Water Resource Management Models

Water-resource models have been used to inform decisions about water supplies, ecological restoration, and water management in complex regional systems.

Water-resources development projects inevitably include economic, environmental, and social considerations, as well as computer-based models that can clarify trade-offs and help identify the plans, designs, and policies that will maximize desired impacts and minimize undesired ones. However, by design, models are simplifications of real systems. Therefore, predictions of how a real system will function under alternative designs and management policies are often controversial and always somewhat uncertain, because they necessarily include assumptions about future events and conditions that are not known. In short, modeling is a necessary part, but only a part, of the information used by decision makers. That said, I am willing to bet that every single major water-resources planning and management activity in the world today, whether focused on flooding problems, reservoir operation, groundwater development, water allocation, or aquatic ecosystem enhancement, includes models.

Thanks in part to advances in computer technology in the past 30 years, the most comprehensive models can now include engineering, economic, ecological, hydrological, and sometimes even institutional and political components of large, complex, multipurpose, water-resources systems. Applications of models to real systems have improved our understanding of how both work. This experience has also taught us the limitations of modeling complex interdependent physical, biochemical, ecological, social, legal, and political water-resource systems. Nevertheless, models are an essential part of the planning process, and models of varying complexity, and thus of varying data requirements, are available, as applicable.

Two major types of computer-based models are simulation models and optimization models. Simulation models address “what if” questions. Given the assumptions for a system design and operation, a simulation model can predict how well it will perform. Optimization models address “what should be” questions, what design and operating policy will best meet the specified objectives. However, the methods (algorithms) used to solve optimization models often limit the amount of detail they can include; thus they are less flexible than simulation models. So, especially for complex systems, optimization may be used first to screen out unsatisfactory alternatives; simulation models can then be used to investigate the remaining more promising alternatives.

Capabilities of Models
Models available today can predict runoff from watersheds due to precipitation and the sediment, nutrient, and other constituent loads in that runoff. Models can predict interactions between ground-water and surface water bodies, as well as flows and their constituents in stream and river channels. These routing models can be based on simple mass balance and advection-dispersion relationships, or they can be based on hydrodynamic computations. They can also include ecological components of aquatic systems. Models can be used to study reservoir operation; to forecast floods and plan flood-control programs; to predict storm surges, embankment erosion, and dam breaks; and to plan for ecosystem restoration.

Federal, state, and local water-resources planning agencies can and do use models for the real-time operation and management of water systems and for assessing water supply availability and reliability, taking into account projections of basin development, administrative and legal requirements, and other constraints related to conveyance and reservoir operations, aquifer pumping, and water demand and distribution. They use models for determining facility-size requirements for water storage and distribution systems given water diversions for domestic, commercial, industrial, rural, irrigation, municipal, energy, and environmental uses. They undertake watershed hydrology studies for improving river and reservoir administration and operation, the conjunctive use of surface water and groundwaters, water banking, and water exchanges and transfers. Many agencies, consultants, and university researchers are also working to improve water-resource analysis tools to meet their modeling and prediction needs.

Models can predict changes in a river bed form and location, including bank erosion, scouring, shoaling associated with construction or changes in the hydraulic regime, and so on. Models involving the transport of sediment have been used for morphological studies in small- and large-scale rivers of all kinds and in all settings, as well in reservoirs. Simulations can cover a few hours to several decades of high runoff events.

Hydrologists increasingly
rely on GIS data and
standardization to solve
a variety of problems on
different spatial scales.

Geographical information systems (GIS) are increasingly being included in planning and management models, and hydrologic models can be directly linked to GIS databases. These tools are useful for 2- or 3-dimensional spatial calculations, such as delineating watershed boundaries and stream and river paths, defining drainage areas and the areal extent of any other data layer, and for modeling distributed runoff. In the future, hydrologists will increasingly rely on GIS data and standardized ways of describing them so that they can be used consistently to solve a wide variety of problems at various spatial scales.

Water-resources planning and management activities are often participatory, involving groups of individuals with different interests, goals, and needs, thus requiring negotiation and compromise. So-called “shared-vision models” can be designed to meet the information needs of all stakeholders. The U.S. Army Corps of Engineers Institute for Water Resources has been actively involved in the development and use of such models, especially for planning for droughts and for water-allocation studies (http://www.svp.iwr.usace.army.mil/). To be most useful, shared-vision models must provide the right information, and the right amount of information to meet all stakeholder needs. Meeting these requirements can be a challenge to model developers.

Hydroinformatics, a term that originated and was popularized in Europe and was associated at first with computational hydraulic modeling, also recognizes the need to communicate model results in more meaningful ways. Hydroinformatics today emphasizes the use of artificial intelligence techniques (e.g., artificial neural networks, support vector machines, genetic algorithms, and genetic programming). These techniques can also be used for mining large collections of observed data or data generated from physically based models for knowledge discovery.

Models were instrumental
in the adaptation of a
cooperative regional plan for
the Washington, D.C., area.

Examples of Modeling Applications
The three examples below illustrate how models have been and are being used. The first involves the Potomac Basin and describes how modeling helped solve a highly visible and highly political water-supply problem. The next two examples, involving the Everglades region and part of the Great Lakes Basin, are ongoing studies that require much more detailed modeling methods. Although all three examples are in the United States, many others could have been chosen from other parts of the world.

The Water Supply for the Washington, D.C., Metropolitan Area
Since the early 1600s, when humans first settled in the Potomac River Basin, concerns have been raised about the river and how it is used. A main reason for these concerns is the rapidly increasing population, particularly in the Washington, D.C., metropolitan area (WMA), where three water-supply agencies provide water to their customers. Traditionally, these agencies have operated independently, concerned only with satisfying the needs of their own constituents.

However, with a rapidly increasing population, but relatively stable river hydrology, it became apparent in the 1950s that frequent, severe droughts could lead to serious problems. To address these concerns, the U.S. Army Corps of Engineers carried out a number of studies and, in 1963, proposed the construction of 16 new reservoirs in the Potomac Basin. Two were subsequently authorized by Congress, and construction was even begun on one of them in spite of substantial local opposition.

A simulation study in the late 1960s and early 1970s by graduate students at Johns Hopkins University showed how the coordinated use of the water stored in existing reservoirs in the Potomac River Basin during droughts would largely eliminate the need for new reservoirs, at least until 2020. However, they pointed out, the public would have to accept some risk of water restrictions during droughts, and the three utilities would have to cooperate with each other and coordinate their operations.

This was the first time a regional perspective was suggested. The idea had widespread political, legal, social, and environmental dimensions, and the models played a major role in convincing everyone that a cooperative policy would work. In the end, models were instrumental in bringing about a consensus supporting the nonstructural plan (Anon, 1983; Hagen et al., 2005).

On July 2, 1982, at a historic ceremony in the District of Columbia Building, contracts were signed ensuring an adequate water supply for WMA until 2050. The regional water-supply system was expected to cost about $31 million, whereas the federal reservoirs that were originally proposed would have cost about $400 million (McGarry, 1990). The savings far exceeded the money spent up to that time to support research on water-systems modeling.

The original simulation model, now named PRRISM (Potomac Reservoir and River Simulation Model), has undergone several modifications and is currently used to evaluate the response of the system to present and future water demands. PRRISM is also used in drought-management exercises, which are conducted every five years (Hagen et al., 2005; Sheer, 1981). The model is an important tool for the evaluation and development of updated drought-management policies. In this sense, WMA is applying an adaptive-management approach to water-supply planning and management.

Greater Everglades Restoration
Historically, the Greater Everglades in south Florida covered an area of about four million acres. During the last 100 years, approximately half of the Everglades have disappeared as a result of drainage, channelization, and other changes made to allow for increased agricultural and urban growth. Today the water-management system in south Florida encompasses 18,000 square miles with more than 1,800 miles of levees and canals, about 200 major water-control structures, and nearly 30 pumping stations. However, the Everglades system today (Figure 1) can no longer meet the environmental and water-supply requirements of the current population.


Figure 1 Historic watershed and current features of the Everglades region.

Unprecedented efforts are under way to restore the Greater Everglades ecosystem, involving the creation of more floodplain wetlands (including from some recently purchased land from the sugar industry), improving water quality, reconnecting natural river channels, and improving the habitat of more than 300 species of fish and wildlife—while simultaneously meeting the needs of the urban and agricultural sectors in the region. Restoration is expected to take more than 30 years and to cost more than $10 billion. The Comprehensive Everglades Restoration Plan, which has more than 60 components and has been described as the largest ecosystem restoration effort in the world, is designed to achieve a balance between ecosystem restoration and urban and agricultural water supply.

Computer modeling has been and continues to be a critical component in the development and implementation of the Everglades restoration plan (Tarboton et al., 1999). The Everglades project requires a multidisciplinary modeling approach, as is evident in the models that have been developed and used thus far:

  • The Everglades Screening Model (ESM) simulates the major hydrologic features and demands of the south Florida water-resources system. The ESM was used extensively during the screening phase of the Everglades restoration plan.
  • The South Florida Water Management Model (SFWMM) simulates the surface and subsurface hydrology under existing and proposed water-management plans in the south Florida region using a fixed grid and climate data for 1965 to 2000. The SFWMM is a premier hydrologic simulation model that has been used for system-wide evaluations of proposed Everglades restoration plans. It is currently being replaced by a finite-volume, variable-grid regional simulation model.
  • The Natural System Model (NSM) simulates the hydrologic condition of the Everglades before they were drained. The NSM uses the same climatic inputs, time step, calibrated model parameters, and algorithms as the SFWMM. The NSM has been essential to setting environmental restoration targets for the Everglades restoration plan.
  • The Everglades Landscape Model (ELM), which combines hydrologic, water quality, and ecological processes in a single model, predicts spatial and temporal patterns of change in the landscape and interactions and feedback among water, nutrients, soils, and wetland plants.
  • The Across Trophic Level System Simulation Model (ATLSSM) simulates the biotic communities of the Everglades/Big Cypress region and the abiotic factors that affect them. The ATLSSM is used to clarify how the biotic communities of south Florida are affected by the hydrologic regime and other, abiotic factors and as a tool for evaluating management alternatives (http://atlss.org).
  • The River of Grass Evaluation Methodology (ROGEM) predicts the relative quality of habitat responses to Everglades restoration alternatives.

The use of these models for decision making is illustrated in Figure 2 (www.sfwmd.gov).


Figure 2 Use of models in the Everglades restoration planning process.

Lake Ontario-St. Lawrence River Study
In April 1999, the International Joint Commission (IJC), which oversees all transboundary waters along the entire Canadian-U.S. border, obtained funds from the governments of Canada and the United States to determine how the management of water levels and outflows in Lake Ontario might be improved in light of public concerns about the environment and adjacent ecosystems and in response to potential climate change. After spending $20 million and taking more than five years to complete a review, the Canadian-U.S. study team submitted three alternative plans to the IJC for consideration (LOSLSB, 2006). The three plans represented different combinations of trade-offs to meet conflicting objectives. The IJC is now in the process of deciding which alternative is “best,” but it appears the plan being proposed by the IJC is not the plan many stakeholders wanted (http://www.greatlakesforall.com/2008/04/ijc-ignores-hea.html; http://www.citizenscampaign.org/campaigns/great_lakes.htm).

Lake Ontario is the most downstream of the Great Lakes, which define part of the boundary between Canada and the United States. The lake receives water from the other four Great Lakes, as well as from the local watershed. Water from the lake is discharged into the St. Lawrence River, which flows northeast past Montreal and Quebec into the Gulf of St. Lawrence and the Atlantic Ocean. Water levels in Lake Ontario and the upper portion of the river, and the flows and water levels in the lower portion of the St. Lawrence River, are regulated, to some extent, by the operation of the Moses Saunders Dam, which separates the upper and lower portions of the river (Figure 3).

Regulation of the Lake Ontario-St. Lawrence River system requires balancing several conflicting water-management objectives, which are inherent in the management of flows and lake levels. For example, alleviating high water levels in Lake Ontario requires releasing more water, which may cause flood-related damage downstream because of high-water conditions in the river. Low water levels in the river can adversely impact shipping, but raising the water levels in the lower river requires releasing water from Lake Ontario, which may cause problems for recreational boaters and municipal water suppliers along the upper river and lake shore. In addition, managing the variability of water levels to accommodate ecosystem needs introduces another level of complexity.


Figure 3 Map of Lake Ontario and the St. Lawrence River and their watersheds.

Uncertainty about future water supplies from the upper Great Lakes and tributaries in the Lake Ontario-St. Lawrence River Basin adds to the difficulty of deciding how much water to release through the dam to balance upstream and downstream needs. For example, if future supplies are unexpectedly low, releases made to alleviate low water levels in the river may drain too much water from Lake Ontario, making it much more difficult to raise river levels later when the impacts may be even worse. Similarly, if future supplies are unexpectedly high, releasing less water to avoid minor downstream flooding may increase damage later, if releases then have to be increased dramatically.

The timing of water availability at different times of year is important, in different ways for different reasons. The level of commercial navigation and recreational boating drops considerably in the winter. The value of energy generated in the summer during periods of peak energy demand can be more than 12 times the value of energy generated in the spring. Larger releases of water from Lake Ontario reduce the water levels of Lake St. Lawrence, which is immediately upstream of the hydropower dam. If too much water is released, the levels in Lake St. Lawrence can be lowered to the point that they become hazardous to navigation and could even cause groundings. On the other hand, high flows can create cross-currents that make it difficult to control vessels. Finally, although more electricity can be generated when a greater volume of water passes through the turbines, this reduces the head (i.e., the level of water in Lake St. Lawrence upstream of the dam), which in turn reduces the amount of electricity generated for each cubic meter of water.

During the five-year study period, considerable physical, economic, environmental, and ecological data were collected, and numerous computer optimization and simulation models were developed and used for economic, environmental, and ecological impact assessments under possible climate-induced changes in regional hydrology. The following list describes models that were developed and used for this study:

  • The Flood and Erosion Prediction System (FEPS) Model was developed to assess shoreline erosion rates and damage over time.
  • The St. Lawrence River Model was used to estimate the impact of water levels on existing shoreline-protection works, such as structural failures or the need for increased maintenance, and the associated economic costs.
  • The Integrated Ecological Response Model (IERM) was used to estimate how different regulation plans would impact plant and animal species in the eco-systems in Lake Ontario and the St. Lawrence River.
  • Various policy-generation models, using stochastic optimization as well as simulation, were used to identify and evaluate real-time operation policies and operating policies that could be implemented without periodic modeling. These numerous plans could then be simulated in more detail, using both FEPS and IERM, and an overall shared-vision model.
  • Statistical hydrologic models were used to generate alternative time series of inflows to the system, which were then used in policy simulations to ensure the reliability, resilience, and robustness of each policy. Some of these time series, which were as long as 50,000 years, were used to analyze four different climate-change scenarios. Each of the candidate plans was thoroughly tested to ensure that none had fatal flaws that would inhibit its performance under plausible extreme conditions.
  • A Shared-Vision Planning Model incorporated the results of all other models; the environmental science, economics, and public responses that had been input into an interactive analytical framework to help the study team and public interest groups explore numerous variations of the plans; operating nuances; and performance impacts. The shared-vision model was developed for stakeholders as well as the study team for use on the Internet via an interactive Excel-based program called the Boardroom (http://losl.org/boardroom/main_e.php).

The Lake Ontario-St. Lawrence River study provided a unique opportunity for making changes to the overall system to see if the operation of the system would be improved, and, if so, to determine how. In the opinion of most of the participants, the study succeeded in developing three plans, all of which should perform better than the current operating regime in terms of overall net economic and environmental benefits. Nevertheless, all three plans involve trade-offs, and some stakeholders may have a higher risk of some impacts (e.g., shoreline erosion) than with the current operating policy. In the end, if negatively impacted stakeholders “yell loud enough,” they might prevent the adoption of any of the three new plans. This brings us back to the fact that information derived from models is just one factor in the decision-making process.

Water-resources planners and managers are becoming increasingly aware of the importance of establishing an open, participatory decision-making process that requires close coordination among the many institutions that manage water resources and a strong focus on the stakeholders impacted by management decisions. This awareness has increased the use of analytical modeling tools, such as decision-support systems, that can promote consensus building and dispute resolution and the integration of water, energy, and land-use planning.

The situation presents numerous intellectual, analytical, and evaluative challenges for the developers of models, and overcoming these challenges will certainly require research. Unfortunately, funding for such research is a rare commodity in the current climate.

Anon. 1983. Water supply. Civil Engineering 53(6): 50–53.
Hagen, E.R., K.J. Holmes, J.E. Kiang, and R.S. Steiner. 2005. Benefits of iterative water supply forecasting in the Washington, D.C., metropolitan area. Journal of the American Water Resources Association 41(6): 1417–1430.
LOSLSB (Lake Ontario–St. Lawrence Study Board). 2006. Options for Managing Lake Ontario and St. Lawrence River Water Levels and Flows. Final report of the International Lake Ontario–St. Lawrence River Study Board. Washington, D.C., and Ottawa, Canada: International Joint Commission.
McGarry, R.S. 1990. Negotiating water supply management agreements for the National Capital Region. Pp. 116–130 in Managing Water-Related Conflicts: The Engineer’s Role, edited by W. Viessman and E.T. Smerdon. Washington, D.C.: American Society of Civil Engineers.
Sheer, D.P. 1981. Assuring water supply for the Washington Metropolitan Area—25 years of progress. Pp. 39–66 in A 1980s View of Water Management in the Potomac River Basin. Report of the Committee on Governmental Affairs, U.S. Congress, Senate, 97th Congress, 2d Session, November 12. Washington, D.C.: U.S. Government Printing Office.
Tarboton, K.C., C. Neidrauer, E. Santee, and J. Needle. 1999. Regional Hydrologic Modeling for Planning the Management of South Florida’s Water Resources through 2050. Proceedings of the Annual International Meeting of the ASAE, Toronto, Ontario, Canada.

Further Reading
Loucks, D.P., and E. van Beek. 2005. Water Resources Systems Planning and Management: An Introduction to Methods, Models, and Applications. Paris, France: UNESCO Press. Available online at http://ecommons.library.cornell.edu/handle/1813/2798.

About the Author: Daniel P. Loucks is a professor, School of Civil and Environmental Engineering, Cornell University, and an NAE member.