The Mid-Pacific Division of the Bureau of Reclamation is responsible for operation of the Central Valley Project (CVP), which provides water to users in central California. Traditionally, decisions about operation of the CVP have been made using conservative 90%-chance inflow forecasts, without explicit modeling of uncertainty in demands or facility performance. Consequently, the probability of meeting or failing to meet water demands in the system has not been evaluated, reported, or considered in the decision making. While the operation has largely been able to meet demands, ever-increasing environmental water allocations have motivated system operators to seek methods for improved decision making.
The Bureau of Reclamations asked us to identify viable uncertainty analysis techniques and the likely "pay off" of their use if they were incorporated in their decision-making processes.
|
We worked with the bureau's operators and planners and developed a tutorial to demonstrate the impact of uncertain inputs and to identify methods for analysis. We reviewed the operation decision making process to identify influencing factors about which operators are uncertain. These include (1) factors associated with total system inflow; (2) factors associated with project storage; (3) factors associated with meeting in-basin needs; (4) factors associated with specific facility operations; and (5) factors associated with Delta export and south-of-Delta conveyance operations. Of these factors, David Ford Consulting Engineers Inc. identified those that could be described with a probability distribution and described how the decision making might be improved with (1) sensitivity analysis; (2) uncertainty analysis; or (3) uncertainty propagation methods. Included in those methods is position analysis.
We recommended position analysis as an appropriate tool for providing additional information for CVP operation decision making. We presented progress reports to Bureau staff throughout the project, and provided a series of detailed reports illustrating the numerical methods used in our recommendations.
|