Journal of Water Resources Planning and Management

A model for the optimal management of a regional aquifer under salinization is developed. The objectives of management are to maximize the total amount of water pumped for use and to minimize the total amount of salt extracted with the water. The model is based on a combination of simulation and an optimization routine, run iteratively. The simulation model uses a finite-element formulation for the flow and a streamline upwind Petrov-Galerkin formulation for the transport and computes the gradient of the state variables (heads and concentrations) with respect to the decision variables (pumping rates at wells). The gradients are then used in a Bundle-Trust nonsmooth optimization procedure to achieve an improved solution. The process ends when termination criteria are met, resulting in a good solution, which cannot be claimed to be the global optimum. The procedure is demonstrated on a 600-km² nonhomogeneous regional aquifer with 12 zones of differing properties and 32 pumping wells.

A hydrological micromodel was developed and applied to a neighborhood typical of urban development on Israel's coastal plain, over the phreatic coastal aquifer. The model was used for analyzing the effects of urban development on infiltration and runoff, and for evaluating a number of practices designed to enhance on-site infiltration. The effect of spatial resolution in the model on computed results was investigated. It was shown that for the range of data examined, simulation at the micro- (residential lot) level can be extrapolated to the neighborhood scale, by adding the responses of the individual microunits. Simulations by the hydrological micromodel showed that connecting roof drains to a yard/garden, and allowing the runoff from the roof to infiltrate through an “infiltration strip” or infiltration trench of an appropriate size, can increase infiltration over a residential lot as much as 18% of the annual rainfall (depending on the soil conductivity and annual rainfall). The dependence of annual infiltration on physical and planning parameters was generalized in functional relations that can be used to assess the effectiveness of measures for increasing infiltration and reducing runoff. This work was part of an effort to develop, test, and recommend policies and practices for water-sensitive urban planning for protecting water resources.

For optimal multireservoir operation, a dynamic programming-based neural network model is developed in this study. In the suggested model, multireservoir operating rules are derived using a feedforward neural network from the results of three state variables' dynamic programming algorithm. The training of the neural network is done using a supervised learning approach with the back-propagation algorithm. A multireservoir system called the Parambikulam Aliyar Project system is used for this study. The performance of the new multireservoir model is compared with (1) the regression-based approach used for deriving the multireservoir operating rules from optimization results; and (2) the single-reservoir dynamic programming-neural network model approach. The multireservoir model based on the dynamic programming-neural network algorithm gives improved performance in this study.

On-site storm-water detention systems can be used to mitigate the hydrologic effects of residential development and to provide a supplemental water supply at the scale of single residences. A three-year rainfall record from a site in the Puget lowland, Washington state, is used in a simple mass-balance model to simulate outflow from single- and multiple-purpose detention systems. Simulations are compared to time series of measured runoff from Evans Creek, a 37 km², rural basis, and a 0.37 km², zero-order forested subbasin. Discharge from a small on-site reservoir is sensitive to both the storage capacity and maximum controlled release rate of the system for extreme high flows (those exceeded 1% of the time) and low flows (those exceeded 80% of the time). An intermediate range of discharges (those exceeded 10–30% of the time) is primarily sensitive to release rate, rather than its storage capacity, suggesting that single-purpose systems with small reservoirs can be effective for hydrologic mitigation over a range of intermediate flows.

Many authors have documented the minimal use of optimization in practical day-to-day multipurpose reservoir operations. The RiverWare decision support system is a flexible general river basin modeling tool that allows water resources engineers to both simulate and optimize the management of multipurpose reservoir systems for daily operations. This paper describes RiverWare's optimization capabilities and its use by Tennessee Valley Authority operations schedulers. Input data requirements include (1) physical and economic characteristics of the system; (2) prioritized policy goals; and (3) parameters for automatic linearization. RiverWare automatically generates and efficiently solves a multiobjective, preemptive linear goal programming formulation of a reservoir system. An advanced feature of RiverWare is that both the physical model of the river basin and the operating policy are defined and easily modified by the modeler through an interactive graphical user interface. Any modifications are automatically incorporated into the linear preemptive goal program. RiverWare's combination of detailed system representation, policy expression flexibility, and computational speed make it suitable for use in routine daily scheduling of large complex multiobjective reservoir systems.

This paper explores the capabilities of genetic algorithms for identifying solutions to waste-load allocation problems. Three optimization model formulations are developed for examining different waste-load allocation problems. Two of the model formulations address problems arising in the planning context while the third model addresses waste-load allocation decisions of use when developing an operational strategy for a river basin. The models can all be used to develop a trade-off relationship for a multiobjective optimization problem that would be of use to a decision maker when selecting a solution for implementation. The models developed are applied to a case study based on the Willamette River in Oregon and the resulting trade-off relationships are discussed both in objective space and in decision space.

Ground-water management models can be computationally intensive and field-scale application of these techniques to complex processes such as in situ bioremediation is not currently possible. A one-way spatial multiscale methodology is investigated for reducing computational effort associated with solving a ground-water management model. The multiscale approach reduces the computational burden by solving the model on a coarse mesh and then using the coarse mesh solution as a starting point for successively finer meshes, proceeding “one-way” from coarse to fine meshes. The performance of the method is presented for two different cases with different dispersivities using a three-level multiscale approach. Over 50% reduction in computing time was achieved in both cases. For the case with low dispersivity, convergence difficulties were encountered on the coarsest mesh that were overcome by using a higher dispersivity on the coarsest mesh and then switching back to the desired low dispersivity on the finer meshes. The choice of penalty weights for constraint violations also proved to be critical to the performance of this approach. Guidelines for selecting appropriate penalty weights are given.