling through 935 feet (280 m) of halite where the hole was terminated.
The sinkhole opened up at the clubhouse golf cart maintenance and storage facility. Apparently a small hole, about 10×10 ft (3×3 m) opened very early Sunday morning that quickly enlarge to a 42-foot (12.8 m) diameter and 40-foot (12.2 m) deep sinkhole by the middle of the day. The opening undermined the southwest corner of a vinyl-tent Quonset where electric golf carts were being stored for the season. Apparently at least two of the small vehicles were lost into the sinkhole as the workers were emptying the structure. Exposed in the hole were a ruptured 24-inch (0.61 m) black plastic corrugated storm-sewer pipe and two smaller PVC pipes of unknown utility. At the time of my inspection, there was water at the bottom of the hole and the walls of the sinkhole were near vertical with the edges very unstable. Rocks were continuously loosening on the walls and falling.
Exposed in the sinkhole were roughly-stratified alluvial fan soils with certain strata very gravelly, while others were predominantly fines. This would be expected since the location is about 500 feet (152 m) below the mouth of a drainage basin that outlets onto the alluvial fan that the development is located on. The bottom third of the hole appeared to be finer material, mostly sand, silt, with some clay. Underlying Eagle Valley Evaporite was not exposed in the sinkhole, nor was mid-Pleistocene Roaring Fork glacio-fluvial terrace gravels.
There was some discussion that a possible leak in the storm sewer could have contributed to this ground movement. Reportedly, water was weeping from the gravel pack around the storm sewer pipe shortly after failure. I think that is unlikely as a sole cause, though. The soils exposed on the sinkhole wall did not appear appreciably wet and these soils have significant hydrocompaction susceptibilities. A leak into the more permeable gravel layers would have quickly spread water laterally and resulted in broad areas of settlement and a general down-warping of the pavement at the site. That was not the case; there was no report of earlier settlement and no evidence in the pavement of distress related to ground settlement surrounding the sinkhole. I believe that a void had propagated upward into the alluvial fan soils in the recent geologic past from chimneying of material down into a large evaporite bedrock void below. The bridging material, or ceiling of the open subsurface void (which had to be at close to the volume equivalence of 40 by 40 by 40 feet (12x12x12 m)) quickly collapsed to the surface. Apparently the void was of such size that the displacement of air when the major part of the sinkhole dropped cause a pneumatic “geyser” of a cloud of finer soil particle that covered the ground along the edges. While leakage was probably not a major factor, it is likely that the overall grading and trenching for the storm sewer at this location weakened the void ceiling and accelerated its propagation to the surface.
For further reading on geo-hazards in the Roaring Fork River Corridor, see MS-34 Collapsible Soils and Evaporite Karst Hazards of the Roaring Fork River Corridor, Garfield, Eagle, and Pitkin Counties, Colorado.