Stochastic Conjunctive Management of Water Resources in Yolo County
Two management models are developed to determine the optimal operating policies for hydraulically connected time-variant surface and ground water supplies in a hypothetical system. The system involves a multipurpose reservoir, a hydraulically connected stream and aquifer, agricultural plot, water supply and observation wells, and an artificial recharge zone so as to address various hydrologic components experienced in Yolo County in the management models. The first model minimizes deviations from a set of rule curves defined for storage in the reservoir and along stream course so as to consider possibilities for storage excess water in wet periods and its distribution in subsequent dry periods. The second model, in addition to the objective of the first model, minimizes total operational costs of surface and subsurface water storages (pricing) while meeting the target storage levels in the surface water supplies. The first model is formulated as a linear programming model whereas the second one is formulated as a nonlinear programming model and a -form approximating model is employed in the solution phase.
The hypothetical system is divided into components which are subsequently integrated using numerically-generated response functions that relate the system's behavior to various system excitations. The management model is formulated and solved for monthly time steps, which include both dry and wet conditions, to determine reservoir release, pumpage rate from supply wells, artificial recharge rate, water diversion from reservoir and stream to the demand area, and storage both in the reservoir and along the stream course at an optimal level in each time step over a six-month planning horizon. Sensitivity analysis of the first management plan with respect to potential use of groundwater supplies is performed to analyze the latter's impact on the operating policies. Furthermore, the performance of the -form approximating model for conjunctive management models applied to the hypothetical system is found to be promising.