Journal of Water Resources Planning and Management

A decision support system (DSS) is presented for conjunctive management of surface water and ground water under prior appropriation. The DSS is constructed around the generalized river basin network flow model MODSIM, providing an open architecture allowing access to input and output databases and modification and verification at all levels of the modeling process. The graphical user interface for the MODSIM DSS provides spatially referenced database capabilities whereby the user can create and link river-basin network objects on the display and populate and import data for that object interactively. Geographic information system tools are used to prepare grid-based spatial data for input into MODRSP, a modified version of the U.S. Geological Survey (USGS) three-dimensional finite-difference ground water model MODFLOW. Response functions generated by MODRSP are provided to MODSIM for simulating spatially varied and time-lagged return/depletion flows from stream-aquifer interactions. Capabilities of the MODSIM DSS are demonstrated on a case study for a portion of the Lower South Platte River Basin, Colorado. Results of the case study indicate significant differences between using ground water response coefficients developed from preassigned stream depletion factor (SDF) values, as currently used in the basin, and those generated using a finite-difference ground water model.

A fuzzy waste-load allocation model, FWLAM, is developed for water quality management of a river system using fuzzy multiple-objective optimization. An important feature of this model is its capability to incorporate the aspirations and conflicting objectives of the pollution control agency and dischargers. The vagueness associated with specifying the water quality criteria and fraction removal levels is modeled in a fuzzy framework. The goals related to the pollution control agency and dischargers are expressed as fuzzy sets. The membership functions of these fuzzy sets are considered to represent the variation of satisfaction levels of the pollution control agency and dischargers in attaining their respective goals. Two formulations—namely, the MAX-MIN and MAX-BIAS formulations—are proposed for FWLAM. The MAX-MIN formulation maximizes the minimum satisfaction level in the system. The MAX-BIAS formulation maximizes a bias measure, giving a solution that favors the dischargers. Maximization of the bias measure attempts to keep the satisfaction levels of the dischargers away from the minimum satisfaction level and that of the pollution control agency close to the minimum satisfaction level. Most of the conventional water quality management models use waste treatment cost curves that are uncertain and nonlinear. Unlike such models, FWLAM avoids the use of cost curves. Further, the model provides the flexibility for the pollution control agency and dischargers to specify their aspirations independently.

Geographic Information Systems (GIS) technology is used to develop automated methods for quantifying the spatial variability of flood hazard. An interface module developed within GIS incorporated flood water elevations predicted from a hydraulic simulation, Ontario depth/damage curves, and simulation functions for evaluating alternative flood control policies. A case study from the town of Dundas, Ontario, Canada, is used to demonstrate the advantages that are gained by the application of this module. The proposed approach will provide water resources managers with improved insight into flooding conditions, strengthening the risk assessment process and the administration of human activities in river floodplains.

Booster disinfection is the addition of disinfectant at locations distributed throughout a water distribution system. Such a strategy can reduce the mass of disinfectant required to maintain a detectable residual at points of consumption in the distribution system, which may lead to reduced formation of disinfectant by-products in particular trihalomethanes. Here an optimization model is formulated for the dynamic schedule of disinfectant injections; this schedule minimizes the total dose required to satisfy residual constraints over an infinite-time horizon. This infinite-time problem is reduced to a solvable finite-time optimal scheduling model by assuming periodicity of mass injections and network hydraulics. Furthermore, this model is linear since the principle of linear superposition is shown to apply to disinfectant concentrations resulting from multiple disinfectant injections over time. A matrix generator code was developed to interface with the EPANET network water quality model. This code automatically generates the linear programming formulation of the optimal scheduling model, which is then solved using the simplex algorithm. Results from application of the model suggest that booster disinfection can reduce the amount of disinfectant required to satisfy concentration constraints, when compared to conventional disinfection only at the source. The optimal booster schedule reduced the average disinfectant concentration within the distribution system and, in some cases, the variability of these concentrations. The number of booster stations, booster location, and distribution system hydraulics were shown to affect the optimal schedule.

State-space and multiple regression methods were compared with each other and with simple monthly averages for the accuracy of their short-term forecasts of urban water demand. Seven sets of 24 monthly forecasts of water demand were computed. Each set is based on a different 7-year historic period, using a total of 15 years of monthly data. Based on a variety of measures of forecast error, the state-space models exhibited less bias than the other models, whereas the size of a typical forecast error was about the same for state-space and simple monthly averages. Forecast errors showed considerable variability within both state-space and multiple regression. The mean absolute forecast error ranged from 7.4 to 14.8% for multiple regression, and from 3.6 to 13.1% for state-space. For this sample data, the multiple regression model forecasts were least accurate and also had larger biases than the other methods.