Decision-Support
Within the next few years, we anticipate being able to build a system that is capable of operating a river basin with little or no direct human intervention required, by tying decision-support models to the automation systems. This will allow a system to operate at maximum efficiency around the clock. To this end, we are working on several decision-support models. Two of these are discussed below.
Decision Support for Water Rights
All western water rights are established by legal decree or statute. In Utah, regulation of water rights is an administrative procedure executed by river commissioners acting under the direction of the State Engineer. In the Sevier River Basin, the river is managed by two commissioners using procedures so complex that few people fully understand them. The most difficult aspect of the allocation procedure involves the determination of the primary flow and the segregation of this water from storage water. Another confusing aspect is the division of the flow into zones.
In view of the importance of water rights in the management of water resources, Dr. Wynn Walker, Head, Department of Biological and Irrigation Engineering at Utah State University (USU), and Roger Walker, retired Sevier River commissioner, developed a computerized water rights allocation model-SEVIER-for use by the river commissioners and others ( Walker, 1991). SEVIER duplicates the computations and record analyses performed by the two river commissioners.
To provide water rights updates in a timely manner, Dr. Walker is currently working with StoneFly to connect his model to the Sevier River monitoring system (river flows, canal diversions, and reservoir storage). This will provide the river commissioners, water users, and others with water rights information that is updated daily. By posting this information on the Web, each irrigator will have continually updated information on the status of his/her water rights and will be notified when additional water is available. This Web software is being tested during water year 2002.
Canal Routing with Artificial Normal Networks
A common requirement for real-time irrigation management is the anticipation of canal inflows that will be required to both satisfy irrigators' demands and minimize wastage. An accurate forecast of the quantity of water to be diverted is an important tool for canal managers. Forecasting the requisite diversions is a difficult task because the flow in a canal depends on physical and hydrological processes that are nonlinear and the exhibit a high degree of variability.
To assist canal managers with inflow forecasts, Abedalrazq F. Khalil and Dr. Mac McKee at USU are developing a canal inflow prediction model using an artificial neural network (ANN). A nice feature of ANN is the capability to extract the relationship between the inputs and outputs without a complete understanding of the physical processes involved. This property makes it well suited to the problem of forecasting flows through fairly primitive delivery systems.
The USU ANN model is currently being applied to the Sevier Valley/Piute Canal in the Sevier River Basin. The canal is approximately 65 miles long and supplies water to about 12,000 acres of irrigated land. The travel time for water flowing to the end of the canal is up to 3 days. Diverting an appropriate amount of water from the river into a canal of this length is a challenging problem for the canal operators.
Model inputs include the real-time flows at monitoring stations along the canal (see Figures 3 and 6) and the associated water orders for each stretch of the canal. Early results from the USU modeling efforts have been encouraging, and the Piute Canal manager is currently field testing the model.
Figure 6. Web time-series plot: hourly flows at five monitoring sites along the Sevier Valley/Piute Canal for the previous 3 months.