Journal of Water Resources Planning and Management

This paper presents results of a large-scale economic-engineering optimization model of California’s water supply system. The results of this 4-year effort illustrate the value of optimization modeling for providing integrated information needed to manage a complex multipurpose water system. This information includes economic benefits of flexible operations, economic valuation of capacity expansion opportunities, estimating user willingness to pay for additional water, economic opportunity costs of environmental flows, and identification of promising conjunctive use and water transfer opportunities. The limitations of such modeling also are discussed. Overall, the results suggest improvements to system operation and water allocations with a statewide expected value potentially as high as $1.3 billion/year. Significant improvements in performance appear possible through water transfers and exchanges, conjunctive use, and various operational changes to increase flexibility. These changes also greatly reduce costs to agricultural and urban users of accommodating environmental requirements. Model results also suggest benefits for expanding selected conveyance and storage facilities.

Decadal climate variability in the form of the Pacific Interdecadal Oscillation (PDO) is analyzed with the El Niño Southern Oscillation (ENSO) to assess impacts on seasonal water supply in the Columbia River Basin and to test for statistical significance. The cross-correlation functions indicate a significant correlation between seasonal runoff and both the Southern Oscillation Index (SOI) and PDO, suggesting a time period where ENSO events have the most significant impact on the January through July volume runoff at The Dalles Dam. Correlation functions indicate volume runoff is directly related to SOI phase and inversely related to PDO phase. Results show there is a statistically significant change in the mean January through July volume runoff only when El Niño events occur during the positive phase of the PDO and when La Niña events occur during the negative phase of the PDO. Precipitation and temperature patterns are illustrated to explain this shift in the mean runoff. While slight shifts in variability are evident for the various climate conditions, significance tests show that the changes in variability do not differ significantly from the neutral condition.

The process of integrating discipline-specific hydrologic, economic, and fish health models in the context of a web-enabled spatial decision support system (SDSS), WebL2W is discussed. As part of a larger research project, this component focused on the issues of integrating assessment models with different spatial and temporal resolution and providing an interface and modeling/decision processing environment. Spatial decision support systems and the advantages of internet-based, cross disciplinary modeling for the watershed management decision process are described. The importance of GIS base and derived data is highlighted in conjunction with a description of the hydrologic, economic, and fish health model components. Users interact with the map interface of WebL2W using the Back Creek Basin near Roanoke, Va. as the sample watershed and choose one of four available housing templates as the development mode. Following creation of a suitable development “scenario” for the watershed, the models are executed and results are displayed to the user. Tools, methods, and architecture utilized for construction of the web-enabled SDSS are delineated, and a case study illustrates the functionality of WebL2W.

Declining fall flows are limiting the ability of the Cosumnes River to support large fall runs of Chinook salmon. Management scenarios linking surface water and groundwater alternatives to provide sufficient fall flows are examined using groundwater flow and channel routing models. Results show that groundwater overdraft in the basin has converted the river to a predominantly losing stream, practically eliminating base flows. Management alternatives to increase net recharge (for example, pumping reductions) were examined along with surface water augmentation options. Using a minimum depth standard for fish passage, average surface water flow deficits were computed for the migration period of Chinook salmon. Groundwater deficits were evaluated by comparing simulated current groundwater conditions with conditions under various scenarios. Increases in net recharge on the order of 200 to 300 million m³/year would be required to reconnect the regional aquifer with the channel and in turn reestablish perennial base flows. Options that combine surface water augmentation with groundwater management are most likely to ensure sufficient river flows in the short term and to support long-term restoration of regional groundwater levels.

The impacts of physical and chemical aquifer heterogeneities on optimal remediation design, costs, and time to compliance are investigated by linking a genetic algorithm with a contaminant transport simulation model. Physical and chemical aquifer heterogeneities were grouped into three levels as follows: (1) hydraulic conductivity (K) heterogeneity only; (2) combined heterogeneity of K and the distribution coefficient $\left(K_d\right);$ and (3) combined heterogeneity of K, $K_d\text{,}$ and the mass transfer rate (α). Various degrees of heterogeneity were considered, ranging from slightly heterogeneous to strongly heterogeneous. Impacts were evaluated using two different optimization models: the optimal design model and the time-to-compliance model. The first model focused on finding optimal aquifer remediation designs and costs under various heterogeneity conditions, and the second model optimized the time needed to meet the water quality goals for a fixed pumping schedule. Results show that the variability in the remediation costs and time to compliance for different realizations of a heterogeneous K-field increases as K-heterogeneity increases. Consideration of $K_d$- and α-heterogeneities results in different policies and costs compared to cases where sorption heterogeneity is neglected. In general, time to compliance increases for systems with both chemical and physical heterogeneity as compared to systems with only physical heterogeneity. The impact of α-heterogeneity on remediation strategies is most apparent when $K_d$-heterogeneity is high. Although an increase in K-heterogeneity decreases the impact of $K_d$- and α-heterogeneities on remediation costs and time to compliance, sorption heterogeneity could still significantly impact the performance of a remediation system, especially when sorption heterogeneity is high.

A new method is presented for detecting the magnitude of leaks in small residential service zones (under 1,000 homes) of a drinking water distribution system. It is assumed that continuous measurements of flow rates through the main supply line into a residential service zone are available during periods of low use. The sample mean and variance from the set of measured flow rates are computed as the set is truncated progressively from below. Trajectories of the sample statistics and their derivatives are plotted versus the level of data truncation. In the presence of leaks, these trajectories diverge from their expected theoretical path when plotted on a standardized graph derived from a mixed truncated normal distribution. The point of departure on the standardized graph indicates where the truncation threshold matches the maximum rate of network leakage. A performance limit for the proposed method is derived to account for network size and time averaging. Several examples, based on observed and simulated pipe flows, demonstrate application of the leak detection method.

Water utilities rely on samples collected from the distribution system to assure the quality of drinking water delivered to customers. However, few utilities employ a systematic sampling protocol to collect these samples. This paper presents a methodology that involves random selection of sample locations and time of sample collection to minimize bias and variance in results. For a given sampling goal, system-specific knowledge is incorporated into the sampling design through definition of appropriate stratification variables, selection of sample size, and sample allocation. The sampling designs produced by this methodology provide a measurement of water quality along with an estimate of statistical precision, with insights into significant factors affecting water quality. The methodology is illustrated using a synthetic data set from a distribution system water quality model to assess sampling designs and results for a mid-sized water utility.

Nonpoint source pollution control requires the implementation of best management practices (BMPs) at various locations in a watershed. One important aspect of the watershed strategy is to find the optimal placement and design of BMPs so that their combined effect is most cost-effective. An innovative method is presented for optimizing the placement and configurations of BMPs at the watershed scale. Heuristic optimization techniques are coupled with a watershed model, which is the agricultural Nonpoint Source Pollution model (AnnAGNPS) in this case, and BMP simulation module to find a least cost set of solutions that meet the pollutant load reduction requirements. An impoundment module imbedded in AnnAGNPS, and modified to address the sediment accumulation and resuspension effect, is used to provide the means of evaluating the long-term performance of detention ponds. A scatter search algorithm is applied to finding the least cost solution. The optimization framework developed herein provides a tool for stormwater management practitioners to examine and analyze the treatment efficiencies of stormwater control alternatives and to determine a robust and cost-effective design of stormwater treatment systems.

A mixed-integer quadratic programming problem is proposed to locate booster stations and to identify their dosage schedules for maintaining disinfectant residual in drinking water distribution systems. Optimal booster locations and disinfectant injections are found by minimizing the second moment of the system residual spatiotemporal distribution. Application of the method to an example network studies the trade-off between booster station number and system water quality improvements. As disinfectant is reallocated in the network by an increasing number of well-placed stations, residual spatial and temporal variability is decreased as much as 30 times smaller compared to a single source. Thus greater flexibility in booster system operation leads to more efficient maintenance of residuals at points of water consumption.