Applying Models to Determine Sediment Transport in Rivers

Sediment transport numerical models can aid in the planning, operation, or decommissioning of a hydro project. Case studies illustrate how using these models can help owners predict where sediment will be deposited, evaluate measures to augment sediment supply downstream from a dam, and forecast how sediment released during dam removal will affect the downstream environment.

By Yantao Cui, Bruce K. Orr, Frank K. Ligon, and David W. Heintzman

Sediment transport in a river is sporadic, depending primarily on periods of higher flow. During lower flow periods, sediment particles settle and form deposits. These sediment deposits, along with such features as bedrock outcrops, form the basic morphologic elements of a river that support a dynamic biological and ecological system.

Human activities in a river – such as water diversion and dam construction – interrupt water flow and sediment transport. As a result, the river adjusts its channel morphology over many years, even after the natural flow and sediment regime are restored. Morphologic changes after construction of dams in mountain rivers include:

– Sediment deposition and channel aggradation upstream of the dam;

– Channel degradation downstream of the dam;

– Coarsening of surface sediment particles and reduced sediment mobility downstream of the dam; and

– Reduced patch size and thickness of gravel deposits in bedrock reaches.

General scientific understanding and past experiences provide a qualitative prediction of how a river may react to development of a hydroelectric project. However, due to the large temporal and spatial scales involved, sediment transport numerical modeling is probably the only tool that allows for a quantitative evaluation.

A sediment transport numerical model generally applies the following principles:

– A combination of mass conservation and energy or momentum conservation for the flow. This will provide physical parameters such as water depth, flow velocity, and shear stress exerted upon the channel bed;

– A sediment transport relation that links the rate of sediment transport with physical parameters of the flow (usually the shear stress); and

– Mass conservation for sediments that tracks bed aggradation or degradation due to the imbalance between the influx and outflux of sediment.

Traditional sediment transport models can be fairly simple in applying the above principles by considering the sediment as homogenous. More complex sediment transport models, such as those discussed in this article, attempt to apply the principle of sediment mass conservation on a grain-size-specific basis. Some models even track the attrition of sediment particles due to collision (i.e., the mechanical breakdown of particles due to abrasion and collision impacts during sediment transport), which can be especially important when simulating long river reaches.

Despite rapid advancements in recent years, there are some inherent limitations for sediment transport numerical modeling. For example, one-dimensional sediment transport numerical models are unable to simulate the detailed channel morphology, such as the formation of pool-riffle morphology. Two- and three-dimensional sediment transport numerical models are still relatively primitive and cannot be applied for large-scale simulations.

In addition, all numerical models need to be examined and/or calibrated before application. In general, a minimum model calibration requirement is that the sediment transport model reproduces the current channel condition (mostly the longitudinal profile, and grain size distribution in certain cases) with the current hydrologic conditions and the best understanding of sediment supply to the river. In certain cases, such as the simulation of fine sediment transport in a bedrock or a gravel-bedded river, even this minimum model calibration cannot be conducted because the question to be examined by the model differs from the processes that resulted in the current channel condition. In this case, our confidence in a particular numerical model often is based on past model performances under similar conditions.

Whether a model is calibrated or not, the results of a numerical model should not be interpreted as the final answer. Instead, professional judgment must be applied to carefully interpret the results.

Figure 1: Results from coring samples show that sediment in the reservoir behind Marmot Dam consists of a coarse gravel deposit at the top and a sandy layer beneath. The bottom layer is sediment in the river before the dam was built. The simulated core, developed using The Unified Gravel-Sand (TUGS) numerical model, closely matches actual conditions. Click here to enlarge image

Sediment transport models can be applied in all stages of a hydro project – such as during planning, operation, and decommissioning. They can be used to answer questions pertaining to many aspects of a hydro project, such as future performance due to sedimentation, estimation of project life, and effects to upstream and downstream geomorphic and ecologic systems. We demonstrate the application of sediment transport models in hydro projects through several case studies.

Case studies

Several case studies of the application of sediment transport models illustrate their use at hydro projects in various stages – planning, operation, or decommissioning.

Predicting sedimentation near an intake

The Sacramento Municipal Utility District (SMUD) in California is designing a 400-MW pumped-storage facility as part of its Iowa Hill Development Project. This project includes construction of an upper storage reservoir. The project will operate by pumping water from the existing Slab Creek Reservoir to the upper storage reservoir during off-peak hours, then discharging water from this reservoir to Slab Creek Reservoir for power generation during peak hours. SMUD was concerned that the delta front of the sediment deposited in Slab Creek Reservoir would migrate down to the vicinity of the intake for the pumped-storage facility during the lifespan of the project. SMUD intended to apply a sediment transport numerical model to answer this question.

SMUD hired Stillwater Sciences to perform this assessment. However, upon examining available field data, Stillwater Sciences personnel determined the question could be answered simply by conducting a mass conservation analysis to project the future growth of the sediment delta. Based on this analysis, the sediment delta will not reach the vicinity of the intake within the project life of 100 years under any of the possible project operation scenarios, as long as future upstream sediment supply remains the same.1

This case study illustrates the importance of combining a good understanding of sediment transport theories and practical applications with basic relevant information available for the project in question before rushing in to apply sediment transport numerical models.

Studying gravel augmentation

The McKenzie River in Oregon is the location of Trail Bridge Dam, which impounds water for a 10-MW hydro project. In 2005, regulating agencies asked the Eugene Water and Electric Board to consider the feasibility of gravel augmentation downstream of the project to mitigate the reduced sediment supply in the river resulting from operation of the dam. To determine if this augmentation would be feasible, the board hired Stillwater Sciences to perform an assessment.

This is another instance where a sophisticated sediment transport numerical model is not needed to arrive at an accurate answer. Instead, Stillwater Sciences used its EASI (Enhanced Acronym Series with Interface) model, which combines Parker’s surface-based bedload equation with simple hydraulic calculation and hydrologic integration to predict an order-of-magnitude accuracy bedload transport rate. This model has been used to evaluate potential gravel augmentation associated with hydro projects in many rivers, including Lewis River in Washington and American River in California. The application of EASI or similar simple models requires minimal field data, such as channel cross section, channel gradient, surface grain size distribution, and hydrologic record. In addition, it can be carried out quickly (i.e., with only a few hours of effort to run the model after data acquisition).

EASI simulation in the McKenzie River indicated that the sediment transport capacity immediately downstream of Trail Bridge Dam is extremely large. Thus, any practical amount of gravel augmentation is unlikely to create additional spawning gravel deposits.

Decommissioning Marmot Dam

The 22-MW Bull Run hydro project includes the 47-foot-tall Marmot Dam on the Sandy River in Oregon. When it came time to pursue a new Federal Energy Regulatory Commission (FERC) operating license for this project, Portland General Electric (PGE) elected to decommission the project. Marmot Dam is the only structure on the Sandy River affecting upstream migration for salmon and steelhead. Adult salmonid upstream passage is possible through a fish ladder at the dam. Removing the dam after project decommissioning will allow anadromous salmonids to volitionally use all of their historical habitat.

To better understand the grain size distribution, volume, and possible chemical contamination of the sediment deposited upstream of Marmot Dam, in 1999 PGE commissioned a coring exercise by Squier Associates. Results of this exercise indicated there were about 1 million cubic yards of sediment accumulated in the reservoir.2 The coring samples showed that sediment is deposited in two large units: a coarse unit about 15 feet thick composed primarily of gravel and some sand, and a unit of varying depth beneath the coarse sediment unit that is composed primarily of sand with some gravel. (See Figure 1 on page 44.) Because of the large volume of this sediment deposit, PGE and others involved in the decommissioning process were concerned that its release after dam removal would result in devastating ecological consequences to the 30-mile river reach downstream of the dam. This reach serves as the most important spawning and rearing habitat for salmon and steelhead in the Sandy River.

Figure 2: Numerical simulation indicated that the sediment behind Soda Springs Dam would be transported out of the impoundment and downstream quickly after dam removal. Channel aggradation would happen quickly (within about two weeks) in a 9-mile reach of the river. Click here to enlarge image

PGE considered mechanical dredging and removal to deal with most of the sediment, but this option was expensive. In addition, it was associated with other environmental and practical problems, such as traffic congestion and excessive noise levels along the hauling route.

To select a dam removal method that would be acceptable to all project stakeholders, PGE needed to better understand the potential effect of sediment release. PGE hired Stillwater Sciences to develop and apply a one-dimensional sediment transport numerical model that simulated several dam removal scenarios. These scenarios included: a one-season removal of the dam with minimal sediment dredging (only enough to allow for removal of the dam), stepped removal in two seasons with dredging of fine sediment in the second season, and one-season dam removal with partial dredging before the removal.3,4 Prior to simulation of dam removal scenarios, the model was examined under the current background conditions and was found to closely reproduce the current river longitudinal profile with the recorded hydrologic data and the best understanding of sediment supply as model input.

Modeling results for a one-season removal with minimal dredging indicated that major sediment deposition after the dam was removed would occur only in a few locations: immediately downstream of the dam where the channel receives the sediment released after dam removal, and in a 2- to 3-mile reach downstream of the Sandy River Gorge where the channel is wide and the slope becomes relatively gentle.

Simulation of two-season stepped removal indicated there are minimal benefits in terms of reducing downstream sediment deposition from this option. Simulation of one-season removal with partial dredging indicated the potential benefits in terms of reducing downstream sediment deposition were small and did not warrant the increased cost.

As a result of this assessment, the stakeholders agreed to the one-season dam removal with minimal dredging as the preferred alternative for the application to FERC. Marmot Dam is scheduled for removal in 2007.

Stillwater Sciences also used the information collected during the Marmot Dam removal study to examine the validity of its TUGS (The Unified Gravel-Sand) model. This model was developed to simulate the transport of gravel and sand in gravel-bedded rivers and can be used to evaluate sediment deposition in reservoirs.5 To examine the model, Stillwater Sciences simulated the sedimentation process upstream of Marmot Dam.6

A comparison of model simulation and field data indicates the TUGS model closely reproduced the sediment deposition process, including the bed profile and the stratified sediment deposit upstream of the dam. Figure 1 shows a comparison of the simulated and observed sediment texture, indicating that the model accurately reproduced the two depositional units: the gravel deposit at the top and the sand deposit beneath the gravel layer. In addition, numerical simulation produced two gravel lenses within the sandy layer that were found to approximate two similar lenses indicated in the coring log. This comparison indicates TUGS can satisfactorily reproduce the stored sediment deposit in great detail.

Studying removal of Soda Springs Dam

PacifiCorp studied its 45-foot-tall Soda Springs Dam for potential removal during FERC relicensing of its North Umpqua River project. About 810,000 cubic yards of sediment, mostly sand-sized particles, is trapped in Soda Springs Reservoir. To understand the potential effect of downstream release of this sediment, Stillwater Sciences developed a one-dimensional sediment transport model.

Soda Springs Dam on the North Umpqua River in Oregon was studied for removal, but sediment transport numerical modeling indicated sediment accumulated behind the dam would wash out quickly in concentrations lethal to all downstream salmonids. The decision was made to leave the dam in place. Click here to enlarge image

Numerical simulation indicated that the sediment in Soda Springs Reservoir would be transported out of the impoundment and downstream quickly after dam removal, resulting in short-term (within about two weeks) channel aggradation in a 9-mile reach. (See Figure 2 on page 46.) During this period of time, the model predicted that the suspended sediment concentration would increase to up to 30,000 parts per million and then decrease quickly and disappear once the sediment shown in Figure 2 passes through the reach. This concentration would be lethal to all salmonids downstream of the dam.7

In this instance, the model cannot be calibrated by reproducing the existing river profile because North Umpqua River is a gravel-bedded river, while the model simulates only the transport of fine sediment. However, similar models have been used elsewhere and found to provide satisfactory results (e.g., the simulation of sediment transport in the Ok Tedi and Fly rivers in Papua, New Guinea, that involves millions of tons of additional fine sediment input to the river due to mining operation).

The model results were one of the considerations that led to the decision not to remove Soda Springs Dam.


The sediment transport numerical models used for studying the removal of Soda Springs and Marmot dams were customized for the projects, focusing on the prediction of sediment deposition downstream of the dams after their removal. Based on these two studies, Stillwater Sciences developed two more versatile sediment transport numerical models, Dam Removal Express Assessment Models (DREAM-1and DREAM-2), for the National Marine Fisheries Service (NMFS).8,9 DREAM-1, DREAM-2, and TUGS simulate multi-grained sediment transport under very large temporal and spatial scales, making them especially suitable for application in natural rivers for evaluation of long-term consequences resulting from a hydro project. DREAM-1 has been applied in the Klamath River for simulation of sediment transport following the potential removal of dams associated with the 18-MW Iron Gate, 47-MW Copco, and 90-MW J.C. Boyle hydro projects.

Drs. Cui and Orr may be reached at Stillwater Sciences, 2855 Telegraph Avenue, Suite 400, Berkeley, CA 94705; (1) 510-848-8098; E-mail: or Mr. Ligon may be reached at Stillwater Sciences, 850 G Street, Suite K, Arcata, CA 95521; (1) 707-822-9607; E-mail: Mr. Heintzman may be reached at Portland General Electric, 121 S.W. Salmon Street, 3WTC-BRHL, Portland, OR 97204; (1) 503-464-8162; E-mail: david.heintzman@


  1. “Iowa Hill Pumped Storage Development Turbidity Analysis,” prepared by Stillwater Sciences for Sacramento Municipal Utility District, Sacramento, Calif., 2004. Available at docs_iowa.htm.
  2. “Sandy River Sediment Study, Bull Run Hydroelectric Project,” prepared by Squier Associates for Portland General Electric, Portland, Ore., 2000.
  3. Numerical Modeling of Sediment Transport in the Sandy River, OR Following Removal of Marmot Dam,” prepared by Stillwater Sciences for Portland General Electric, Portland, Ore., 2000. Available at whatwedo/sedtranspubs.
  4. Cui, Yantao, and A. Wilcox, “Development and Application of Numerical Models of Sediment Transport Associated with Dam Removal,” Sedimentation Engineering, Theory, Measurements, Modeling, and Practice, ASCE Manual 110, American Society of Civil Engineers, Reston, Va., In press.
  5. Cui, Yantao, “The Unified Gravel-Sand (TUGS) Model: Simulating Sediment Transport and Gravel/Sand Grain Size Distributions in Gravel-Bedded Rivers,” manuscript submitted to Water Resources Research, 2006.
  6. Cui, Yantao, “Examining the Dynamics of Grain Size Distributions of Gravel/Sand Deposits in the Sandy River, Oregon with a Numerical Model,” River Research and Applications, 2007. Available at rra.1012.
  7. Newcombe, C.P., and J.O.T. Jensen, “Channel Suspended Sediment and Fisheries: A Synthesis for Quantitative Assessment of Risk and Impact,” North American Journal of Fisheries Management, Volume 16, No. 4, November 1996, pages 693-727.
  8. Cui, Yantao, C. Braudrick, G. Parker, W.E. Dietrich, and B. Cluer, “Dam Removal Express Assessment Models (DREAM). Part 1: Model Development and Validation, and Part 2: Sample Runs/Sensitivity Tests,” Journal of Hydraulic Research, Volume 44, No. 3, March 2006, pages 291-323.
  9. Cui, Yantao, et al, “Sediment Pulses in Mountain Rivers: 1. Experiments,” Water Resources Research, Volume 39, No. 9, September 2003. Available at 001803.

Yantao Cui, PhD, is a senior scientist and hydraulic engineer, Bruce Orr, PhD, is a senior ecologist and principal, and Frank Ligon is a senior fisheries biologist and chairman at Stillwater Sciences. Cui developed and applied the numerical models discussed in this article, Orr provided ecological consultation and project management for the Marmot Dam removal study, and Ligon provided biological consultation and project management for the Soda Springs Dam removal study. David Heintzman, project manager at Portland General Electric, managed implementation of the Marmot Dam removal.

µ Peer Reviewed

This article has been evaluated and edited in accordance with reviews conducted by two or more professionals who have relevant expertise. These peer reviewers judge manuscripts for technical accuracy, usefulness, and overall importance within the hydroelectric industry.

Previous articleKeeping up with the Growing U.S. Demand for Renewable Energy
Next articleDutch Power Company to Use Coffee Husks as Fuel
Renewable Energy World's content team members help deliver the most comprehensive news coverage of the renewable energy industries. Based in the U.S., the UK, and South Africa, the team is comprised of editors from Clarion Energy's myriad of publications that cover the global energy industry.

No posts to display