When older dams are removed, what happens to the sediment trapped behind them?

By Linda Robinson

When Three Gorges Dam is built to curtail flow on the great Yangtze River in China, it is estimated that more than 26 million m³ of concrete will be used in the massive project. That's twice as much concrete as was used in Itaipu Dam in Brazil, currently the largest hydroelectric dam in the world. Three Gorges Dam itself is 1.5 mi. wide and more than 600 ft. high–and already under criticism.

The reservoir for the dam is several hundreds of feet deep and 400 mi. long and will inundate 395 mi.² of land. The Yangtze River has a history of massive floods that have claimed more than a million lives in the last century. The dam is one attempt at harnessing the 3,937 mi. of water to provide energy for China.

Smoke and dust rise after demolition efforts begin to make way for the Three Gorges Dam project.

Will the social and environmental benefits outweigh the possible damages? Some experts are concerned that health problems might surface when the reservoir is filled. At the present time, no attempts have been made to test for and remove accumulated toxic materials. Claims have been made that the dam will contribute to silt and sediment accumulation in the Yangtze River. So whether it is with awe or concern, the world will keep a keen eye on the Yangtze River and the Three Gorges Dam project.

Throughout history, humans have tried to harness rivers for flood control, irrigation, and power production. Some dams in the United States are currently being studied, and some are being removed after their projects have either failed or outlived their usefulness.

A survey in January sponsored by the National Hydropower Association found that, of 1,000 registered voters polled, 93% supported hydropower and believed it has an important place in America's energy future. However, the drawbacks to damming certain rivers have become increasingly apparent in the last 20 years or more. The two most damaging are the decline of anadromous fish populations and sediment buildup in the reservoirs behind the dam structures.

Protecting the Rivers, Fish, and Environment

University of Colorado Law Professor David H. Getches states in Water Law in a Nutshell that Columbia River harvests of salmon are now only about 8% of their size compared to 100 years ago. He concludes, "The primary reason for the destruction of anadromous fisheries has been construction of hydropower facilities on major rivers, which obstruct upstream spawning migration."

Laws require careful examination of fish and wildlife in an area when planning water projects. The Federal Power Act requires the Federal Energy Regulatory Commission (FERC) to find that a proposed project is "best adapted to a comprehensive plan" for water development for navigation, water power, "and for other beneficial public uses, including recreational purposes" (16 USCA Sec. 803[a]). This includes forethought for any negative consequences to fish populations.

In Idaho, FERC broke new ground in the legal arena last April by denying preliminary permits for two hydropower projects based on the consideration of potential for future environmental harm. The studies involved areas on the Snake River referred to as Star Falls and Eagle Rock.

"This unprecedented move by FERC illustrates how environmentally important these sites are," says Sara Denniston Eddie, director of hydropower and energy programs at Idaho Rivers United.

The Star Falls project was rejected based on a previous 1984 decision. The commissioners denied the application because "preservation of the natural scenic beauty, wildlife habitat and last undeveloped waterfall on this stretch of the Snake River Canyon in its historic condition is a far more valuable use of the resource than the proposed development of the site's potential for generating hydroelectric power." In denying the Eagle Rock project, the commission said completion of the project would "eliminate a Class I trout fishery and harm wildlife and riparian areas."

"Eagle Rock and Star Falls are Idaho treasures that should be preserved for recreation, fish, and wildlife," says Eddie. "The Snake River has been dammed, developed, and overworked for decades. FERC recognized that we must protect the last free-running stretch of the river."

Safely Removing the Old, Outdated, and Unused Dams

February 23, 1999, view of Savage Rapids Dam. River flow was 7,400 ft. 3/sec.

Robert Hamilton, P.E., regional activity manager at the Bureau of Reclamation Pacific Northwest Regional Office in Boise, ID, says one of the problems with most dams on America's rivers is age. It is estimated that a quarter of all US dams are more than 50 years old–the average life expectancy for a dam–and with 50 years of service comes 50 years of sediment build up in the reservoirs.

Hamilton points out that for reservoirs that have become laden with sediment, the problem is not only how to remove it but how to dispose of what is removed.

Bureau of Reclamation and US Army Corps of Engineers experts agree that there is no one-size-fits-all approach to dam removal. Each site is unique with its own set of problems in designing a removal project. Timothy Randle, hydraulic engineer with the Technical Services Center of the Bureau of Reclamation Denver Sedimentation and River Hydraulics Group, sums up the stream dynamics and reservoir sedimentation problems: "The size of the reservoir is relative to the annual flow of the stream. The width of the reservoir determines the velocity of the river, and with a wider reservoir there is more sediment deposited."

One example Randle offers is the Elwha River channel on the Olympic Peninsula in Washington. The channel is 100-200 ft. wide, with the reservoir on Lake Mills being 1,000-2,000 ft. wide–10 times wider than the channel. The annual volume a stream carries, Randle adds, is relative to the size of the reservoir.

"The reservoir volume can be compared to the annual streamflow volume," he says. "The larger the reservoir volume is–relative to the annual streamflow volume–the larger the amount of reservoir sedimentation."

At Arizona's and Utah's Lake Powell, for example, the reservoir is two times the annual flow of the river, but at Lake Mills the reservoir is only 4% of the average annual flow. And this, Randle says, has to do with the amount of sediment trapped by the reservoir. If storage is small relative to flow, then less sediment is trapped in the reservoir. If the reservoir is large relative to average flow, then all sediment will be trapped there. He adds that a key element for dam removal is the rate at which the dam is removed.

Savage Rapids Dam

Located on the Rogue River just 5 mi. upstream from Grants Pass in southwestern Oregon, Savage Rapids Dam was built in 1921 to divert river flow for irrigation. A combination gravity and multiple-arch concrete dam, it has a crest width of 464 ft. and a height of 39 ft.

During the nonirrigation season, the dam creates a backwater pool that extends 0.5 mi. upstream. Here a natural formation in the river has created a small riffle. During irrigation season it extends 2.5 mi. upstream. After irrigation season the stoplogs are removed and this section returns to a free-flowing river during the winter months.

The reservoir is fairly narrow–only two to three times wider than the river. The Savage Rapids River Dam Sediment Evaluation Study (included in the Josephine County Water Management Improvement Study [JCWMIS] authorized and funded by Congress in 1989) lists the annual mean flow for the Rogue River at 3,372 ft.³/sec. and the total drainage area as 2,459 mi.² Annual mean runoff is 19 in.; the highest recorded peak flow was in 1962 and measured 152,000 ft.³/sec. The dam has fish ladders in place, but they are old and don't meet current fisheries criteria. Dam removal has been proposed to restore fish passage to natural conditions. The JCWMIS recommends two pumping plants that would deliver water to the irrigation canals to replace the dam. The Grants Pass Irrigation District (GPID) asked that a sediment study be undertaken to model the potential sediment-related impacts of dam removal.

Things to consider for the study, Randle says, include how much sediment there is to be removed, the quality of the sediment, and the transport capacity of the river downstream. Significant concerns listed by the JCWMIS regarding the removal of Savage Rapids Dam include the particle-size gradation and spatial distribution of sediment accumulated within the reservoir, chemical composition of the reservoir sediment, and the rate at which the reservoir sediment would be eroded if the dam were removed. Other concerns the study will address include the rate at which the eroded reservoir sediment would be transported downstream and the location and magnitude of deposition downstream from the dam. Specifically there is apprehension regarding the potential for sediment deposition downstream at the proposed GPID irrigation pumping plants and at the water intake and treatment operations for the City of Grants Pass.

The primary objectives of the JCWMIS were to find a permanent solution to salmon and steelhead passage problems at Savage Rapids Dam and to help resolve conflicts over water use in Josephine County. The Bureau of Reclamation distributed the fishery portion of the report and a report on GPID water management in 1992. In 1994, the GPID Board voted to remove Savage Rapids Dam if capital and operational funding, water-availability guarantees, and protection from liability exposures were ensured.

A 1995 final environmental statement (FES) and a 1997 record of decision concentrated on salmon and steelhead passage at the dam and on the associated diversion facilities. The FES study concluded that fish passage and protective facilities at Savage Rapids Dam were inadequate, resulting in significant losses of both species, and recommended a preferred alternative that included removal of the existing dam.

After the FES completion, the Oregon legislature directed the establishment of a task force to review the recommendation of the planning report/FES. After reviewing documented examples of sediment damage to North American rivers when dams were demolished or breached, the task force recommended the dam be retained, based on sediment-related concerns.

The mid-Rogue is surrounded by mountains, with forest and timberland covering more than three-quarters of the river basin. The Rogue is designated a wild and scenic waterway from where it enters the Applegate River (west of Grants Pass) downstream to Lobster Creek Bridge, which is approximately 10 mi. upstream from the mouth of the river.

Irrigated farmland in southern Idaho is watered by the diversion of water from the Snake River via the Milner Dam.

Thirty percent of the total drainage area upstream of Savage Rapids Dam is regulated by Lost Creek Reservoir, which was built by the US Army Corps of Engineers primarily for flood control. A few other small reservoirs exist that might trap some sediment, but they are considered small relative to the Rogue River. The Lost Creek Reservoir is important in that it reduces flood peaks at Savage Rapids Dam by storing water during the high flood peaks and traps essentially all sediment transported into the reservoir by the river during these peak flows. Therefore, the study concluded "virtually no sediment from the uppermost Rogue River drainage gets past Lost Creek Dam."

Water storage behind a diversion dam is typically small, and the pools fill with sediment in the first few years of operation. After that, all sediment transported into the reservoir passes the dam. Sediment probably filled in Savage Rapids Reservoir to its storage capacity within the first few years, and the reservoir is now full. During periods of high flows on the Rogue River, almost all of the sediment is naturally transported downstream. River conditions that exist upstream from the Savage Rapids Park boat ramp cause high velocities relative to the reservoir velocities behind the dam. According to the Bureau of Reclamation report, these high velocities mean the dam does not cause sediment deposition in the upper 2 mi. of the reservoir during this period. The bureau confirmed this by sending drill crews and divers to the site. The conclusion is that any sediment deposition caused by Savage Rapids Dam is within the half-mile reach upstream of the dam to the park boat ramp.

Visual observations made by bureau personnel confirm that gravel-size sediment, along with the finer sediments, is being transported past the dam. Coarser sand and gravel, traveling as bedload, has been deposited in the half-mile area immediately upstream of the dam. This permanent deposition is probably the filling that occurred in the first few years after the dam was built.

The characteristics of the Rogue River provide a continual scouring effect on the river's pools. After several studies were completed and compared, it was determined that the current volume of reservoir sediment is estimated to be 200,000 yd.³ To put this in perspective, says Randle, if the same volume were placed on a football field, it would reach 100 ft. high. The volume is roughly two years of sediment load transported by the Rogue River and accounts for 70% of the river's transport capacity in the Grants Pass area, assuming that the remaining 30% is trapped upstream in the Lost Creek Reservoir.

The fairly steep gravel and cobble bed of the Rogue River has a series of pools, riffles, and rapids. Eight of the pools in the 12.5-mi. reach of the river between the reservoir and the conjunction with the Applegate River are 10-20 ft. deep. The other 10 pools are shallow at less than 10 ft. deep.

During the period with low flow, the velocity slows down and the pools fill slowly with sediment. During high flows, such as spring runoff or during the winter storms, however, the speed within the pools increases and the sediment is scoured from the pools and transported downstream with the river. From 1996 to 1997, a US Geological Survey gauge cross-section near Grants Pass evidenced this occurrence during a winter storm that virtually scoured out a 6-ft. depth and then refilled the channel bed the following year during the low-flow period.

To arrive at the sedimentation estimates, experts look first for older studies, including topographical maps. In the case of Savage Rapids there were no such maps available. "If there is an old topographical map to refer to, then a survey can be done and the two can be compared," Randle explains. "If no pre-dam survey exists, then we have to use other methods and means to measure the river bottom. Then we put that in context. With these items in hand, it's possible to design a removal program that is slow enough not to impact the river.

"We're in the second phase of the study," he says. "Before dam removal, they have to first build pumping plants. The study now is where they should be built. But before anything else, the pumping plants for irrigation would go in first."

 Elwha and Glines Canyon Dams

Elwha and Glines Canyon Dams are located on the Elwha River, which flows on the Olympic Peninsula in northwestern Washington. The federal government acquired both dams nearly three years ago, with the goal of removing them to restore the native anadromous fisheries in the Elwha River. Neither dam has fish passage facilities.

Completed in 1913, Elwha Dam is a 108-ft.-high concrete gravity dam with gated spillways on both abutments. A powerhouse contains four generating units with a combined capacity of 14.8 MW. The dam impounds Lake Aldwell, which has a surface area of 267 ac. and a storage capacity of 8,100 ac.-ft. at 197-ft. elevation.

Glines Canyon Dam, completed in 1927, is a 210-ft.-high, single-arch concrete structure with varying width, with a thrust block on the right abutment and a gated spillway on the left. A powerhouse with one generator has a 13.3-MW capacity. The dam impounds Lake Mills, which has a surface area of 415 ac. and a storage capacity of 40,500 ac.-ft. at 590-ft. elevation.

The plan calls for allowing river flows through diversion channels and notches during the actual dam-removal process and managing sediment through controlled releases.

Proposed Removal Methods

Salmon Falls Dam is primarily a diversion dam for farmland irrigation.

Area of Star Falls in southern Idaho was recently rejected by FERC for a hydropower project based on protection of the environment and wildlife.

Wildlife is abundant in the Auger Falls area on the Snake River, which is currently protected from installation of new hydropower dams.

Elwha Dam. The concrete gravity section of Elwha Dam would be removed first by lowering the reservoir behind the dam by about 50 ft. through construction of a diversion channel at the left-abutment spillway. This drawdown would also allow for a major portion of the upstream fill materials to be excavated. With the removal of the remaining rock fill within the original river channel in specific increments, the reservoir would be lowered another 40 ft. At completion, removed structures would either be hauled off or buried at the dam site. In addition, the temporary diversion channel would be filled in and contoured to provide a natural look.

Tim Randle explains the plan for Elwha Dam. "A tentative plan would be to lower the water-table reservoir level in the 7.5-foot lifts in low-flow periods but not during certain ‘fish window' periods, which are about two months at a time."

For Elwha, fish windows fall during November to the end of December, and Randle says the river flows are too high during this time for dam removal activities anyway. Fish windows also occur during all of May and June and then again from August through September 15.

The reservoir at Elwha Dam is much wider than Savage Rapids Dam, and rather than scouring itself every two years or so, it has had about 60-75 years to store sediment. Such a large volume of sediment has posed huge challenges to removal.

"In terms of the width of the reservoir, it's harder to flush out a wider reservoir because of the sheer volume that you are dealing with," Randle says. "There is a limit as to how wide the river will erode a channel across the reservoir sediments. For example, at Savage Rapids, the reservoir is narrow and small and completely filled with sediment in its first two years of operation. But at Lake Mills on the Elwha River, the reservoir is much wider and larger–trapping 75 years' worth of sediment that could potentially erode. So the questions become: How will it be removed? Will the sediment be eroded by the river? Should all of the dam be removed, or is there a possibility that some structures could be left in place for historical purposes?"

Robert Hamilton adds still yet another angle to the challenge of the Elwha Dam removal. "A more intriguing question is what are the long-term impacts?" he says. "In the case of Elwha Dam, the bottom of the river will be restored to its pre-dam elevation–an estimated 3 to 5 feet higher than it currently is. This will change the floodplain dynamics."

Glines Canyon Dam. Because of flooding concerns, the task force proposed to retain Glines Canyon Dam until the surface diversion channels at Elwha Dam are completed. At construction time, the concrete-arch part of Glines Canyon Dam will be removed in 7.5-ft. layers by a combination of blasting and diamond sawcutting. The 7.5-ft. layers will come out beginning at elevation 590.33 ft. down to the final level of the streambed at 400 ft. The Bureau of Reclamation report also describes that the currently existing power penstock would be used for streamflow diversion for reservoir levels above 530 ft. In addition, a series of notches excavated to a 15-ft. depth alternating from left to right sides of the arch at 7.5-ft. intervals would be constructed. These notches would go in between elevations of 522.83 and 410.33 ft. Structures the task force chose to retain include the concrete thrust block and the gravity wall. The embankment dike that's on the right abutment will be updated and become a public overlook. For historical value the gated spillway, penstock, gatehouse, and powerhouse structures on the left side will remain intact.

Using the US Bureau of Reclamation Composite Index for 1995, it was estimated that physical removal of the two dams would cost approximately $20.2 million. Using the Index for 2002, the cost goes up to $23.5 million for dam removal alone. However, experts add that significant upstream and downstream impacts of the dam removals could double or even triple that figure by completion of the project.

Tentative schedules for beginning the actual physical removal of Elwha Dam are set for November 2005, and it will take an estimated two and a half years to complete. Hamilton concludes that from that time it would take another three years for the sediment to be eroded out.

Guest author Linda Robinson is a journalist specializing in agriculture and land-use planning.

EC - September/October 2002