The development of a management plan for future utilization of Nile river basin water involves people and groups who may or may not have a well-developed, intuitive understanding for the geography and hydrology of the river system. This diversity of training and experience suggests a need for a highly intuitive, easily used Nile River basin model that would allow these people and groups to examine the effects of policy options on the behavior of the river system by direct, interactive experimentation. This pedagogical method of simulation brings concerned individuals with varying backgrounds and levels of technical training to a common understanding of the physical reality on which project decisions in part depend.

The simulation approach to understanding river basin behavior necessitated a physically realistic simulator that could be easily used by both technical and non-technical people. NileSim, a PC-based, interactive simulator of the Nile River basin (from the Equatorial Lakes of east Africa and the highlands of Ethiopia to the Delta of Egypt) attempts to meet this need. The user of NileSim may observe the effect of changes by dynamically manipulating river system controls and regulating water usage. This capacity for direct experimentation will foster an intuitive understanding of the hydraulic behavior of the river basin. In a sense, NileSim is like a flight simulator that gives the user a safe locale within which to learn by experimenting. In this simulated environment, the user can try different decision strategies-perhaps even injudicious ones-and safely observe the effects.

A useful simulator has several potential purposes; it may be used as an investigative tool, or for policy-making, or for engineering design (McCuen, 1989). It should allow control over variability and should lead to an improved understanding of data needs. Most practically, the simulator should be cheaper, safer, and easier to manipulate than the system itself.

If a model represents a physical system with a suitable level of accuracy and detail, then the simulator is a means by which "what-if" scenarios can be posed and answered. For example, to understand how the Jonglei diversion canal around the el-Sudd swamp would affect water availability at Lake Nasser, the user can insert a canal in the simulator and observe the corresponding downstream changes. One might ask not only about long-term changes in annual water volumes in Lake Nasser, but also about the system-wide relation between water management policies, societal benefit, and water-borne disease such as schistosomiasis. When the water availability is affected by natural variation, the outcomes or system changes are themselves subject to variability and must be described probabilistically.

Egypt has traditionally been the major user of Nile River water. Under the 1959 treaty, Egypt was allocated 55.5 billion cubic meters (BCM) per annum, and the Sudan, 18.5 BCM. Ethiopia was not a party to the 1959 treaty, but its rapidly growing population and its desire for industrialization suggest that Ethiopia's need for Nile River water will increase significantly. Other riparian states, such as Uganda, currently use little Nile River water; but they, too, will demand a greater share in the future. Major projects being planned or in progress will expand water demand beyond the nominal 84 BCM per annum. The most noteworthy of these are the Western Desert project and Sinai Canal and Siphon in Egypt, and Ethiopia's plan for a series of dams on the Blue Nile.

With today's population and water constraints, Egypt, Ethiopia, the Sudan, and other Nile states are net importers of food. While the management of Nile River water could be improved, the projected rise in population will create difficulties even under the most efficient allocation scenario. Thus, everything possible must be done to manage the limited Nile River waters in an optimal way.

The concept of using river basin simulators as a means of improving the intuitive understanding of people involved in negotiations is becoming more widely accepted. Simulators of the Nile for scientific and engineering purposes have been built by the US National Weather Service (Barrett, 1993) and by Georgakakos and Yao (1997), among others. These simulators, which include distributed parameter modeling of spatial rainfall-runoff relations, as well as sophisticated hydrograph routing, are complex and not easily accessible to non-technical people. Thus, the simulators tend to be used mostly by engineers and scientists, are apt to be data-intensive, and often have long computational cycles. As a result, few river forecasting simulations built for engineering purposes are useful where pedagogy is a main goal.

The challenge of illustrating hydrological concepts such as water mass balance, flow rates, and hydrographs to non-technical users lies not so much in the creation of a model of the Nile river system, but in building a simulator that can be easily used. A chief requirement of this usability criterion is the appearance of the simulator in its graphic user interface (GUI).

Upon starting NileSim, the user sees a GUI that shows a full-screen image of the Nile River basin from the Equatorial Lakes and Ethiopia to the delta (Figure 1). The GUI is the level at which information is sent to and received from the underlying simulation. While the NileSim program is a complicated set of algorithms, the user only interacts with the GUI of Figure 1. This image is color-coded, so that lakes, rivers, and reservoirs appear in dark blue, seas are light blue, political boundaries are yellow, and man-made features are red.

Large regional sections are also outlined in Figure 1. When these sections are selected with the mouse, simulator controls and plots of flow and volume with time appear in pop-up windows. The large pop-up windows correspond to the geographical regions of the Equatorial Lakes, el-Sudd, Lake Nasser, Upper Egypt, and Delta; smaller pop-up windows show controls and time-series plots for the Roseires dam, the Nile River at Khartoum, and the Atbara River.

When the simulator is running, the user can view three types of dynamic flow condition data:

River discharge as a function of time is tracked at a discrete number of points within the river network. Hydrographs can be viewed by clicking on the corresponding section of the river on the Nile map. These hydrographs are dynamic and evolve as the user modifies the simulator conditions. Currently, the discrete points at which hydrographs can be observed are coded directly in the simulation. In the future, the user will be able to click on any reach of any river in the basin and view its corresponding hydrograph.

Along the course of the lower Nile in Egypt, flows can be distributed among the major existing canals and branches of the river. At the head of the Delta, the main Nile River flow is divided among the Rosetta and Damietta branches of the river, and among the four major delta canals taking water off from the river at the barrages just north of Cairo. Default values for the fractions of flow entering of these five distributaries (Said, 1993) are coded into the simulation but may be altered by the user (Figure 2). In the example shown in Figure 2, one-fifth of the default 2.5 bcm/mo allocated for Deltaic use is distributed to the Menoufi rayah.

The geographic and hydrologic features of the Nile River system were reduced to a flowchart schematic for the purpose of model development (Figure 3). Inflows from the White Nile and Blue Nile merge at Khartoum and are joined by the Atbara further downstream. This is represented by a converging tree structure entering upstream of Lake Nasser (Figure 3).

Lake Nasser itself is modeled with four effluents: downstream flow to Egypt, water withdrawals by the Sudanese, losses to evaporation and seepage, and overflow released into the Toshka depression. These inflows and outflows are summed at the node representing Lake Nasser.

Incorporation of existing flow containment structures, such as the Aswan high dam and the Roseires dam, enables users to examine the effects of various flow release rates. In addition to the modification of flow rates from existing structures, the model also allows the construction of proposed flow control structures within the time duration of a simulation. These future structures include canals near the Sobat marshes, Jonglei, Bahr el-Ghazel, and the Toshka depression, and dams at Lakes Edward and Victoria.

The simulator is based on a detailed description of the physical hydrology of the river system and is calibrated to empirical records of the basin. Its output reproduces the descriptive statistics of observed hydrographs, reservoir levels, and travel times of flood waves along river reaches. The NileSim model incorporates monthly river flow variability from analyses described in the literature (Said, 1993; Shahin, 1985). The monthly flow and volume estimates include a stochastic component. The filling and draining of reservoirs obey mass balance, and the travel times of flood waves downstream are described by Manning's equation and geographically distributed reach geometries (McCuen, 1989).

One important feature that distinguishes NileSim from earlier hydraulic simulation models is its use of simulation technology drawn from other engineering disciplines besides hydrology and civil engineering. This technology transfer is possible because of the emergence within the past five years of computer-aided software development tools for electrical engineering, mechanical design, and other large market sectors. These software tools allowed NileSim to be built in a functionally acceptable and cost-effective way. This produced a simulator of the Nile that: (i) is easy to use; (ii) allows an intuitive visual interaction between the user and the simulation; (iii) is physically realistic; and (iv) was quick and inexpensive to develop.

The underlying network simulation of the river basin was developed using VisSim (Visual Solutions, Inc., 1995), a network simulation toolkit from the electrical engineering industry. VisSim combines a graphical development environment for rapidly building network diagrams with a set of mathematical and logical tools for describing analytical relationships within such networks.

The Vissim simulation engine underlying NileSim uses circuit-like logic to construct an analytical representation of the Nile river basin from a series of river reaches, reservoirs, mathematical operators, and time series functions. Some indication of the complexity of the Vissim engine is suggested by Figure 4. The Nile module in Vissim is shown in partial, expanded form to illustrate the connection of lakes, reservoirs, and reaches in the Equatorial Lakes region. The Semliki River reach (i.e., model channel 03) is further expanded to show the representation of the depth-discharge function for that reach (Leopold and Maddock, 1953).

The NileSim GUI was built using Delphi3.0 (Borland, 1997). The final code constructed using these tools was compiled into a C-based executable for users to download from the Internet. This strategy of using CASE tools rather than writing original code allowed NileSim to be developed within a few months and to have both the functionality and look-and-feel of standard Windows software; this consistency with industry standards is important in simplifying training and encouraging use.

During the simulation, the user may create a dam, increase a dam height (e.g., Owen Falls dam), and specify release discharges from any reservoir in the system. The effect of these changes is to raise the respective lake level upstream and reduce the variability of all flows throughout the river system downstream. Other structures, such as dams on the Blue Nile or consumptive uses for agricultural development anywhere within the basin, can be similarly modified in the simulation. Finally, the user may add or remove water-bearing canals.

Three proposed canal projects have been included in the simulation: the Jonglei canal, a diversion of the Bahr el-Ghazel, and the Sobat marsh canal. In the default simulation of the Nile River, these canals do not exist. The user may at any time turn on one or more of these canals, and from that point in the simulation forward (until turned off) the network is affected by the new diversions.

An example of experimentation can be seen in Figure 5, where the outflow hydrograph from the Sudd into Lake No is shown over a duration of 40 years. During the first 20 years, changes to the basin are not made. At year 20, the Jonglei Canal is built (i.e., turned on). Relatively quickly, the effect of the canal is seen on the discharges from Lake No, even though the stochastic nature of the flow from year to year makes this effect difficult to see one year at a time. Subsequent effects of modification of flow from Lakes Victoria and Edward change the average annual discharge and a reduction in variability.

The annual discharge data reported by NileSim (see Figure 6a) show the individual effects of the construction of the Jonglei Canal and the regulation of outflow from Lake Victoria at the Owen Falls Dam. Implementation of the Jonglei Canal reduced the average annual discharge measured at Lake No from 15 bcm/yr (Case 1) to 11 bcm/yr (Case 2). Construction of the Owen Falls Dam with a steady release of 2.25 bcm/mo (Case 3) reduced the variability of the annual discharge from 1.75 bcm/yr to 0.49 bcm/yr; this large drop in variability is largely because the Victoria Nile is the principal source of water to the el-Sudd region. Finally, the combination of the canal and the dam (i.e., Case 4) reduces both the central tendency and variability of the annual discharge. The effects of the engineering modifications on the characteristics of the discharge-frequency curves (see Figure 6b) are clear from the respective differences in central tendency and variation.

Another interesting experiment that can be performed is to test the effect of diverting water from Lake Nasser to the Western Desert Project. When the image of Lake Nasser in the map is clicked on, NileSim returns a dialog box. This box shows dynamic bar graphs for the volume of water in Lake Nasser and the volume of water released to the Toshka depression, a hydrograph of the downstream release of water from the high dam, and slider bars by which to change downstream releases, consumptive use from Lake Nasser, and fractional release into the Toshka depression. The user may allow excess storage in Lake Nasser to flow out into the Toshka depression; this outflow may be prevented by clicking on the image of Lake Nasser and within its corresponding dialog box, checking on or off the toggle box for the Toshka depression outfall.

Setting consumptive use to 4 BCM per annum, this amount of water is allocated as an out-of-basin transfer. If the user attempts to keep downstream releases constant, the volume of water in Lake Nasser will fall with time. Conversely, if the user attempts to maintain predetermined volumes of water in the lake, discharges to the downstream Nile and canal system will decrease.

Any user, with or without training in hydrology, can carry out these experiments and see the results. Using the simulator, the user can even make wild and irresponsible changes to the river system and quickly see the effect; this is an important characteristic of interactive simulation when non-technically trained professionals are involved in decision making and may propose publicly popular alternatives. A principal goal of such gaming is the development of intuition about the river basin as a dynamic system. This has not been generally possible using highly sophisticated forecasting simulators because most untrained users cannot get immediate feedback from such complex tools, no matter how valuable those tools may be to engineers and scientists. Similarly, the static and tedious review of massive numbers of data tables, graphs, and equations is an almost hopeless way for most people to develop an understanding of the river system.

Of greater interest would be the ability to incorporate within NileSim future demands for water use based on population and economic indicators. The simulator should be expanded to allow modeling based on water demand, agricultural demand, or other economic criteria. Clearly, many "what if" scenarios have to do with population projections and economic development. As a first approximation, it should be relatively straightforward to incorporate population effects on water demand and water availability effects on economic development.

Less pressing but interesting additions to the simulator would include algorithms to simulate health effects and disease incidence, rates of siltation and erosion, and possible changes to flora and fauna of the Nile environment. Ultimately, one would like to base the simulation on a spatial database with a geographic information system (GIS) interface. This would lead to modeling of rainfall-runoff with real-time, spatially distributed rainfall and runoff information and to predicting of spatially distributed downstream impacts, such as agricultural production.

The changing nature of today's technologically oriented world requires that people learn not only the craft of their discipline, but develop an appreciation for the interconnectedness of natural systems. Of few instances is this more true than in river basin planning. Difficult future negotiations with respect to the Nile River and other internationally shared rivers will require that more of the parties involved develop an understanding of important aspects of river engineering that had heretofore been relegated to the domain of experts.

A number of current planning issues related to the utilization of multinational water resources lend themselves to investigation using decision support simulators like NileSim. Prominent among these are the effects of climate change due to global warming, and how these effects can be managed. Decision support simulation can also be used to assess alternative approaches to sustainable development by incorporating the effects of population growth and industrialization on spatial water demand, and correspondingly on environmental conditions. The corollary to such studies is the corresponding evaluation of the effects of changes in water availability on economic and social indices.