Report of the Community Sediment Transport Modeling Workshop

June 22-23, 2000 

Executive Summary

A group of 55 scientists and engineers from nine countries met in Woods Hole, Massachusetts on June 22-23, 2000 to explore a community approach to improving deterministic models of sediment transport in coastal seas, estuaries, and rivers. The motivation was that there did not appear to be any well tested models that had been accepted by the research community, and that identification or development of such a model would enhance our ability to use sediment transport models as research tools.

It was the consensus of the workshop participants that:

• Wide acceptance of a community sediment transport model would make the model a more effective tool for scientific research. Hydrodynamic models that incorporate sediment transport processes are available, but are not widely used by the sediment transport research community. This is because no existing model has the features that would allow wide access and acceptance by the community: freely available code; state-of-the-art hydrodynamic and sediment algorithms with modern, modular coding; comprehensive documentation; and demonstrated performance on a suite of community defined test cases. A significant effort is needed to make models more usable by the research community, and to test available and developing models.
  
• A better understanding is needed of the basic physics that control sedimentary processes such as bottom roughness, aggregation or flocculation and disaggregation, erosion and deposition, and bed consolidation. While the requisite observations and theoretical studies are beyond the scope of a modeling effort, a well tested community model would provide a valuable platform for testing and comparing emerging parameterizations of sedimentary processes. Our ability to realistically simulate sediment transport is often limited by our ability to formulate laws for essential sediment processes from first principles, not by the efficiency and sophistication of our numerical models. Because we may not achieve this ability for some years, we need to use models as a context for developing and testing improved parameterizations of these processes. First, however, the models must have a demonstrated ability to correctly model basic problems. Small working groups should be formed from the community to identify the most-important basic research directions relevant to realistic modeling and propose guidelines for appropriate and specific tests.
  
• Coordination of the community modeling effort should be by an impartial organization with long-term stability. Although the development effort can be shared by the community to some extent, there needs to be sustained support of core activities to guide development efforts, facilitate evaluation by providing test cases, maintain the source code and user documentation, develop a forum for user group discussion, and maintain a web site. Ideally, these activities should be performed by an established and impartial organization with coastal science expertise and a mission that involves sediment transport.

Introduction

Community models are valuable tools in many scientific disciplines. Their availability provides a diverse group of researchers and engineers with a common medium for investigating physical phenomena, and a starting point for derivative models that can be developed to address specialized questions. This approach has succeeded in the hydrologic community, with widespread use of the groundwater model MODFLOW, and in the atmospheric community with models such as MM5. Collaboration within the communities that built these models has allowed scientists to focus on the questions of their choice, rather than on building a model from scratch. These models have provided platforms for testing sub-models, and enabled nesting in a well designed and tested framework.

Within our community, there are several models and ongoing development efforts, and candidates for a public-domain model already exist. Some of those models were discussed and evaluated at the workshop. Workshop participants were tasked with assessing the present state of sediment transport modeling in rivers, estuaries and coastal seas, and charting the optimal path towards establishing community models.

The first day of the workshop was informational. The morning was devoted to plenary presentations highlighting challenges in understanding processes associated with cohesive sediments, noncohesive sediments, coastal waves, and coastal circulation. In the afternoon, presentations describing representative examples of existing deterministic sediment transport modeling systems were made, and reports on the progress and lessons of three ongoing community modeling efforts were provided.

The second day of the workshop was devoted to exploring the major issues and directions that should be taken as a community. During the morning, five working groups met independently to address the same two questions:

• What is limiting our ability to realistically simulate sediment transport?
• What is limiting our ability to use models as scientific tools?

Input from every meeting participant was solicited, and through the group leaders and the following plenary summary of the workgroup findings, consensus on the major needs and goals of the community obtained. In the afternoon, there was a plenary discussion on what the best approaches, over both short- and long-term time scales, should be to address the needs and goals of the community.

Summary of Talks on Processes, Models, and Community Efforts

The first day of the workshop included summaries of the state-of-the-art understanding of sediment transport processes, existing sediment transport models, and community modeling efforts. Experts in the fields of non-cohesive and cohesive sediment transport, sediment entrainment, near-shore processes, and oceanographic circulation provided summaries of their fields, and discussed the challenges being addressed by current research. Experts from all fields agreed that:

• Much remains to be learned about small-scale sediment transport processes such as particle flocculation, sediment entrainment, bed cohesion, and near-bed turbulence.
• If sufficient site-specific data are available (e.g., bottom roughness, bed sediment size distributions, sediment settling velocities, critical bed shear stress), then available sediment transport formulations can be calibrated to predict sediment flux with reasonable accuracy.
• The ability to extend these calculations to a regional scale is limited by difficulty in prescribing boundary conditions, in specifying the flow field, and in knowledge of local sediment properties such as size distributions, settling velocity, and entrainment rate.

Presentations describing five sediment transport models provided a sampling of the many available models:

• EFDC, a freely-available, curvilinear orthogonal coordinate, coupled hydrodynamic, water-quality, and sediment model developed by TetraTech, Inc. and currently being improved by the U. S. Environmental Protection Agency.
• STP, the non-cohesive sediment transport program included in both the LITPACK (1D coastal processes) and MIKE 21 (2D wave, hydrographic and sediment transport processes in estuaries and coastal areas) packages marketed by the Danish Hydraulic Institute.
• CH3D-U.Fl, a non-orthogonal curvilinear grid academic model developed under the leadership of Y. Peter Sheng at the Univ. of Florida.
• CH3D-ACOE, derived from an earlier version of the Univ. of Florida model and used for engineering studies by the U. S. Army Corps of Engineers.
• ECOM-SED, built around the Blumberg-Mellor hydrodynamic model and commercially marketed by HydroQual, Inc. and Delft Hydraulics.

Each presentation summarized model capabilities, discussed model parameterization of non-cohesive and cohesive sediments, noted particular strengths of the model, summarized efforts at assessing the model’s performance, and listed improvements that would most enhance the model. The presentations illustrated that:

• Coupled hydrodynamic-sediment transport models are being widely applied, particularly by the engineering community.

• Available models span a broad range of process parameterizations, numerical sophistication, model availability, level of documentation, and the confidence with which the model is applied.

• Proper application of their model required considerable insight in specifying model parameters and inputs, without which the model could give spurious results.

Three community efforts were then discussed; a National Ocean Partnership Program (NOPP) funded effort to develop a nearshore sediment transport model, an Office of Naval Research (ONR) funded effort aimed at building a next generation terrain-following coordinate ocean model, and an European Union funded project to produce COHERENS, a community model for hydrodynamic, sediment and biological simulations. The three efforts are in different stages of development:

• COHERENS is at the most advanced stage of development, now being distributed on CD-ROM (100+ copies worldwide).  Much of the recent COHERENS effort, in fact, was aimed at generating a "product" that could be widely used. This required that the developers and scientists take time to produce a comprehensive users’ guide, which not only tells the user how to run the model, but also describes the formulations of model algorithms.  It also includes over 50 test cases covering physics, sediments and biology.

• The ONR project is using ROMS (developed by Rutgers and UCLA) as a starting point.  ROMS includes 3-d hydrodynamics and bottom boundary layer calculations, and has been distributed to a limited number of users. Efforts to date have concentrated on developing the sophisticated model code and ensuring the model will run efficiently on parallel architectures. Efforts in the future will include model tests, linking with the operational atmospheric model COAMPS, and incorporation of sediment transport algorithms.

• The U. Delaware led NOPP modeling effort is in the early stages of development, with scientists concentrating on developing formulations for sediment transport valid for the nearshore zone. Future efforts as a part of this project will include field experiments to evaluate the performance of these algorithms.

Results of the Working Groups

The working groups determined that one or more community sediment transport model system(s) would be extremely useful, and listed the characteristics that are desirable in such a system.

Evaluate existing models: Our ability to numerically predict sediment transport would be most improved by evaluating existing models and critically examining existing parameterizations of sediment transport processes. The sediment transport modeling community should evaluate the sensitivity of calculations to uncertainties in process dynamics, site-specific information, and forcing functions. These efforts require collaboration between sediment transport modelers and observationalists.

Develop rigorous model tests: Existing and developing transport models must be evaluated against a variety of tests, including analytical solutions, laboratory data sets and field observations. A group of modelers and observationalists therefore need to collaborate to identify existing data sets and analytical solutions for comparison with model calculations, and recommend field or lab measurements to obtain data sets that do not yet exist but are needed to fully test models. Once these data sets have been identified and made available, existing models and formulations should be tested against them. The group should encourage appropriate agencies to fund efforts to fill gaps in data sets deemed critical for model evaluation. The sediment transport modeling community can help to guide this effort by conducting sensitivity tests to determine which parameters or processes are most poorly constrained and have the most significant impact on sediment transport predictions. By completing this effort in a community forum, guidelines and test sets will be available with which to evaluate new sediment transport models as they are developed.

Improve understanding of basic physics: Many sediment transport processes are poorly understood, including bedform formation and migration, bottom roughness, particle aggregation, dissaggregation and flocculation, bed compaction, and sediment entrainment. The sediment transport modeling effort should use sensitivity tests to evaluate how important each of these processes are in overall sediment transport calculations. This effort should help to direct process studies by providing insight into which of the processes have the most significant impact on sediment transport calculations. Links between process studies and development of a community sediment transport model would be encouraged by collaboration between observationalists and modelers, but the primary funding of process studies is expected to come from other organizations and efforts.

Improve numerical methods for sediment transport modeling: Data assimilation is an exciting area of development in oceanographic and atmospheric forecasting efforts, and could also be powerful as applied to coastal sediment transport. The physical oceanographic community has rigorously evaluated the numerical methods used to solve 3-d hydrodynamic problems, but a similar effort has not been applied to sediment transport calculations. The evaluation component of the community modeling task should determine whether widely available numerical methods (e.g., advection schemes) are sufficient for solving sediment transport calculations, or whether efforts need to be directed at improving these numerical schemes. The ability to run the calculations in a nested framework was also discussed as important for tackling regional-scale sediment transport problems.

Develop Guidelines for Developing Model Input and Assessing Uncertainty: The working groups furthermore recommended that guidelines be developed that list the site-specific information needed to accurately predict sediment transport, and the sensitivity of the calculations to uncertainty in this information. The site-specific information covered by these guidelines could include sediment properties, seabed properties, and flow characteristics.

House and maintain the community model with an institution with a long-term commitment: The sediment transport modeling framework must be modular and extensible for effective community contribution, but should be centrally maintained. The modeling system must be flexible enough to be applied to a variety of research and engineering questions, and to evolve along with our understanding of sediment transport processes. A central location for model source code and documentation was recommended, however, to maintain some consistency in the system. The workshop participants felt that the leadership and funding for this support should come from a group that has a stable source of funding for sediment transport issues, and that does not have a proprietary or commercial interest in the model.

Distribute model code freely with clear documentation and user support: For a community model to be embraced, it must be freely accessible, run on many operating systems, and be well documented. Simple test cases should be delivered with the model source code so that new users can test their implementation of the model. The community also voiced a need for pooling resources to develop visualization tools. User support is important for a widely used model, and workshop participants advocated centralized, long-term support. At a minimum, this includes maintenance of a user mailing list and model web site to foster the community of trained users vital for a model to realize its potential.

Develop a modular, flexible system with open source code: The model should be capable of addressing many research and engineering questions. This requires that it be modular and easily extended, and participants agreed that open access to source code was a requirement. The model should be as flexible as possible, and include the ability to run simple calculations (e.g., 1- or 2-d cases, or steady-state situations) as well as complex practical applications in 3-d with time-dependent forcing. The flexibility and modularity should also allow for users to modify the sediment transport processes that are contained in the code, and even to take the sediment transport modules and place them into a different hydrodynamic model.

Pool resources to expand model capabilities: One advantage of a community model is the ability to pool resources. This should be taken advantage of by having a clearing house to which users can contribute new model components or model tools such as visualization and I/O tools.

What Next? (action items to move ahead)

The final plenary discussion of the workshop was to determine future directions for the community sediment transport modeling effort; that is, to answer the question: What next? The discussion provided the following list of issues and planned action items:

Issue: Dissemination of workshop results.

Action: This report is the first draft of a white paper reporting the community consensus regarding future modeling efforts. Subsequent drafts will reflect comments received by the research and engineering community.

Issue: Expand model community.

The workshop participants represent a small (but hopefully representative) slice of the intended user group for a community model. The success of the community model effort will depend on broad support and participation.

Action: Advertise workshop results via web and publication of a workshop report in EOS. Directly communicate with funding agencies (ONR, NSF, NOPP, others?) and management in federal agencies (USGS, EPA, MMS, NOAA, ACOE, DOE). Plan future meetings or workshops, with expanded invitations.

Issue: Process oriented test problems have been developed for ocean circulation models (e.g. Haidvogel and Beckman, 1999), but similar tests need to be developed for sediment transport models so we can quantify the behavior of different models and methods.

There should also be test cases that are real-world applications from studies where abundant field data is available. These test problems should to be chosen by the community, and test data should be made publicly available for model evaluation and comparison.

Action: An ad hoc subcommittee will generate a strawman list of test problems and will circulate it to the community for comment.

Issue: Sediment transport models require a hydrodynamic modeling kernel.

Suitable short-term and long-term hydrodynamic modeling efforts need to be identified. Important factors are modular, structured code, free availability, comprehensive user documentation, and documented verification of model performance on simple test problems.

Action: The USGS will generate an assessment of the strengths and weaknesses of several candidate models (EFDC, CH3D, COHERENS, ROMS) and will circulate it to the community for comment.

Issue: Although model testing and improvements can come from the community, there needs to be an institutional home for the modeling system that coordinates releases, establishes and maintains standards, hosts a web site and mailing list and provides clearing house for model discussion and information.

Ideally, the institution would be unbiased and have long-term stability.

Action: Government agencies are ideal institutions for housing community model developments. USGS will explore with EPA, MMS, NOAA, and the ACOE the roles that are appropriate for each organization.

A timeline for completing these action items has been developed and will be updated as progress is made.

References

Haidvogel, D. B. and A. Beckmann (1999) Numerical Ocean Circulation Modeling, Imperial College Press, London, 318 pp.

 
List of Attendees
Timeline of Action Items