COMPUTER MODELS FOR INTEGRATED HYDROSYSTEMS MANAGEMENT

COMPUTER MODELS FOR INTEGRATED HYDROSYSTEMS MANAGEMENT

MESSELE Z. EJETA

Graduate Student

LARRY W. MAYS, Ph.D., P.H., P.E.

Prof. of Civil and Environmental Engineering

Department of Civil and Environmental Engineering

Arizona State University

Tempe, Arizona 85287

Abstract

The work in this paper is directed towards two aspects: review of the concepts of integrated hydrosystems management and computer models used for integrated hydrosystems management. The term "integrated hydrosystems management" is used in order to be all inclusive of various types of water systems ranging from water distribution systems and storm water management systems to entire watershed systems and river basin systems. Review of hydrosystems computer models developed starting prior to the 1960s to the present day shows enormous evolution/revolution of computer programming. These efforts which started in the early days of computer programming for the simplification of calculation of analytical functions have now reached the age of what is being referred to in computing technology as "artificial intelligence" whereby it has become possible to write computer programs that not only evaluate a hydrosystems problem, but also draw preliminary conclusions based on the results and recommend appropriate actions based on the conclusions.

An attempt has been made herein to categorize computer programming techniques and models useful for hydrosystems management into simulation models, optimization techniques and decision support systems (DSS). A taxonomy of some of the more widely used simulation models in the U.S. is given. The discussion of different optimization techniques ranging from mathematical programming to heuristic search techniques including genetic algorithms and simulated annealing shows the potential resources available for computer programming for integrated hydrosystems management. Incorporating established water policies that take into account the balancing out process of water among competing users in simulation and optimization models help develop DSS that can be used as models used for integrated hydrosystems management. The study of a few of such models manifests the relative importance of these computer programs for integrated hydrosystems management. Only a limited number of DSS for this purpose have been developed and used in the past. However, the availability of technical resources including database management systems, simulation models, optimization techniques and advanced computing technology provide the opportunity for more exploration to develop DSS for integrated hydrosystems management.

1. Introduction

The fact that every living being depends on water to live and its limited availability in terms of both quantity and quality makes it a resource that living beings compete for to live. This precious resource has competitors that need it in one way or another as a result of which it often becomes challenging in space and time to fully satisfy the needs of these competitors for water. The viable solution under such conditions is "balancing out". This may be achieved through integrated hydrosystems management.

Various definitions have been given in the past to integrated resource management in general and water management in particular by different individuals and institutions involved in the management and/or study of water resources. In addition, various terms such as hydrosystems management, integrated water management, integrated regional water management, water resources management, river basin management, watershed management, total water management, and so on have been used to refer to the management of water resources in conjunction with other resources on a large scale, that is, on a river basin or watershed scale. For the purposes of this paper, the term integrated hydrosystems management is consistently used unless otherwise specified.

This paper reviews the concept of integrated hydrosystems management. The definition of integrated water management as used by various institutions and individuals are cited and an attempt is made to give a definition that considers the wide range of aspects of integrated hydrosystems management. The evolution of simulation models and the structure of optimization models for hydrosystems problems are revisited. Examples of a relatively new set of computer models, generally termed as decision support systems (DSS), for hydrosystems problems are reviewed. These models, being capable of incorporating water policies, are believed to be promising computing methodologies for integrated hydrosystems management. Some of the examples of DSS given for integrated hydrosystems management manifest the possibility of incorporating or at least monitoring water policy issues in the process of allocating water to all the competing users.

2. Integrated Hydrosystems Management

2.1. DEFINITION

Mitchell (1990) noted that integrated water management may be contemplated in at least three ways: 1) the systematic consideration of the various dimensions of water: surface and groundwater, quality and quantity; 2) the implication that while water is a system it is also a component which interacts with other systems; and 3) the interrelationships between water and social and economic development. In the first thought, the concern is the acceptance that water comprises an ecological system which is formed by a number of interdependent components. In the second one, the interactions between water, land and the environment, which involve both terrestrial and aquatic issues, are addressed. Finally, the concern is with the relationships between water and social and economic development, since availability or lack of water may be viewed as an opportunity for or a barrier against economic development.

The provision of water resources management include: providing ports, harbors, and usable channels for water transport; supplying water and electricity for cities, industry and agriculture; providing flood control for cities; and cleaning up visibly polluted rivers and lakes (Hall, 1998). Jamieson and Fedra (1996) also indicate that river basin management includes all aspects such as water supply, land drainage, hydropower generation, effluent disposal, recreation and amenity.

Each aspect of integrated hydrosystems management depends on and is affected by other aspects. Loucks (1996) points out "Integrated water resources systems planning and management focuses not only on the performance of individual components, but also on the performance of the entire system of components".

Water policy issues, of which limited effort was made in the past to incorporate into hydrosystems models, are some of the major factors that affect integrated hydrosystems management. Grigg (1998) describes water policy as dealing with finding satisfactory ways to allocate resources to balance between diverse and competing objectives of society and the environment. He refers to "integrated water management" as blending together actions and objectives favored by different players to achieve the best total result. Mitchell (1990) states that integration in water management deals with "… problems that cut across elements of the hydrological cycle, that transcend the boundaries among water, land and environment, and that interrelate water with broader policy questions associated with regional economic development and environmental management". The policies that are needed for integrated water resources management require coordination and collaboration among governments and agencies engaged in water management (Viessman, Jr., 1998). Grigg (1998) notes that improving coordination is the most promising route to the conceptual and perhaps utopian vision of integrated water management.

AACM (a consulting company in Australia) and Center for Water Policy Research, Australia, in 1995 defined integrated resource management (of which water resources is a part) as the coordinated management of land and water resources within the region, with the objectives of controlling and/or conserving the water resource, ensuring biodiversity, minimizing land degradation, and achieving specified and agreed land and water management and social objectives (Hooper, 1995). This definition is also appealing to water resources which is just a component of the resources of a watershed.

The American Water Works Association Research Foundation (AWWARF) (1996) defined the concept of total water management which comprehends wide aspects of integrated hydrosystems management through the following statements.

"Total Water Management is the exercise of stewardship of water

resources for the greatest good of society and the environment.

A basic principle of Total Water Management is that the supply

is renewable, but limited, and should be managed on a sustainable

use basis. Taking into consideration local and regional variations,

Total Water Management:

Encourages planning and management on a natural water systems basis through a dynamic process that adapts to changing conditions;
Balances competing uses of water through efficient allocation that addresses social values, cost effectiveness, and environmental benefits and costs;
Requires the participation of all units of government and stakeholders in decision-making through a process of coordination and conflict resolution;
Promotes water conservation, reuse, source protection, and supply development to enhance water quality and quantity; and
Fosters public health, safety, and community good will."

Table 1 shows an elaboration by Grigg (1998) of the definition of total water management as related to the concept of coordination. He emphasized on what is implied by each of the important phrases used in the definition. These phrases which are apparently the central aspects of integrated hydrosystems management include society and environment, stakeholder, watershed and natural water systems, means of water management, time-wise, intergovernmental, water quality and quantity, local and regional concerns and competing uses. Integrated hydrosystems management is as much challenging as compromising between these different aspects in making decisions.

The foregoing definitions and discussions indicate that integrated hydrosystems management is multi-objective. It is necessary both for economic efficiency (which is measured in monetary units) and for environmental quality (which is measured in terms of pollutant concentration). Shortly, it balances between societal welfare and ecosystem sustainability. To summarize, integrated hydrosystems management in a watershed involves a multi-disciplinary approach of developing and using water resources by making possible balances between all the competing water uses and through coordination between all parties without causing detrimental consequences to the ecosystem and/or future requirements.

Table 1. Types of coordination from total water management definition (Grigg, 1998)

Type of coordination	Phrase from Total Water Management definition	Discussion	Effective-ness Ranking
Society and environment	The exercise of stewardship of water resources for the greatest good of society and the environment	This statement provides a general organizing framework for balancing. It is adequately understood, but needs more explanation.	1
Stakeholder	Requires the participation of all… stakeholders in decision-making through a process of coordination and conflict resolution	Process is known as stakeholder and public involvement. Good and improving. A central issue of democratic government.	2
Watershed and natural water systems	Encourages planning and management on a natural water systems basis	It is recognized and currently popular that water management on a basin or watershed basis is desirable. Further progress will require more effort.	3
Means of water management	Promotes water conservation, reuse, source protection, and supply development	This means to coordinate different ways to meet needs and sustaining the environment. A central planning and management issue.	4
Time-wise	Through a dynamic process that adapts to changing conditions	This requires valid planning methods to preserve institutional memory and keep processes on track and requires much improvement.	5
Intergovernmental	Requires the participation of all units of government … in decision-making through a process of coordination and conflict resolution	Intergovernmental coordination is given as separate from stakeholders because of the different kinds of authorities that government has.	6
Water quality and quantity	To enhance water quality and quantity	This is handled through water quality law and regulation. Many problems still require solution.	7
Local and regional concerns	Taking into consideration local and regional variation	This is a difficult issue requiring intergovernmental cooperation in arenas which lack adequate incentives and often can not be mandated. It is not working too well.	8
Competing uses	Balances competing uses of water through efficient allocation that addresses social values cost effectiveness, and environmental benefits and costs	This is handled through state and federal water law regulations, court decisions, and other institutions. A very difficult arena.	9

2.2. HISTORY

The history of integrated hydrosystems management is, perhaps, not as clear as we would wish it to be. Jamieson and Fedra (1996) report that the concept of integrated hydrosystems management has been recognized by practitioners since the early 1970s. This perception was endorsed by the United Nations in the Dublin Statement in 1992.

The history of integrated hydrosystems management on a regional basis is even less clear, because the definition of a region is often ambiguous. River basin boundaries usually differ from political boundaries. Groundwater flow has obviously never been dictated by political boundaries, and neither has the movement of atmospheric water. Furthermore, the question of the size of a region has been a challenge and will probably remain so in the near future. Viessman, Jr., (1998) states that it is not clear that integrated regional water plans can be fitted within the geographic limits of large river basins or watersheds. Vlachos (1998) poses a very important question: Can integrated planning and management work in the vast expanses of the Nile, the Amazon, the Parana/LaPlata, or should it be restricted to more regional, specific socio-political conflicts of rather well-defined geographic, cultural, environmental, physiographic, and economic boundaries?

Defining a water resources region now appears to be driven more by the watershed approach than the other factors mentioned above. A national forum convened in January 1994 by the Conservation Fund and the National Geographic Society clearly recognized the critical need for the watershed approach for integrated hydrosystems management rather than political jurisdiction or boundaries. Similarly, the Environmental Advisory Board (EAB) of the US Army Corps of Engineers (USACE) recommended in 1994 to use the watershed/ecosystem approach as the holistic, integrated concept on which to base (water resources) planning (Bulkley, 1995). Furthermore, the US General Accounting Office (1994) listed the importance of the watershed approach for integrated management. Accordingly, watershed boundaries:

are relatively well defined;
can have major ecological importance;
are systematically related to one another hierarchically and thus include smaller ecosystems;
are already used in some water management efforts; and
are easily understood by the public.

Many water resources projects in the past lacked the integrated planning aspect. Hall (1998) states that throughout history, water management "systems" have been developed in a linear fashion, i.e., it had a piecemeal development in which the components of integrated water management were put into place as the need for the component arose. As a result, these systems have not been sufficient and effective enough.

2.3. IMPORTANCE

We are becoming more increasingly aware, with time, of the fact that our water supplies are limited both in quantity and quality. Because water has multiple and often competing uses, hydrosystems are interrelated with other physical and socio-economic systems. In some locations, when water supplies become extremely limited, its further use is based on the determination of which user has the oldest "right" to it, or on a judgment about which uses have the highest priority (Hall, 1998). He also warns that unless dealt with appropriately, the forces of population growth, urbanization and increased water demands for home, industry and agriculture, coupled with an increasingly global economy and culture, will produce in the future spreading, perilous degradation of water quality everywhere, and a continuously widening gap between water needs and the availability of useful water in all too many locations. As a solution to this problem, he suggested a different approach which includes: 1) management across political boundaries, 2) the collective management of atmospheric water, surface waters and groundwater, and 3) the combined management of water quality and water quantity.

Schultz (1998) brings into picture what the criteria for water resources management projects at present are and those criteria emerging as new ones in the future. Accordingly, the factors that have to be satisfied include: 1) economic benefits; 2) technical efficiency; and 3) performance reliability. The approach which seems to become more and more dominant includes:

the principle of sustainable development;
ecological quality;
consideration of macroscale systems and effects; and
planning in view of changes in natural and socioeconomic system.

It is evident from these comparisons that hydrosystems projects are geared towards integrated management.

In a different argument, an integrated hydrosystems project needs to be evaluated on the following important factors: technical, economic, financial, environmental and socio-political. Technically, it must be feasible to build; economically, it must be reasonably affordable; financially, it must have source; environmentally, its effect must be mitigated with ease; and socio-politically, it must be acceptable to the public. The project can be successful if effective coordination prevails between the parties involved and if such parties are mandated to monitor clearly defined scope and regional coverage.

Therefore integrated hydrosystems management is found to be a viable approach in planning efficient water resources projects. Integrated hydrosystems management makes it even easier and more efficient for such projects to succeed. In England and Wales, for example, regional water authorities whose boundaries were defined by the watersheds of the country enabled the replacement of 1600 separate water service entities with ten regional watersheds (Bulkley, 1995).

3. Computer Modeling Tools for Integrated Hydrosystems Management

If the ideals of integrated hydrosystems management can be sought after, analytical tools become essential to simplify or assist in the balancing out process. Water policies need to be transformed into such forms that can be "understood" and "interpreted" using analytical tools such as computer models. Consequently, robust computer models that not only solve the problems that have analytical structure or mathematical formula but also capable of reducing and incorporating water policies into the analytical structure are required. Furthermore, these models may be required to interpret the result of the computations, give conclusions based on the result and make appropriate recommendations based on the conclusions reached.

A review of the computer models for solving hydrosystems problems show that although tremendous work has been done in the past to develop such models, only a few models exist that address the overall framework of problems associated with integrated hydrosystems management. A few of the reasons may be attributable, among others, to:

the lack of clear definition and better understanding of integrated hydrosystems management;
the variation of water needs with space and time; and
the evolution (revolution) of computer programming.

Most of the existing hydrosystems simulation models solve problems that can be readily expressed in a form of mathematical functions. Similarly, hydrosystems optimization models search for optimal solutions of problems defined by mathematical functions. To use such models for integrated hydrosystems problems, they must also have the capability of considering different water policies and incorporating them into the solution.

Computer modeling approaches that at least partly tried to address some of the concepts of integrated hydrosystems management are highly based on interfacing simple computer models programmed and used for the analysis of specific hydrosystems problems. At the core of some advanced computer models used for integrated hydrosystems management lie simple simulation modules, rule-based simulation modules (also known sometimes as expert systems) and optimization modules of hydrosystems problems. While many simulation and optimization modules have been developed and interfaced over the years by different institutions and agencies, the incorporation of rule-based simulation modules in computer models for integrated hydrosystems management appears to have emerged as a sound approach recently. By incorporating rule-based simulation modules, it has become easier to manage decisions that involve several factors and water policies.

The following section discusses some of the computer models that emerged in the US over the past few decades for the simulation of various types of hydrosystems problems. Real time event hydrologic models are discussed in this Section and subsection 3.2 discusses the basic mathematical structure of optimization models, which may be viewed as generic functions that can be customized to specific hydrosystems problems.

3.1. SIMULATION

3.1.1. Development of Hydrosystems Simulation Models

In the advancement of information technology, hydrosystems simulation models have generally gone through an evolutionary process. Figure 1 depicts the evolution of hydrosystems models as classified into five generations (derived from the explanation given by Jamieson and Fedra, 1996). The first generation codes (models) which tremendously simplified calculation of analytical functions through generic computer codes are but mediocre by today’s standards. One may draw an analogy between the coming into being of these codes and the transition of computation methods from using the slide rule to scientific calculators. In both cases, similar jobs are done but the new method highly reduced the time required for numerical computations. The succeeding generations of models successively enhanced the robustness of the models and/or the ease with which the model can be used. The fifth generation of models are embodied with artificial intelligence that not only perform analytical computations but also draw some preliminary conclusions and recommend appropriate actions.

3.1.2. Taxonomy of Hydrosystems Simulation Models

Over the past few decades, water resources professionals have witnessed the development of quite a number of hydrosystems simulation models. Wurbs (1995) points out that a tremendous amount of work has been accomplished during the past three decades in developing computer models for use in water resources planning and management. The majority of these models, perhaps most of the earliest computer models to be developed for water resources problems, may be viewed as simulation models. A taxonomy of some of the popular hydrosystems simulation models in the US are summarized in Table 2.

Table 2. Taxonomy of some of the most popular hydrosystems simulation models in the US

1. Surface water systems
Model name	Developed by	Model purpose	Remarks
a) Watershed runoff system HEC-1 HEC-HMS TR-20 HYMO A & M Watershed Model WMS	US Army Corps of Engineers Hydrologic Engineering Center (HEC) HEC US Department of Agriculture Soil Conservation Service (SCS) and Agricultural Research Service US Department of Agriculture Agricultural Research Service and Texas A & M University USACE Waterways Experiment Station Brigham Young University	Precipitation- runoff processes Precipitation- runoff processes Precipitation-runoff processes Precipitation-runoff processes Precipitation-runoff processes Precipitation-runoff processes	Streamflow hydrographs at desired locations in the river basin are computed. Part of the Next Generation (NexGen) models developed by the HEC. Surpasses HEC-1. New capabilities include a linear distributed transformation that can be applied with grid (e.g., radar) rainfall data, optimization options, and so on. Uses the SCS curve number method and SCS curvilinear dimensionless unit hydrograph to develop the runoff response. Includes option to compute watershed sediment yields using a modified version of the universal soil loss equation. Accepts radar readings as well as conventional gauged rainfall data. Capabilities also include standard step method water surface profile computation. Automatically delineates watershed boundaries using TINs.
b) Streamflow systems HEC-2 WSPRO HEC-RAS	HEC US Geological Survey (USGS) HEC	Water surface profile in rivers Water surface profile in rivers Water surface profile in rivers	Computes water surface profile for gradually varied flow. Uses the standard step method solution of the energy equation. Part of the NexGen models. Surpasses HEC-2. Current version performs one dimensional steady state flow; future versions will perform unsteady flow and sediment transport calculations.

Table 2. Cont’d.
FLDWAV UNET FESWMS-2DH	Hydrologic Research Laboratory of the National Weather Service R. L. Barkau USGS, Water Resources Division, for Federal Highway Administration (FHWA)	Dynamic routing of flood One dimensional unsteady open channel flow Two-dimensional river flow	FLDWAV combines the capabilities of DWOPER and DAMBRK models which are one dimensional unsteady flow models based on an implicit finite difference solution of the St. Venant equations. Used for unsteady flow through a full network of open channels with external or internal boundary conditions. Based up on RMA-2 model which is a finite element model used for either steady or unsteady flow.
2. Ground-water systems
MODFLOW UN Groundwater Software Package (GW1 - GW11) PLASM WHPA SUTRA	USGS UN Department of Technical Cooperation for Development, Natural Resources and Energy Division Illinois State Water Survey EPA USGS	Simulation of two- or three-dimensional saturated flow Varies; depends on which model is used Simulation of two dimensional unsteady flow Delineation of Wellhead Protection Areas, defined by the Safe Water Drinking Act (1986) Fluid movement and solute and energy transport	Three dimensional, finite difference groundwater model. Each model in the packet solves a specific groundwater flow problem. Has capabilities for simulating two-dimensional unsteady flow in hetrogeneous anisotropic aquifers under water table, nonleaky and leaky artesian conditions. Delineates capture zones and contaminant fronts assuming steady-state horizontal flow in the aquifer. Consists of four particle tracking modules. Can be used to analyze groundwater contaminant transport and aquifer restoration problems.
3. Surface-ground water systems
MODBRANCH	USGS	Combining surface and groundwater flow	Formed by coupling together two simulation models: MODFLOW-96 (latter version of MODFLOW) and BRANCH (a steady and unsteady surface water flow model).
Table 2. Cont’d.
4. Storm water systems
SWMM STORM	Metacalf and Eddy, Inc., University of Florida and Water Resources Engineers under the auspices of EPA HEC	Simulation of urban runoff quantity/quality Simulation of storage, treatment, overflow and runoff	Can simulate hydrographs and pollutographs which can be used as input to river and reservoir water quality models. Can simulate the interations of rainfall/snowmelt, runoff, dry-weather flow, pollutant accumulation and washoff, land surface erosion, treatment and detention storage. Water quality parameters include suspended and settleable solids, biochemical oxygen demand, total nitrogen, orthophosphate, and total coliform.
5. Water distribution/quality
EPANET KYPIPE2/ KYQUAL QUAL2E WQRRS	U.S. Environmental Protection Agency University of Kentucky Texas Water Development Board HEC	Water quality and hydraulics in water distribution Flow and water quality in pipe networks Water quality Water quality for river-reservoir systems	Performs extended period simulation of hydraulic and water quality conditions. In addition, water age, source tracing and chlorine decay can be simulated. Consists of several packages for different purposes. Simulates both steady state flows and extended period simulation along with water quality in pipe distribution networks. Allows simulation of 15 water quality constituents, including dissolved oxygen, biochemical oxygen demand, temperature, organic nitrogen, and so on. A package of three programs: Stream Hydraulics Package (SHP), Stream Water Quality (WQRRSQ) and Reservoir Water Quality (WQRRSR).
6. Bay/Estuary Systems
SHARP	USGS	Freshwater-saltwater flow	A quasi-three dimensional, finite difference models that simulates freshwater and saltwater flow in layered coastal aquifer systems.
7. Flood Mitigation/Forecasting Systems
HEC-FDA	HEC	Flood damage reduction analysis	Part of the Next Generation (NexGen) models developed by the HEC. Performs plan formulation and evaluation for flood damage reduction studies.

Some of the earliest simulation models included in Table 2 such as HEC-1 and TR-20 are lumped parameter hydrologic rainfall-runoff models. These models, which were developed in the late 60’s and early 70’s, continue to be the accepted standards. There have been many advances in the distributed watershed modeling over the past several years that now permit the more comprehensive and sophisticated distributed modeling. The development of collection and management of overwhelming data required to derive these models have been made easier with the emergence of more user friendly software and geographic information systems (GIS).

The Watershed Modeling System (WMS, formerly known as GeoShed) developed at Brigham Young University (Nelson, et al., 1995) is a graphically based software tool with an interface to HEC-1 and an interface to CASC2D, a two-dimensional, grid-based, distributed hydrologic model. In addition, features include triangulated irregular network (TIN) generator from scattered and digital elevation model data source, automated watershed and sub-basin delineation from TINs. CASC2D, developed through the U.S. Army Corps of Engineers, is a physically based rainfall/runoff model which uses rectangular grid cells to represent the distributed watershed and rainfall domain (Julien, et al., 1995). This model uses a two-dimensional diffusive wave equation to simulate overland flow and a one-dimensional diffusive wave equation to simulate channel flow.

3.1.3. Real-time Rainfall Runoff Analysis Using GIS and Radar Data

Watershed rainfall-runoff computation requires determination of the general hydrologic processes within the watershed. This, in turn, requires not only the topographic information of the watershed but also information about other hydrologic variables such as the temporal and spatial distribution of precipitation. Use of GIS has made it possible to represent spatial distribution of elevations using Digital Elevation Models (DEM). Three principal methods are available in most GIS models for structuring a network of elevation data: 1) square-grid networks; 2) contour-based networks; and 3) triangulated irregular networks (TIN) (Moore, et al., 1991).

Precipitation data can be obtained by means of remote sensing such as radar at desirable time intervals so that real-time runoff (flood) simulation can be performed. Using the DEM data (available for the entire United States from the USGS), GIS can compute the aspect (direction of maximum slope) at a given location within the watershed. With other hydrologic parameters for abstraction, infiltration, routing and so on available in GIS or other database systems, the watershed runoff processes can be easily simulated. In effect, this approach can be used to forecast flood events at desired locations on a real-time basis provided that instantaneous rainfall data can be directly obtained using radar or other means. Figure 2 shows a general procedure that can be used for modeling a general real-time operation (adapted from Loucks, 1996).

The WMS discussed in Section 3.1.2 is an advanced model used for a more comprehensive watershed modeling system. This model incorporates digital terrain modeling, GIS data, and analytical hydrologic models in a single environment. It has the capabilities of automatically delineating watershed and sub basin boundaries from TIN and then computing geometric parameters such as area, slope and runoff distances

for each basin. Figure 3 shows the representation of a watershed by grids for which different data can be stored in GIS. WMS can determine different parameters of the watershed from the stored grid data. HEC-1 is directly interfaced in WMS for performing rainfall/runoff analysis (Nelson, et al., 1995).

As shown in the WMS interface in Figure 4, runoff hydrographs at desirable locations can be computed and viewed. This can be a very useful tool especially in dealing with flood mitigation efforts. If one or more detention facilities exist within the watershed, it may be possible to adjust release policies on a real time basis such that threatening flood peaks can be reduced.

Fig 4. WMS interface

3.1.4. Real-time Flood Management Model for the Lower Colorado River Authority

Developed at the University of Texas at Austin by Unver, et al. (1987) for the Lower Colorado River Authority (LCRA), this model can be used for flood routing and rainfall-runoff modeling on a real-time framework. It has several modules that interact with one another. Real-time data that are managed by the data management module of this model include rainfall collected at recording gages, streamflow collected at automated stations, headwater and tailwater elevations at each dam, information on which rivers and reservoirs are to be simulated in flood routing, and current reservoir operations. The model’s subsystems constitute the three basic subsystems of a DSS. Figure 5 depicts the structure of the model as given by the LCRA.

3.2. OPTIMIZATION FORMULATIONS

Various optimization techniques in general and their application to various hydrosystems problems in particular have shown remarkable progress over the past three decades. The progress of the application of these techniques has gone alongside with the revolution of computer models and as such similar explanations can be given to the development of simulation models and optimization techniques over the past three or more decades. Figure 6 gives the development of the application of optimization techniques to hydrosystems problems, in an analogy that is similar to Figure 1, which was given for simulation models.

The general formulation for optimization problems in water resources can be expressed in terms of state (or dependent) variables (x) and control (or independent) variables (u) as (Mays, 1997; Mays and Tung, 1992)

Optimize f(x, u) (1)

subject to process simulation equations

G(x, u) = 0 (2)

and additional constraints for operation on the dependent (u) and independent (x) variables

(3)

The term Optimize in Eq. (1) refers to either maximization or minimization whereas the constraint equations (Eqs. (3)) dictate the feasibility of the objective with respect to each and all of the constraints. In other words, the solution to the simulation equations (Eqs. (2)) must satisfy the constraints defined by Eqs. (3). The process simulation equations basically consist of the governing physical equations of mass, energy and momentum.

Many hydrosystems problems can be formulated as discrete-time-optimal control problems. The basic mathematical definition of a discrete-time-optimal control problem is stated as

(4)

subject to

, t = 1, 2, … T. (5)

. (6)

where is the vector of the state variables at time t, is the vector of the control variables at time t, and T is the number of decision times.

A few possible optimization formulations for different hydrosystems problems are given below.

3.2.1. Groundwater Management Subsystems

The general groundwater management problem can be expressed mathematically as (Mays, 1997)

Optimize Z = f(h, q) (7)

subject to

G(h, q, c) = 0 (8)

(9)

(10)

(11)

w(h, u) 0 (12)

where h and q in the objective function are vectors of heads and pumpages (or recharges), respectively. C is a parameter that measures quality such as chlorine content and so on. Eqs. (8) are the general groundwater flow constraints, which represent a system of equations governing groundwater flow and transport. Eqs. (9) and (10) represent, respectively, the upper and the lower bounds on the pumpages (recharges) and on the heads. Eqs. (11) are the ground water quality constraints whereas Eqs. (12) may be taken as additional constraints which can be included to impose restrictions such as water demands, operating rules, budgetary restrictions and so on. It may be noted here that the lower and upper bounds on pumpages may or may not exist whereas those on the head can be the bottom elevation of the aquifer and the groundwater surface elevations for the unconfined cells respectively.

3.2.2. Real-time Operation of River-Reservoir Systems for Flood Control

Mays (1997) states the optimization problem for the real-time operation of multireservoir systems under flooding conditions as

Minimize Z = f(h, q) (13)

subject to

G(h, Q, r) = 0 (14)

(15)

(16)

(17)

w(r) 0 (18)

where h and Q are the vectors of water surface elevations and discharges, respectively. Eqs. (14) are the hydraulic constraints defined by the Saint-Venant equations for one-dimensional gradually varied flow and other boundary conditions. Eqs (15) - (17) define the bounds on the discharges (reservoir releases), the elevations of the water surface and the physical and operational bounds on the spillway gate operations, r being the fraction of the gate opening. Eqs. (18) are other constraints such as operating rules, target storage, storage capacities, and so on.

The objective of the optimization in this case can be to minimize (a) the total flood damages, (b) deviations from target levels, (c) water surface elevations in the flood areas, or (d) spills from reservoirs or maximizing storage in the reservoirs.

3.2.3. Reservoir System Operation for Water Supply

The optimization for this kind of hydrosystems problem can be expressed as (Mays, 1997)

Maximize Benefits = (19)

subject to

, t = 0, …, T - 1 (20)

, t = 1, …, T (21)

, t = 1, …, T (22) , t = 1, …, T (23)

, t = 1, …, T (24)

(25)

where S_t and U_t are the vectors of reservoir storage and releases and t represents discrete time period. Eqs. (20) define the system of equations of conservation of mass for the reservoirs and river reaches. and are respectively the vectors of reservoir storage at the beginning of time period t + 1 and t, is the vector of hydrologic inputs and is the vector of reservoir losses. Eqs. (21) and (22) define the bound constraints on reservoir releases and storage respectively. Eqs. (23) and (24) define the bound constraints on reservoir storage in probabilistic form where P[ ] denotes the probability and and represent the minimum and the maximum reliability or tolerance levels. Eqs. (25) express the other constraints on reservoir operation.

3.2.4. Water Distribution System Operation

Mays (1997) defines the optimization problem for water distribution system operation in terms of the nodal pressure heads, H, pipe flows, Q, tank water surface elevations, E, pump operating times, D, and water quality parameter, C, as follows.

Minimize energy costs = f(H, Q, D) (26)

subject to

G(H, Q, D, E, c) = 0 (27)

w(E) = 0 (28)

(29)

(30)

(31)

(32)

where Eqs. (27) and (28) express the energy and flow constraints and the pump operation constraints. The remaining equations express the bound constraints on the nodal pressure head, Eqs. (29), pump operating times, Eqs. (30), tank water surface elevations, Eqs. (31), and water quality, Eqs. (32).

3.2.5. Freshwater Inflows to Bays and Estuaries

The optimization problem is to minimize freshwater inflows, or to maximize harvest or both, expressed mathematically as

Optimize Z = f(Q, s, H) (33)

subject to

G(Q, s) = 0 (34)

h(Q, s) = 0 (35)

(36)

(37)

where Q is inflow to an estuary, s is the salinity of the estuary and H is the fish harvest. Eqs. (34) are the hydrodynamic transport equations that relate the salinity at a given point in an estuary to inflow whereas Eqs. (35) are regression equations that relate inflow to fish harvest. The last two equations are the bound constraints that define the limitations on freshwater inflows and salinity.

3.3. INTERFACING OPTIMIZATION AND SIMULATION MODELS

The general form of the objective functions and the constraints in hydrosystems problems including the foregoing examples can be linear, non-linear or differential equations. Each of such equations needs different approaches for solution. Several computer codes have been written for each of these types of formulations.

For those hydrosystems optimization problems which involve solving general governing differential equations of mass, energy and momentum (as is the case with most of the above formulations), the approach used can be solving the optimization problem directly by embedding finite differences or finite element equations of the governing process equations (Mays, 1997). This approach is relatively tedious to apply to real world problems. Alternatively, an appropriate process simulator can be used to solve the constraints process simulation equations when they need to be evaluated for the optimizer. Consequently, the following general and simpler optimization problem can be used.

Minimize F(u) = f(x(u), u) (38)

subject to

(39)

Different techniques have been successfully applied to solve optimization problems that are formulated in the above form. The most common techniques are given below.

3.3.1. Mathematical Programming

Table 3. Summary of some of the most popular optimization models in the U.S.

Model name

Developed by

Model purpose

Remarks

LINDO

Lindo Systems, Inc.

Solves linear, quadratic and integer programming problems

A user friendly Linear Interactive and Discrete Optimizer (hence, the name LINDO).

LINGO

Lingo Allegro USA, Inc.

Solves linear and nonlinear programming problems

A sophisticated matrix generator; helps the user create large constraints objective function terms by writing one line code.

GRG2

Univ. of Texas

Solves nonlinear programming problems

Uses the generalized reduced gradient algorithm to find the optimal solution.

GINO

Solves nonlinear programming problems

This model is a microcomputer version of GRG2.

GAMS

GAMS Development Corporation

Solves linear programming problems

MINOS

Saunders and Murthagh

Solves linear and nonlinear programming problems

Uses different algorithms when the problem has linear objective function and constraints, nonlinear objective function and linear constraints, and nonlinear objective function and constraints.

GAMS/ZOOM

Solves mixed integer programming problems

Adapted ZOOM (Zero/One Optimization Method).

GAMS/MINOS

Solves linear and nonlinear programming problems

Adapted MINOS (Modular In-Core Nonlinear Optimization System).

A modified form of DDP, known as Successive Approximation Linear Quadratic Regulator (SALQR), has been used for optimization problems in which nonlinear simulation equations are made linear in the optimization step (Culver and Shoemaker, 1992).

Example applications of DDP have been made by Carriaga and Mays (1995) to reservoir release optimization to control sedimentation, and SALQR to operation of multiple reservoir systems to control sedimentation in alluvial river networks by Nicklow and Mays (1998); to operate soil aquifer treatment systems by Tang, et al. (1999); and to optimal freshwater inflows to bays and estuaries by Li and Mays (1995)

3.3.3. Genetic Algorithms and Simulated Annealing

Genetic Algorithms (GA). Genetic algorithms are non-conventional search techniques patterned after the biological processes of natural selection and evolution (Tang and Mays, 1999). GA can be useful for the selection of parameters to optimize the performance of a system and for testing and fitting quantitative models (Chambers, 1995). Every solution of the optimization problem is represented in the form of a string of bits (integers or characters) that consist of the same number of elements, say n. Each candidate solution represented as a string is known as an organism or a chromosome. The variable in a position on the chromosome and its value in the chromosome are called the gene and the allele, respectively. For example, if n = 3, a general chromosome is x = (x₁, x₂, x₃) where x₁, x₂, and x₃ are the genes on this chromosome in the three positions (Murthy, 1995).

Genetic algorithms for optimization problems are developed by first transforming the problem into an unconstrained optimization problem so that every string of length n can be looked upon as a solution vector for the problem (Murthy, 1995). Five tasks are required in the performance of a GA to solve the optimization problem: encoding, initialization of the population, fitness evaluation, evolution performance and working parameters (Adeli and Hung, 1995).

The decision variable vector is encoded as a chromosome using mostly binary number coding method. Therefore if there are m decision variables and if each decision variable is encoded as an n-digit binary number, then a chromosome is a string of n x m binary digits as shown in Figure 7.

A population of chromosomes is initialized which require randomly generating the initial population in such a way that all values for each bit have equal probability of being selected. The fitness measure at every feasible solution is equal to the objective function value at that point. Thus, fitness evaluation is used to determine the probability that a chromosome will be selected as a parent chromosome to generate new chromosomes. Evolution performance involves selection, crossover and mutation. Selection chooses the chromosome to survive for a new generation. Crossover is used to recombine two chromosomes (parent strings) and generate two new chromosomes (offspring strings) based on a predefined crossover criterion. Mutation serves as an operator to reintroduce "lost alleles" into the population based on a predefined mutation criterion. Working parameters guide the genetic algorithm and include chromosome length, population size, crossover rate, mutation rate and stopping criterion.

Simulated Annealing (SA). SA stems from an algorithm that is used for the application of statistical thermodynamics concepts to combinatorial optimization problems. A solution to a combinatorial optimization problem is based on a statistical mechanics in which the best solution is obtained from a large set of feasible solutions.

In essence, it is a type of local search (descent method) heuristic that starts with an initial solution and has a mechanism for generating a neighbor of the current solution. For minimization problems, if the generated neighbor has a smaller objective value, it becomes the new current solution, otherwise the current solution is retained. The process is repeated until a solution is reached with no possibility of improvement in the neighborhood (Murty, 1995).

This algorithm has the disadvantage that the local search stops at a local minimum (see Figure 8). This can be avoided by running the local search several times starting randomly from different initial solutions. By doing so, the global minimum can be taken as the best of the local minima found.

A better approach to find the global minimum was introduced in 1953 by Metropolis et al. (Murty, 1995). In this attempt, annealing was applied to the search of minimum energy configuration of a system after the system is melted. At each iteration, the system is given a small displacement and the change in the energy of the system, d , is calculated. If d < 0, the change in the system is accepted; otherwise, the change is accepted with probability exp (-d /T) where T is a constant times the temperature.

This optimization technique has been applied to different problems in engineering, such as groundwater restoration (Skaggs and Mays, 1999), operation of water distribution systems (Sakarya, and Mays, 1999; Goldman and Mays, 1999), for water quality purposes (Sakarya, et al., 1998).

3.3.4. Comparison of Heuristic Search Methods (GA and SA) to Other Optimization Techniques

Whereas the heuristic search methods involve trial solutions, mathematical programming and DDP/SALQR follow some given procedures. On the other hand, mathematical programming and DDP/SALQR require derivative information. The optimal solution found by mathematical programming approach may result in a very short operating time during one time interval that can not be followed for practical purposes. In the simulated annealing approach, this problem can be minimized by setting minimum period of operation (Sakarya, et al., 1998).

The mathematical programming approaches find the optimum solution in much shorter operating times than the heuristic search approaches. Tang and Mays (1999) have developed a new methodology for the operation of soil aquifer treatment systems, formulated as a discrete-time optimal control problem. This new methodology is based upon solving the operations problem using a genetic algorithm interfaced with the one-dimensional unsaturated flow model HYDRUS (Kool and van Genuchten, 1991). The same problem has been solved by Tang, et al. (1996) using SALQR interfaced with the HYDRUS model. The computer time for a ten cycle operation with the SALQR algorithm was reported as 654 CPU seconds, while with the genetic algorithm, it needed about 46600 CPU seconds (about 13 hours) on the same computer to obtain the optimal solution for a three cycle operation.

Sakarya, et al. (1998) have compared two newly developed methodologies, a mathematical programming approach and a simulated annealing approach, for determining the optimal operation of water distribution system considering both quantity and quality aspects. Both methodologies formulate the problem as a discrete-time optimal control problem. The mathematical programming approach interfaces the GRG2 model (Lasdon and Warren, 1986), a generalized reduced gradient procedure, with the U.S. Environmental Protection Agency EPANET model (Rossman, 1994) for water distribution system analysis. The simulated annealing approach is also interfaced with the EPANET model. The study showed that while different results were obtained for total pump operation hours, the total 24 hr energy costs were comparable.

3.4. COMPUTER BASED INFORMATION SYSTEMS

3.4.1. Supervisory Control Automated Data Acquisition (SCADA)

SCADA is a computer-based system that can control and monitor several hydrosystems operations such as pumping, storage, distribution, wastewater treatment and so on. Several such systems have been developed in the past for different water supply agencies. For instance, the Metropolitan Sewer District of Cincinnati planned to integrate a SCADA system in the 1980s to monitor its wastewater treatment plants and pump stations. This system was planned for an area which consisted of seven major treatment plants, 30 package wastewater plants serving individual subdivisions and about 130 pump stations (Clement, 1996). A SCADA system developed in 1986 for Honolulu, Hawaii, had the capability of controlling and monitoring 57 source pumping stations, 126 storage reservoirs, and 73 booster pumping stations (Wada, et al., 1986). In general, SCADA systems are designed to perform the following functions:

acquire data from remote pump stations and reservoirs and send supervisory controls;

allow operators to monitor and control water systems from computer controlled consoles at one central location;

provide various types of displays of water system data using symbolic, bar graph, and trend formats;

collect and tabulate data and generate reports; and

run water control software to reduce electrical power costs.

Remote terminal units (RTUs) are used to process data from remote sensors at pump stations and reservoirs. The processed data are transmitted to the SCADA system also by the RTUs. Conversely, supervisory control commands from the SCADA system prompt the RTUs to turn pumps on and off and open and close valves.

3.4.2. Geographic Information System (GIS)

All hydrologic processes relate to space making it plausible to associate geo-information with hydrologic processes. Survey of some of the recent literature shows several attempts that have been made to incorporate GIS into hydrologic analyses. Greene and Cruise (1995) classify these attempts into four groups: 1) calculation of input parameters for existing hydrologic models; 2) mapping and display of hydrologic variables; 3) watershed surface representation; and 4) identification of hydrologic response units. Since several GIS database layers can be overlain, GIS can be a very useful tool to integrate the analyses of hydrologic processes of watersheds.

The study by Greene and Cruise (1995) formed a GIS database of such hydrologic/hydraulic variables as storm water inlet locations, soil moisture characteristics of layered soils, etc. to determine the discharge hydrograph at desired outlet points. The results obtained from this analysis showed reasonable accuracy.

3.4.3. GIS as a Tool for Flood Damage Analysis

Buffering applications in GIS – delineating the area in a river system that is affected by a flood of certain magnitude – help to perform sensitivity analysis to the risk from flooding. This can be done in two major ways. First, a series of "what if" questions can be analyzed before the flooding occurs. Putting in various flood levels and analyzing can help forecast the associated damages thereby assisting the management body to make better decisions before the flood occurs. Second, if landscape coverage is readily available in a GIS database, the effect of the disaster from a flood event can be analyzed very quickly, thus permitting the management body to respond rapidly. Such analyses can save lives and property (Davis, 1996). Figure 9 shows how rivers and buffered flood zones can be visualized or represented on GIS desktop.

3.5. PROSPECTS OF COMPUTER MODELS FOR INTEGRATED HYDROSYSTEMS MANAGEMENT

No doubt that the first computer models developed to solve hydrosystems problems targeted specific problems such as catchment runoff simulation, streamflow characterization, water quality monitoring, and so on. With the enhancement of computing efficiency and speed over the past several years, more sophisticated and user friendly computer models for hydrosystems problems have been developed. However, the objective of most of the computer models was not to address the problems of integrated hydrosystems management inasmuch as a consensus exists as to the definition of integrated hydrosystems management given in Section 2.

More recently, computer models that attempt to provide support for decision makers have been brought into the picture. One can safely say that such computer models, generally termed as decision support systems (DSS), have manifested themselves at this time as promising models for integrated hydrosystems management. The following topic discusses the DSS applications for integrated hydrosystems management.

4. Decision Support Systems (DSS) as Tools for Integrated Hydrosystems Management

4.1. DEFINITION OF DSS

Decision support systems (DSS), as might be inferred from the name, do not refer to a specific area of specialty. It is not easy to connote a specific definition to DSS based on their uses. Reistma, et al. (1996) point out that although some consensus exists as to the purpose of DSS, "a single, clear, and unambiguous definition is lacking". Generally, however, a DSS gives pieces of information, sometimes real-time information, that help make better decisions. Sprague and Carlson (1982) defined a DSS as an interactive computer-based support system that helps decision makers utilize data and models to solve unstructured problems.

4.2. BASIC STRUCTURE OF DSS

DSS generally consists of three main components: 1) state representation, 2) state transition, and 3) plan evaluation (Reitsma, et al., 1996). State representation consists of information about the system in such forms as databases and geographic information systems. State transition takes place through modeling such as simulation. Plan evaluation consists of evaluation tools such as multi criteria evaluation, visualization and status checking (Reitsma, 1996). The above three components comprise the database management subsystem, model base management subsystem and dialog generation and management subsystem, respectively. Figure 10 depicts these subsystems including their specific purposes and functions. Some examples of DSS for different integrated hydrosystems management are presented later in this Section.

Jamieson and Fedra (1996) elaborated on the basic structure of the WaterWare DSS (Figure 11). It is shown in this Figure that each subsystem is made up of different components. The data management subsystem can use different tools such as GIS as well as other simplistic data. The model base subsystem basically consists of

simple simulation models, optimization techniques and expert systems (also sometimes known as rule-based simulation models). The dialog generation and management subsystem helps in visualization and making decisions through interactive user interface.

The structure of DSS discussed above has, perhaps, made them the best structured and most promising computer models for integrated resource management. These models are believed to contribute largely to this objective. Reitsma, et al. (1996) pointed out that "… the next few years will be most interesting" for DSS. This stems from the fact that DSS are promising computer models for integrated hydrosystems management and the advance in the computing and information technology is remarkable.

4.3. EXAMPLES OF DSS FOR INTEGRATED HYDROSYSTEMS MANAGEMENT

4.3.1. Trinity River Basin, Texas

One of the integrated DSS in regional hydrosystems management was developed for the Trinity river in Texas (Ford and Killen, 1995). This DSS has the capability of integrating three major hydrosystems problems. Accordingly, it has three components which perform the following tasks: 1) retrieve, process and file rainfall and streamflow data; 2) estimate basin average rainfall and forecast runoff; and 3) simulate reservoir operation in order to forecast regulated flows basinwide. Each of the tasks is done by the DSS subsystems which use existing models. The first subsystem, data-retrieval, processing and filing subsystem, retrieves data that are collected from an existing precipitation and streamflow gauge network, and stores the data using a time-series database-management system (DBMS) designated as HEC-DSS. The second subsystem, rainfall estimating and runoff forecasting subsystem, uses the following computer programs: 1) PRECIP to compute catchment areal-average rainfall, and 2) HEC-1F for forecasting runoff. The third subsystem, reservoir simulation subsystem, uses HEC-5 that is customized and fitted to basin conditions.

Figure 12 shows different components of this DSS that are used for forecasting streamflow. TRACE (Trinity River Advanced Computing Environment) is the forecaster’s interface of the DSS. It executes programs PRECIP, HEC-1F and HEC-5 with the proper input. It also serves as a file manager, input processor and DBMS interface. Furthermore, it executes, behind the scenes, programs PREFOR and PREOP to complete the HEC-1F and HEC-5 files, respectively. The DBMS-interface

component of TRACE executes program EXTRCT to create working copies of data records, program DISPLAY to graph data, and program DWINDO to tabulate and edit data (Ford and Killen, 1995).

The size of the Trinity river basin for which this DSS was developed is approximately 4.6 million ha (17, 800 sq. mi.). Seven multipurpose major reservoirs having a total capacity of approximately 13.63 billion m³ (11,080,000 acre-ft) are found in the basin.

4.3.2. TERRA (TVA Environment and River Resource Aid)

TERRA is a DSS developed for the Tennessee Valley Authority (TVA) and the Electric Power Research Institute (EPRI) (Reitsma, et al., 1996). It was developed for the management of the TVA river, reservoir and power resources. TERRA has the following characteristics:

consists of geo-relational data base;

serves as the central data-storage and retrieval system;

records the TERRA information flow;

supports interfacing specialized data management software;

has various visualization tools; and

checks the data entering the database or data from both resident and non resident models against various sets of operational constraints (environmental, recreational, special/emergency, navigational and so on).

TERRA consists of the three essential components of a DSS, namely, 1) management of the state information of the TVA river basin, 2) the models for conducting simulations and optimizations, and 3) a comprehensive set of reporting and visualization tools for studying, analyzing and evaluating current and forecast states of the river system.

4.3.3. PRSYM (Power and Reservoir System Model)

This model is used for river, reservoir and power systems. It provides a tool for scheduling, forecasting and planning reservoir operations. It integrates the multiple purposes of reservoir systems such as flood control, navigation, recreation, water supply, and water quality, with power system economics by solving the problem based on pure simulation, rule-driven simulation or a goal programming optimization (Zagona, et al., 1995).

Shane, et al. (1995) note that PRSYM represents a major advance in modeling flexibility, adaptability and ease of use, which enable the users to:

Visually construct a model of their reservoir configuration using "icon programming" with icons representing reservoir objects, stream reach objects, diversions, etc.;

Select appropriate engineering functions, standardized by the industry, to reflect object characteristics needed for schedule planning, e.g., reservoir and stream routing methods;

Replace outdated functions with improved versions developed by industry;

Develop and include functions that are unique to their system;

Experiment with operating policies; and

Use data display and analysis objects to customize data summary presentations.

4.3.4. Conjunctive Stream-Aquifer Management

This DSS is used for conjunctive management of surface water and groundwater under the prior appropriation water right (Fredericks, et al., 1998). It has the three components which are typical of a DSS: database management subsystem, model base management subsystem, and a dialog generation and management subsystem or user interface. It is possible to prepare input data files for this DSS using GIS. The overlay of the GIS raster or grid database with other aquifer grid data enabled the finite groundwater model MODFLOW to readily read these data.

4.3.5. RiverWare

Developed by the Center for Advanced Decision Support for Water and Environmental Systems (CADSWES) at the University of Colorado, this DSS was designed for a general river basin modeling for a wide range of applications (Zagona, 1998). It has three fundamental solution methods: simple simulation, rule-based simulation and optimization.

To abate the problems of complicated water policies, a different programming language (from the usual programming languages such as FORTRAN and C/C++) called RiverWare Rule Language (RWRL) is used. Policy descriptions can be designed as structured ruleset in RWRL. Once these policy descriptions are saved as ruleset files, a simulation may be guided by the ruleset (Dumont and Lynn, unpublished). Furthermore, the policies can be modified between runs, without requiring the simulator to be changed or rebuilt (Wehrend and Reitsma, 1995).

Wehrend and Reitsma (1995) gave the following examples of how water policies can be formulated and interpreted.

IF Mead’s elevation > 1229.0 THEN

Mead’s release = Mead’s inflow

END IF

This approach gives a conditional water policy, which may be considered to be easy enough to be incorporated in a general simulation model.

IF Mead’s elevation > value THEN

Mead’s release = mead’s inflow

END IF

In this approach, the user has the choice of changing value at run-time without rebuilding the program. However, the policies expressed in this fashion may be still very specific.

A more comprehensive approach is to allow policies to be completely modifiable without requiring the underlying system to be rebuilt. As such, policies can be written in a rule language that interprets the policies and be interfaced with the simulation models. The policies are interpreted during run-time, which makes the running time of the program longer.

The general architecture of RiverWare program employs the representation of a river basin by objects. The objects that are included in Riverware include the following (Zagona, et al., 1998):

Storage Reservoir – mass balance, evaporation, bank storage, spill;

Level Power Reservoir – Storage Reservoir plus hydropower, energy, tailwater, operating head;

Sloped Power Reservoir – Level Power Reservoir plus wedge storage for very long reservoirs;

Pumped Storage Reservoir – Level Power Reservoir plus pumped inflow from another reservoir;

Reach – routing in a river reach, diversion and return flows;

Aggregate Reach – many Reach objects aggregated to save space on the workspace;

Confluence – brings together two inflows to a single outflow as in a river confluence;

Canal – bi-directional flow in a canal between two reservoirs;

Diversion – diversion structure with gravity or pumped diversion;

Water User – depletion and return flow from a user of water;

Aggregate Water User – multiple Water Users supplied by a diversion from a Reach or Reservoir;

Aggregate Delivery Canal – generates demand and models supplies to off-line water users;

Groundwater Storage Object – stores water from return flows;

River Gage – specified flows imposed at a river node;

Thermal Object – economics of thermal power system and value of hydropower;

Data Object – user specified data: expression slots or data for policy statements.

Table 4 shows user methods for selected objects in RiverWare.

4.3.6. AQUATOOL

Developed at the Universidad Politécnica de Valencia (UPV), Spain, as a result of a continuing research over a decade, AQUATOOL is a generalized decision support system that has attracted several river basin agencies in Spain (Andreu, et al., 1996). Andreu, et al. (1996) also note that AQUATOOL has various capabilities that can be used in water resource systems to:

screen design alternatives by means of an optimization module, obtaining criteria about the usefulness and performance of future water resource developments;

screen operational management alternatives by means ofUser Methods
Reservoirs
Evaporation and precipitation
No evaporation

Pan and ice evaporation

Daily evaporation

Input evaporation

CRSS evaporation
Spill
Unregulated spill

Regulated spill

Unregulated plus regulated

Regulated plus bypass

Unregulated plus regulated plus bypass
Power Reservoirs
Power
Plant power

Unit generator power

Peak base power

LCR power
Tailwater
Tailwater base value only

Tailwater base value plus lookup table

Tailwater storage flow lookup table

Tailwater compare

Hoover tailwater
Reaches
Routing
No routing

Time lag routing

Variable time lag routing

SSARR

Muskinghum

Kinematic wave

Muskingum-Cunge

MacCormack
Water User (on AggDiversion)
Return flow
Fraction return flow

Proportional storage

Variable efficiency

check and refine the screened alternatives by means of a simulation module;

perform sensitivity analysis by comparing the results after changes in the design or in the operating rules;

use different models, once an alternative is implemented, as an aid in the operation of the water resource system, mainly for water allocation among conflicting demands and to study impacts of changes in the system; and

perform risk analysis for short and medium term operational management to decide, for instance, the appropriate time to apply restrictions and their extent.

AQUATOOL has been accepted by the Sagura and Tagus river basins agencies in Spain as a standard tool to develop their basin hydrologic plan and to manage the resource efficiently in the short to medium term (Andreu, et al., 1996).

4.3.7. WaterWare

This decision support system is a comprehensive model for integrated river basin planning. It has the capabilities of combining geographical information systems, database technology, modeling techniques, optimization procedures and expert systems (Jamieson and Fedra, 1996). The aspects of integrated river basin management that this DSS incorporates are briefly as follows (Fedra and Jamieson, 1996).

Groundwater pollution control: simulation of flow and contaminant transport, and reduction of the level of contaminant in the aquifer and/or protecting groundwater resources.

Surface water pollution control: estimation of the level of effluent treatment required to meet the river water quality objectives.

Hydrologic processes: estimation of ungaged tributary for use in the water resources planning component (see No. 5 below); assessment of daily water balance for ungaged subcatchments, and the impact of land-use changes on runoff; and evaluation of the effects of conjunctive use of surface and groundwater.

Demand forecasting: Use of rule-based inference models which use generic expert system.

Water resources planning component consisting of

a model capable of simulating the dynamics of demand, supply, reservoir operations and routing through the channel system; and

a module for reservoir site selection which assesses ten problem classes which include:

landscape and archeological or historical sites;

land-use restrictions;

drainage, soil and microclimate;

natural habitats and associated communities;

water quality, aquatic biology and ecology;

water resources and cost implications;

reservoir construction;

reservoir operations;

socio-economic effects of reservoir operations; and

recreational provisions.

5. State of Practice of Hydrosystems Models

"Although the principle of integrated river basin management models has been aspired to in many countries, more often than not the problems have been considered in a piecemeal fashion, with experts from different disciplines using separate models (water resources, surface-water pollution control, groundwater contamination, etc.), to tackle parts of the overall problem in a reactive way" (Jamieson and Fedra, 1996). Uncoordinated hydrosystems modeling efforts often result in incompatibilities.

The new planning approaches for integrated hydrosystems management necessitate new ways of modeling. Schultz (1998) states that new planning tools are required to plan and design water resources systems on the basis of the new criteria which, include: 1) the principle of sustainable development; 2) ecological quality; 3) consideration of macroscale systems and effects; and 4) planning in view of changes in natural and socio-economic systems. He concludes that "since no planning tools following the four new criteria are available, we are faced with a vacuum."

This argument shows that the concept of integrated water resources management is a comprehensive representation of several components each of which requires sufficient representation or modeling within the whole system. Modeling needs to be driven by coverage of all aspects of integrated hydrosystems management, not by the convenience or simplicity of the modeling of each aspect of the problem. Loucks (1996) clearly puts that "an integrated view of water-resource systems can not be compartmentalized into either surface water or groundwater and either water quantity or water quality just because the respective time and space scales make the modeling or study of such divisions convenient".

On the contrary, as mentioned earlier in this paper, computer programming generally started out with the simplification of calculations of analytical functions that required very long times to solve by hand. Through time, the capability enhanced to the level of tackling complex hydrosystems problems. It is through improvements of the programming methodologies and new technological discoveries that more sophisticated hydrosystems models have been developed. Therefore, hydrosystems computer models have been approaching the essence of integrated hydrosystems management from bottom up.

The important aspects of integrated hydrosystems problems which have been tackled using computer programs include simulation, database management systems, data collection and storage systems and so on. These efforts have reached a level of promising prospect and have diminished the gap between the concept of and computer models for integrated hydrosystems management. For instance, GIS generally provides facilities for storage and management of very large geo-information. It has been possible to represent the terrain of the entire U.S. as a database of Digital Elevation Model (DEM). Automatic data collection systems such as SCADA and radar provide readily available input data for real-time analysis of integrated hydrosystems problems. Some computer models such as HEC-HMS and WMS are capable of accepting radar data.

By integrating together different computer models, it has been possible to develop DSS that have manifested to address these issues. A few of these systems have been designed not only to solve the problem, but also to attempt to interpret the result. Jamieson and Fedra (1996) point out that DSS have the capabilities of predicting what may happen under a particular set of planning assumptions and of providing expert advice on the appropriate course of action.

In summary, most of the available computer models for hydrosystems problems address only a specific issue of the general concept of integrated hydrosystems management. While they have been found satisfactory tools to solve the particular problem they are designed for, only a few DSS currently available such as TERRA, RiverWare, AQUATOOL and WaterWare are useful as stand-alone computer models for integrated hydrosystems management. Therefore, it can be inferred that because of the availability of only a limited number of DSS for integrated hydrosystems management, the state of practice of DSS for integrated hydrosystems management is premature, yet evolving.

6. Prospects for Integrated Hydrosystems Management Models

Advances in software engineering appear to be promising for integrated hydrosystems management models. It has enabled the development of models that not only incorporate easy-to-use analytical capabilities, but also offer expert advice and intelligent interrogation facilities. With these types of models, the artificial intelligence involved can be provided by a mixture of optimization techniques and expert systems that can evaluate, draw preliminary conclusions and recommend appropriate actions. This stage of development of hydrosystems models is the emergence of what has been referred to as the fifth generation of hydroinformatics system (Jamieson and Fedra, 1996).

The efforts made in the past to develop simulation models have been tremendous. Almost every specific hydrosytems problem has been modeled, albeit the limited focus of the objective of many of these models. In other words, many hydrosystems models were written to address specific hydrosystems problems such as reservoir operation, water distribution, urban drainage, streamflow, and so on. However, the painstaking task of integrating these simple models as we see it fit is still to demand of us the commitment. The parts are out, yet we are faced to put them together to bring out the wagon.

Some promising efforts in this regard have already been undertaken. The successful developments of TERRA, WaterWare, RiverWare, AQUATOOL and so on are very good examples. The efforts made at the USACE Hydrologic Engineering Center to enhance the old models to the new ones, generally known as the Next Generation (NexGen) models, may form one of the strong cores of DSS, simulation models.

DSS in general are, perhaps, the most promising approach to integrate the simple models and use for integrated hydrosystems management. The three subsystems of DSS – database management subsystem, model base management subsystem, and dialog generation and management subsystem – constitute a logical construct of the concept of integrated hydrosystems management. Figure 13 shows a representation of most of the possible components of a typical DSS that one can aspire for to develop. The dotted lines in the Figure show the components that can be included in the DSS in the future or enhancement to its current proposed structure.

The data base management subsystem provides the opportunity for easy collection, storage and alteration of data, including on real-time basis. GIS and SCADA, among others, are important systems for this purpose. The proliferation of simulation models and the availability of some advanced optimization techniques provide valuable resources in dealing with different aspects of hydrosystems problems. The graphics supported user-friendly interface environment also helps to draw appropriate conclusions and make necessary decisions that agree with predefined integrated hydrosystems management policies.

If there are challenges to overcome to use DSS for integrated hydrosystems management problems, one of the most difficult challenges, perhaps, will be not having appropriate integrated hydrosystems management policies clearly defined. It may be noted that it is possible to code any policy in a computer program. However, no code may be written for a policy that does not exist. Likewise, it can not be easy to write a clear computer code for an ambiguous or ill-defined policy.

A computer programming language specifically used for hydrosystems management policy called ruleset has been developed at CADSWES. Ruleset is a collection of rules that control simulation (Dumont and Lynn, unpublished at the time of reference).

7. Summary and Conclusions

Water being a precious, but limited, resource poses the question of how to allocate a sufficient amount to all the competing users efficiently and effectively. An integrated hydrosystems management approach enables us to have knowledge in space and time of what water is needed for and in what amount it is needed, thereby allowing for balancing out between the competing needs. Through integrated hydrosystems management, viable water policies compromising to all parties or satisfying all objectives can be formulated.

Design of multi-dimensional, multi-objective hydrosystems projects require formulation of sound water policies. As discussed herein, an integrated hydrosystems management may be the most promising means to provide the water requirements of all the competing users, requiring the involvement of all parties concerned. The scope and regional coverage of hydrosystems agencies need to be clearly defined. To this effect, a river basin or watershed approach for regional coverage is a sound strategy.

Computer models for integrated hydrosystems management can be very important tools that are helpful for fast computations, easy data management and drawing conclusions about certain water policies. Such models, generally termed as Decision Support Systems (DSS), have been introduced recently by different institutions. As computing speed and ease become more powerful, more complex yet more comprehensive computer models are being developed. Such computer models as TERRA, RiverWare, AQUATOOL and WaterWare are examples of DSS that are used for integrated hydrosystems management.

These DSS are embodied with water policies in the form of rulesets (to use the term used in RiverWare) or expert systems (to use the term used in WaterWare). These models have become successful as models of integrated hydrosystems management by the incorporation of water policies that are formulated in a form understandable in the computation processes.

At the center of DSS are found simulation and optimization models. A tremendous amount of work has been done in the past to develop simulation and optimization computer models that solve problems in the areas of hydrology, hydraulics and water resources. Effort was also made to interface simulation and optimization computer models to solve optimal control problems in water resources. Although DSS are highly based on these models, they also introduce water policy issues such as water rights, ecosystem sustainability, amenity and so on. These additional aspects have been incorporated in DSS models in such forms as rulesets or expert systems. In this regard, much more effort is needed not only because rulesets or expert systems have been recently introduced, but also because the concept of integrated hydrosystems management approach is yet to come to fruition.

In conclusion, some useful computer models in the form of decision support systems that address integrated hydrosystems management problems have been written. Some of these programs such as TERRA, which have been in use for some time now, have proved the importance of DSS in integrated hydrosystems management problems. The availability of various hydrosystems models that address specific hydrosystems problems and different optimization techniques, in conjunction with the advance in the information technology, provide a wealth of resources that are useful in designing DSS. Thus, we may conclude that not enough work has been done to develop DSS for integrated hydrosystems management. However, we have the technical resources – database management systems, simulation models, optimization techniques and advanced computing technology – and we are faced to make use of these resources to bring out more DSS for integrated hydrosystems management.

The requirements of writing DSS for integrated hydrosystms models would be more complete if the ideals of integrated hydrosystems management are clearly defined and understood, and if the policies can be easily interpreted so as to code in computer programs. The challenge in this regard is yet to be fully overcome. Heathcote (1998) points out that although the concept of integrated hydrosystems management is a strategy that is increasingly advocated in the literature, it is still relatively new. Because the concepts of integrated hydrosystems management can be best explained in terms of hydrosystems policies or rules and because such policies can be interpreted and coded in computer programs, it is very important to have these policies clearly defined for a given watershed. It may be noted that it is these policies that we begin with to deal with integrated hydrosystems management. Furthermore, the scope and areal coverage of integrated hydrosystems management that is mandated to an institution or water agency should be unambiguously defined. The authors agree with the watershed approach strategy for integrated hydrosystems management already recommended by different institutions. This approach entails hydrosystems policies that transcend political boundaries for the purpose of integrated hydrosystems management and, therefore, it is necessary that this approach be acceptable by different parties so that the best overall result is attainable.

Finally, lack of efficient techniques in the past that could be used to code hydrosystems policies in computer programs might have had negative impact on the development of computer models for integrated hydrosystems management. The advance in computing technology appears to be at a stage where it is capable of overcoming such problems. Today, a computer programming language specifically used for rulesets (a set of simulation rules) have been developed at CADSWES and therefore can be helpful for modeling integrated hydrosystems problems, should such languages become the requirement of the state-of-the-art for this purpose.

References

Adeli H. and Hung, S. L. (1995), Machine Learning – Neural Networks, Genetic Algorithms, and Fuzzy Systems, John Wiley & Sons, Inc., New York.

American Water Works Association Research Foundation (1996), Minutes of Seattle Workshop on Total Water Management, Denver, CO.

Anderson, M. P., et al. (1993), Computer Models for Subsurface Water in D. R. Maidment (editor in chief), Handbook of Hydrology, McGraw-Hill, Inc., New York.

Andreu, J., Capilla, J. and Sanchis, E. (1996), AQUATOOL A Generalized Decision Support System for Water-Resources Planning and Operational Management, Journal of Hydrology 177, pp. 269 – 291.

Bao, Y. X. and Mays, L. W. (1994b), New Methodology for Optimization of Freshwater Inflows to Estuaries, Journal of Water Resources Planning and Management 120(2), pp. 218 – 236.

Brion, L. M. and Mays, L. W. (1989), Methodology for Optimal Operation of Pumping Stations in Water Distribution Systems, Journal of Hydraulic Engineering, ASCE, 117(11), pp. 1551 – 1569.

Bulkley, J. W. (1995), Integrated Watershed Management: Past, Present and Future, Water Resources Update, Issue No. 100, Universities Council on Water Resources, Carbondale, Illinois.

Carriaga, C. C. and Mays, L. W. (1995), Optimization Modeling for Simulation in Alluvial Rivers, Journal of Water Resources Planning and Management, ASCE, 121(3), pp. 251 – 259.

Carriaga, C. C. and Mays, L. W. (1995), Optimal Control Approach for Sedimentation Control in Alluvial Rivers, Journal of Water Resources Planning and Management, ASCE, 121(6), pp. 408 – 417.

Chambers, L. (1995), Practical Handbook of Genetic Algorithms Applications, Vol. 1, CRC Press.

Clement, D. P. (1996), SCADA System Using Packet Radios Helps to Lower Cincinnati’s Telemetry Costs, Water Engineering and Management 134(8), pp. 18-20

Culver, T. B. and Shoemaker, C. A. (1992), Dynamic Optimal Control for Groundwater Remediation with Flexible Management Periods, Water Resources Research 28(3), pp. 629 – 641.

Davis, B. E. (1996), GIS: A Visual Approach, OnWord Press, Santa Fe, NM.

DeVries, J. J. and Hromadka, T.V. (1993), Computer Models for Surface Water in D. R. Maidment (editor in chief), Handbook of Hydrology, McGraw-Hill, Inc., New York.

Dumont, A. and Lynn, P. (unpublished at the time of reference), Creating a Ruleset, CADSWES, University of Colorado, Boulder, CO.

Essaid, H. I. (1990), The Computer Model SHARP, A Quasi-Three-Dimensional Finite Difference Model to Simulate Freshwater and Saltwater Flow in Layered Coastal Aquifer Systems, Water-Resources Investigation Report 90-4130, U.S. Geological Survey, Menlo Park: CA.

Fedra, K. and Jamieson, D.G. (1996), The ‘WaterWare’ Decision Support System for River-Basin Planning. 2. Planning Capability, Journal of Hydrology 177, pp. 177 - 198.

Fredericks, J. W., et al. (1998), Decision Support System for Conjunctive Stream-Aquifer Management, Journal of Water Resources Planning and Management 124(2), pp. 69 – 78.

Ford, D. T. and Killen, J. R. (1995), PC-Based Decision-Support System for Trinity River, Texas, Journal of Water Resources Planning and Management 121(5), pp. 375 – 381.

Goldman, F. E. (1998), The Application of Simulated Annealing for Optimal Operation of Water Distribution Systems, Ph.D. Dissertation, Arizona State University, Tempe: AZ.

Goldman, F. E. and Mays, L. W. (1999), Simulated Annealing Approach for Operation of Water Distribution Systems Considering Water Quality, ASCE (in review).

Greene, R.G. and Cruise, J.F. (1995), Urban Watershed Modeling Using Geographic Information System, Journal of Water Resources Planning and Management 121(4), pp. 318 – 325.

Grigg, N. S. (1998), Coordination: The Key to Integrated Water Management, Water Resources Update, Issue No. 111, Universities Council on Water Resources, Carbondale, Illinois.

Hall, Millard W. (1998), Extending the Resources: Integrating Water Quality Considerations into Water Resources Management, Water Resources Update, Issue No. 111, Universities Council on Water Resources, Carbondale, Illinois.

Heathcote, Isobel W. (1998), Integrated Watershed Management Principle and Practice, John Wiley & Sons, Inc., New York.

Hooper, B. (1995), Towards More Effective Integrated Watershed Management in Australia: Results of a National Survey, and Integrated Implications for Urban Catchment Management, Water Resources Update, Issue No.100, Universities Council on Water Resources, Carbondale, Illinois.

Khalil, H. M. (1998), Proposed Water Management System for the Great Man-made River Project in W. R. Blain (ed.), Hydraulic Engineering Software VII, , pp. 361 – 379.

Kirkpartick, S., et al., (1983), Optimization by Simulated Annealing, Science, American Association for the Advancement of Science, 220(4598), pp. 671 – 680.

Jamieson, D.G. and Fedra, K (1996), The ‘WaterWare’ Decision-support System for River-basin Planning. 1. Conceptual Design, Journal of Hydrology 177, pp. 163-175.

Jeter, M. W. (1986), Mathematical Programming An Introduction to Optimization, Marcel Dekker, Inc., New York.

Julien, P. Y., et al.. (1995), Raster-Based Hydrologic Modeling of Spatially Varied Surface Runoff, Water Resources Bulletin 31(3), pp. 523 – 536.

Kool, J. B. and van Genuchten, M. Th. (1991), HYDRUS: One-dimensional Variable Saturated Flow and Transport Model, Including Hysteresis and Root Uptake, U. S. Department of Agriculture, Agriculture Service, Riverside, CA.

Lasdon, L. S. and Waren (1986), GRG2 User’s Guide, Department of General Business, The University of Texas at Austin, Austin, Texas.

Li, G. L. and Mays, L. W. (1995), Differential Dynamic Programming for Estuarine Management, Water Resources Planning and Management 121(6), pp. 455 – 462.

Loucks, D. P. (1996), Surface Water Resource Systems in L. W. Mays (ed.), Water Resources Handbook, McGraw Hill, Inc., New York.

Mays, L. W. (1991), Flood Simulation for a Large Reservoir System in the Lower Colorado River Basin, Texas in National Water Summary 1988-89 – Hydrologic Events and Floods and Droughts, USGS Water Supply Paper 2375, US Government Printing Office, pp. 143 –146.

Mays, L. W. (1997), Optimal Control of Hydrosystems, Marcel Dekker, Inc., New York.

Mays, L. W. and Tung, Y. K. (1992), Hydrosystems Engineering and Management, McGraw-Hill, Inc., New York.

Mitchell, B. (1998, ed.) Integrated Water Management: International Experiences and Perspectives, Belhaven Press, London.

Moore, I. D., Grayson, R. B., and Ladson, A. R. (1991), Digital Terrain Modeling: A Review of Hydrological, Geomorphological, and Biological Applications, Hydrological Processes 5(11), pp. 3 – 30.

Murty, K. G. (1995), Operations Research Deterministic Optimization Models, Prentice Hall, Englewood Cliffs: NJ.

Nelson, E. J. (1995), A Comprehensive Environment for Watershed Modeling and Hydrologic Analysis in Espey, W. H., Jr. and Combs, P. G. (eds.), Water Resources Engineering, Vol. 1, Proceedings of the First International Conference, ASCE, San Antonio, Texas, Aug. 14 – 18, 1995, pp. 829 – 833.

Nicklow, J. W. and Mays, L. W. (1999), Optimal Control of Reservoir Releases to Minimize Sedimentation in Rivers and Reservoirs, Journal of Hydraulic Engineering, ASCE (in review).

Nicklow, J. W. and Mays, L. W. (1999), Operation of Multiple Reservoir Systems to Control Sedimentation in River-Reservoir Networks, Journal of Hydraulic Engineering, ASCE (in review).

Philip, E. G. (1988), GAMS: A User’s Guide, The Scientific Press, Redwood City, CA.

Reitsma , R. F., et al. (1996), Decision Support Systems (DSS) for Water Resources Management in L. W. Mays (editor-in-chief), Water Resources Handbook, McGraw-Hill, Inc., New York.

Rossman, L. A. (1994), EPANET Users Manual, Project Summary Report, Risk Reduction Engineering Laboratory, U. S. Environmental Protection Agency (EPA), Cincinnati, OH.

Sakarya, A. B. and Mays, L. W. (1999), Optimal Operation of Water Distribution Systems for Water Quality Purposes, Journal of Water Resources Planning and Management, ASCE (in review).

Sakarya, A. B., Goldman, F. E. and Mays, L. W. (1998), New Methodologies for Optimal Operation of Water Distribution Systems for Water Quality Purposes in Blain, W. R. (ed.), Hydraulic Engineering Software VII, Wessex Institute of Technology Press, Boston, pp. 101 – 110.

Schultz, G. A. (1998), A Change of Paradigm in Water Sciences at the Turn of the Century?, Water International, Journal of the International Water Resources Association 23(1), pp. 37 – 44.

Skaggs, R. W. and Mays, L. W. (1999), Simulated Annealing for Groundwater Restoration, Journal of Water Resources Planning and Management, ASCE (in review).

Shane, R. M., et al. (1995), The INTEGRAL PROJECT: Overview in Computing in Civil Engineering, Proceedings of the Second Congress, Vol. 1, pp. 203 – 205, ASCE, June 5 - 8, Atlanta, GA.

Sprague, R.H. and Carlson, E. D. (1982), Building Effective Decision Support Systems, Prentice-Hall, Inc., Englewood Cliffs: NJ

Tang, A. and Mays, L. W. (1999), Genetic Algorithms for Optimal Operation of Soil Aquifer Treatment Systems, Water Resources Management, Kluwer Academic Publishers, The Netherlands, to be published, 1999.

Topping, B.H.V, et al., (1993), Topological Design of Truss Structures Using Simulated Annealing in Topping, B.H.V. and Khan, A. I. (eds.), Neutral Networks and Combinatorial Optimization in Civil and Structural Engineering, pp. 151 – 165, Civil-Comp Press, Edinburgh: UK.

Unver, O., Mays, L. W., and Lansey, K. (1987), Real-time Flood Management Model for the Highland Lakes System, Journal of Water Resources Planning and Management 113(5), pp. 620 – 638.

U.S. Army Corps of Engineers Hydrologic Engineering Center (HEC) (1998), HEC-FDA Flood Damage Reduction Analysis, User’s Manual, Version 1.0, January 1998.

U.S. Army Corps of Engineers Hydrologic Engineering Center (HEC) (1998), HEC-HMS, Hydrologic Modeling System, User’s Manual, Version 1.0, March 1998.

U.S. Army Corps of Engineers Hydrologic Engineering Center (HEC) (1997), HEC-RAS River Analysis System, User’s Manual, Version 2.0, April 1997.

U.S. General Accounting Office (1994), Ecosystem Management Additional Actions Needed to Adequately Test a Promising Approach, GAO/RCED-94-111.

U.S. Geological Survey (1998), Summary of MODFLOW96, User’s Manual.

Viessman, W., Jr., (1998), Water Policies for the Future: Bringing It All Together, Water Resources Update, Issue No. 111, Universities Council on Water Resources, Carbondale, Illinois.

Vlachos, E. C. (1998), Practicing Hydrodiplomacy in the 21^st Century, Water Resource Update, Issue No. 111, Universities Council on Water Resources, Carbondale, Illinois.

Wanakule, N., Mays, L. W., and Lasdon, L. S. (1986), Optimal Management of Large Scale Aquifers: Methodology and Applications, Water Resources Research 22(4), pp. 447 – 465.

Wehrends, S. C. and Reitsma, R. F. (1995), A Rule Language to Express Policy in a River Basin Simulator in Computing in Civil Engineering, Proceedings of the Second Congress, Vol. 1, pp. 392 – 395, ASCE, June 5 - 8, Atlanta, GA.

Wada, R. N., et al. (1986), Honolulu’s New SCADA System, Journal of American Water Works Association 78(8), pp. 43 - 48.

Winston, W. L. (1994), Operations Research Applications and Algorithms, Duxbury Press, Belmont: CA.

Wurbs, R. A. (1995), Water Management Models A Guide to Software, Prentice Hall PRT, Englewood Cliffs: NJ.

Zagona, E. A. (1995), The INTEGRAL PROJECT: The PRYSM Reservoir Scheduling and Planning Tool in Computing in Civil Engineering, Proceedings of the Second Congress, Vol. 1, ASCE, June 5 - 8, Atlanta, GA.

Zagona, E. A., (1998), RiverWare: A General River and Reservoir Modeling Environment, Proceedings of the First Federal Interagency Hydrologic Modeling Conference, April 19 - 23, Las Vegas, NV.

Zhao, B. and Mays, L. W. (1995), Estuary Management by Discrete-Time Stochastic Linear Quadratic Optimal Control, Journal of Water Resources Planning and Management 121(5), pp. 382 – 391.

Home | Resume | Books | Classes | Ancient Water | Research | Students | Research Papers