Optimal Staging of Russian River Basin Development

A planning model for a river basin development is constructed and applied to the Russian River Basin in California. Using dynamic programming methods, the problem is set in both deterministic and stochastic frameworks. In the latter, future population in the region and the corresponding time-dependent, price-sensitive demand relationships for water are projected with stochastic disturbance. A learning mechanism is incorporated in the algorithm so that population projections are updated with time by a Bayesian rule, and an optimal investment strategy is obtained for each time period conditional upon observed population at that period. Application to the Russian River Basin recognizes explicitly development of the three major multipurpose water projects of the Basin. Optimal timing and scheduling of these projects strongly depends upon population growth, but ordering of their construction does not. Results indicate that it would be optimal to construct only two of these three projects within the next 50-year period-a second stage of the currently existing Lake Mendocino and the Warm Springs projects. The third and largest project (Knights Valley) should be postponed beyond the year 2020 (the planning horizon in the present study). For economically reasonable discount rates, the Lake Mendocino enlargement should precede the Warm Springs project. The economically ~fficient solution calls for construction of the former between 1980 and 1990. The latter should not be constructed prior to the year 2000, its optimal timing depending on population growth. Under current projections of population growth by the California Department of Finance, the optimal timing for constructing the Warm. Springs project is between 2005 and some time after 2020, the end of the planning period.