FLOOD06904 - Laserfiche WebLink

Home Browse Search

62 LAWN LAKE DAM AND CASCADE LAKE DAM FAILURES, COLORADO Park, the flood peak was 2.1 times the 50o-yr flood. Geomorphic and sedimentologic evidence suggest this was probably the largest flood in these basins since at least the retreat of the glaciers about 10,000 years ago. Traveltimes of the flood wave were determined from eyewitness accounts. The leading flood wave took 3;28 hours to travel 12.5 mi (average 3.8 milh). This slow traveltime was attributed to the extreme valley rough- ness and damming of the flow by large amounts of debris in the Roaring River, and to the long, flat, glacial lake valley of Horseshoe Park, which attenuated the flood. Because of the steep channel, debris-laden flows, and unusual and poorly understood characteristics of un- steady flash floods, the theory and application of con- ventional energy and flow-resistance concepts probably are not applicable for these conditions. When hydraulic analyses are made on these fl9ods, the effects of un- steady flow, Manning's n-values, high sediment concen- trations, and scour and fill that affect the cross-sectional flow area need to be analyzed. To varying degrees these factors may influence hydraulic computations, par- ticularly indirectly determined peak discharges. Geomorphic effects of the dam-failure flood were cata- strophic. Channels were widened tens of feet and scoured from 5 to 50 ft locally. In the Roaring River valley, alter- nate river reaches were either scoured or filled, depen- ding on valley slope. Generally, reaches steeper than 7 percent were scoured, and reaches less than 7 percent were filled. In the Roaring River, 56 percent of the chan- nel was scoured as much as 50 ft, and 44 percent was filled with coarse sediments 2 to 8 ft thick. Antidune backset beds were preserved in a sand splay 700 ft downstream from the dam; horizontal stratified coarse sand at the flow margins all along the Fall River indicated upper-flow regime conditions. An alluvial fan of 42.3 acres, containing 364,600 yd' of material, was deposited at the mouth of the Roaring River. The fan had a maximum thickness of 44 ft and an average thickness of 5.3 ft. The largest boulder known to have moved in the flood, I4X17.5X21 ft and weighing an estimated 452 tons, was located on the alluvial fan. Areas of sediment deposition shifted from the east to the west side of the fan during the flood. Down the flow axis, average particle size changed from 7.5-ft boulders to fine sand and silt in a distance of 1,900 ft. The alluvial fan dammed the Fall River, forming a lake of 17 acres upstream from the fan. On the west side of the Roaring River alluvial fan, below Cascade Lake dam and upstream from the Aspenglen Campground, boulder berms similar to those described from other catastrophic floods in steep moun- tain channels were well-developed. They are believed to have originated from non-cohesive, turbulent, coarse sediment-gravity flows. Upstream from Aspenglen Campground, floodwaters formed large-scale concentra- tions of boulders across a part of the channel, forming a step-pool profile. Downstream on the Fall River, chan- nel scour and sediment deposition were much less significant than downstream from Cascade Lake dam. The U.S. Bureau of Reclamation spent $80,000 for debris removal in Lake Estes following the flood, but estimated at least 10 times more sediment was deposited in the lake during high spring runoff in 1983. The source of the sediment was primarily from erosion of exposed river banks. The channels of the Roaring River, Fall River, and Big Thompson River are not yet stabilized and will continue to transport large amounts of sediment into Lake Estes for some time in the future. A dam-break flood model was used to evaluate the model's performance on high-gradient streams, to pro- vide supplemental hydrologic information, particular- ly peak flows for these dam failures, and to evaluate various scenarios of dam.breach development and the probable impact of the failure of Cascade Lake dam. Satisfactory results were obtained, but not without significant difficulties in getting the model to run prop- erly. To calibrate the model, Manning n-values between 0.1 and 0.2, or an average of 78 percent greater than field-selected values, were required, and subcritical flow was assumed and substantiated. Locally, very short reaches of supercritical flow occurred. Model results would have been significantly different without calibration. Results of the calibration phase indicated that the model had the potential to simulate dam-break floods in high-gradient stream channels. Peak discharges from dam-break modeling reflected water-only discharges; total discharge may have been considerably higher on the Roaring River from sediment and debris. The range of difference of observed and modeled peak discharges varied from -3,200 ft'ls to 600 ft'/s. At worst, the model underpredicted peak discharge by 27 percent 3.61 mi downstream from Cascade Lake dam. Consider- ing the dynamics of the breach development and errors in the indirect measurements of peak discharge, the results were quite reasonable. The range of difference of the observed and modeled maximum flood depth was -1.3 to 2.6 ft; it averaged 1.0 ft. The range of difference of the observed and modeled leading edge of traveltime was -0.4 and 0.05 hour. Comparisons were made for hypothetical breach widths of 25 ft and 200 ft, which were compared with model results of the actual breach width of 55 ft. or a breach width of 25 ft, the peak discharge would have been 7,000 ft'ls less downstream from Lawn Lake dam to 1,300 ft'ls less at river mile 12.5. Maximum flood depths averaged 0.6 ft lower, and the flood wave would