Technical Integrity / Collaboration / Exceptional Results
Featured

Numerical Modeling Q&A with Marcia Greenblatt, Ph.D., P.E.

Dr. Marcia Greenblatt, a principal water resources engineer with Integral, answers questions on modeling studies, tools, and trends.
 
Dr. Greenblatt is known for her experience in conducting hydrodynamic, water quality, and sediment investigations. She has extensive involvement in Superfund remedial investigations and feasibility studies at large sediment sites and experience in designing data collection programs and performing modeling studies. Dr. Greenblatt has testified in court as a technical expert on environmental issues and has mediated public conflicts.
 
Q: Integral conducts a wide variety of environmental modeling. What is your specialty area, and what types of clients and projects do you support?
 
My focus is on surface water modeling, which includes rivers, lakes, estuaries, oceans, and watersheds. I’ve worked with industrial clients, either individually or as part of a performing parties group, and for law firms to support litigation, both as a consulting and testifying expert. Other clients are state and federal agencies and watershed groups. I work on projects that range from large, multiyear studies involving fieldwork and data collection designed to answer broad or complex environmental questions, to short, focused studies that are completed in a few weeks. These studies may address flow, runoff, sediment and contaminant fate and transport, and water quality—such as temperature and nutrients. I have developed and applied models for TMDLs and NPDES discharge permitting and have evaluated impacts of various industrial discharge practices on water quality.
 
Q: What are the most important things you take into consideration when performing a modeling study?
 
Developing a robust conceptual site model, or CSM, to describe the important processes in the system before initiating a modeling study is essential to the development of a good numerical model. The CSM uses available data to provide an understanding of what’s happening at a site, and it identifies the relevant or dominant processes that the model needs to capture. The CSM also provides a means to evaluate the ability of the model to simulate the important site processes. Conversely, once a model is developed, it may be used to further our understanding of particular behaviors.
 
A model is one line of evidence to characterize system behavior and response. It should be used with other information from the site, including data and history, to more fully understand the system. Using multiple lines of evidence to explain observed behavior provides a much more compelling story than relying on a model alone.
 
Another important consideration in developing and applying a model is to recognize the uncertainty inherent in any model. The degree of uncertainty will vary among model studies and is a function of the extent of available data and the complexity of the system. Uncertainty doesn’t mean that a model should not be developed, or that model predictions should not be used for decision making. Rather, uncertainty should be acknowledged and considered when using model predictions. The degree of uncertainty that is acceptable will depend on the magnitude and importance of the decision, as well as the strength of the additional lines of evidence that contribute to site understanding.
 
Finally, communication of the model and its results to relevant audiences can be a challenge. Decision makers or fact finders often have limited or no experience with modeling or no technical background. Modeling studies are most useful when the highly technical work can be explained clearly and concisely.
 
Q: Are there times when a model is not the most appropriate tool?
 
A model is a great tool for supporting system understanding, organizing and presenting data about a site, and simulating system behavior for conditions that do not exist today. For example, models are the only way to explore system response to historical discharges or proposed remediation or restoration. But there are times when it may not be appropriate to use a model, or when the use of the model should be limited in application.
 
In complex systems that are not well understood or where there are significant data limitations, a model may not be the best initial step to support decision making. A robust data collection program that can provide critical information to characterize system behavior may be more appropriate. A thoughtful monitoring program can clarify spatial and temporal system responses, for example, how high flow events affect surface water contaminant concentrations. These data can then be used to support and validate a numerical model on the basis of a strengthened understanding of important system processes.
 
In some cases, there are insufficient site-specific data available to support the setup and development of a numerical model. I would caution clients about the pros and cons of using a model in these situations. A model may be helpful to perform a comparative analysis of possible outcomes or scenarios, but may not be appropriate for developing predictions of specific scenarios or conditions.
 
Q: Are there unique challenges to using modeling to support litigation?
 
Modeling to support litigation is most successful when study questions are clearly articulated and the modeling study is set up and developed specifically to answer the study questions. A defined modeling study scope can serve to bound the extent of deposition and cross-examination inquiry. The ability to communicate the model results and implications in lay terms is of increased importance in litigation. Decision makers and fact finders such as judges, juries, or mediators often do not have a technical background, and may be overwhelmed by large amounts of information regarding the matter.
 
Another consideration when performing a modeling study as part of a litigation matter is how the model can best support the legal strategy. A well thought-out modeling study that is clearly and concisely presented to advance cogent technical arguments and address the relevant questions at issue will best serve to support the case.
 
Q: Do you see a change in the way models are being used to support decision making?
 
Computing power is continually becoming faster while getting less expensive. Large models that were previously prohibitive to apply due to enormous computational demands are now feasible. In fact, for big projects, multiple computers are often purchased to accommodate simultaneous model runs, with little impact on the overall project budget. We have come a long way from the days of MS-DOS command lines and small hard drives churning for days or months on the model computations!
 
At the same time, however, more comprehensive models of complex systems often include more uncertainty, as inclusion of additional and/or more detailed processes means that more supporting data may be needed to characterize the relevant processes. There is increasing recognition that, while models provide an invaluable tool to support environmental investigations, they are only one part of the decision making process. For example, a memorandum issued by the Office of Land and Emergency Management to EPA Regional Administrators in January 2017 recommended that, at complex contaminated sediment sites, the limitations of modeling in predicting future concentrations for purposes of decision making be considered. At Integral, we recognize the value of numerical models as well as the associated uncertainty, and always make sure numerical modeling is considered in the context of a larger, holistic view.
 
For more information on Integral’s data collection and modeling capabilities, contact Dr. Greenblatt at mgreenblatt@integral-corp.com.