This focus area links the research from the prior TA5 areas with the findings from other thrust areas in modeling systems developed collaboratively with water managers. These system dynamic models act as frameworks for integrating physical and social science into decision tools for the management of scarce water resources. The individual research efforts include:

· Drought preparedness plan for the Rio Conchos basin
· Extending URGWOM to include dissolved oxygen using demand-side management
· A surface water-groundwater mass balance model for the Verde River basin using demand-side management
· Evaluating Sierra Vista conservation alternatives using demand-side management
· Dynamic Simulation Model for the Albuquerque basin
· Multi-objective water resources management optimization model for the San Pedro river basin

Drought preparedness plan for the Rio Conchos basin
J. Valdés (UA-CEEM), F. Aparicio (IMTA)

The characterization of droughts poses a significant problem for hydrometeorologists and water resources engineers. One of the reasons is because droughts can last longer and extend across larger areas than other climatic extreme events. Droughts in the Conchos River basin are of great interest to both the U.S. and Mexico because of the International Treaty of 1944 regarding the sharing of the waters of the Rio Bravo/Rio Grande watershed.

The general objective of this research is to create a decision support system (DSS) for the Rio Conchos basin that will help to simulate and improve water resource management and planning. The DSS will help to clarify to the decision-makers what is necessary in order to increase the water supply and reduce the water demand both in the short and long term through the analysis and applications of various scenarios. Drought indices were developed to characterize droughts in semi-arid and arid regions. We also provide statistical approaches to examine the spatial influence of droughts and to estimate the return periods of droughts. The results of this research provide a framework for sustainable water resources management in the basin.

Activities and Results

The previous years were focused on analyzing temporal and spatial extents of droughts as well as estimating the relative frequencies and recurrences of droughts in the Conchos River basin. During the current reporting period the characterization of the droughts was completed with a bivariate representation of the drought severity and duration. In addition a forecasting model was developed that outperforms others statistical models in the application case, the Conchos River Basin. These characterization and forecasting components will be used in the development of a decision support system for the basin using system dynamic techniques. This decision support system will allow the consideration of socio-economic indicators for the basin and for the Lower Rio Grande/Bravo basin in conjunction with the research being carried out by other members of TA5.

In addition to completing the characterization of droughts, a forecasting model for droughts was developed. The forecasting model provides indexed regional drought forecasts for the Conchos River basin. This model is based on dyadic wavelet transforms and artificial neural networks. Preliminary results of the model are promising as may be seen in the following section, and we plan to continue the development of the forecasting model. Improved forecasts of the indexed drought will allow water resources decision makers to develop drought preparedness plans far in advance to mitigate drought impact. Another graduate student is spending his summer getting familiar with this particular catchment.

The coupled (ANN-DD) forecasting model developed in this reporting period provides significant improvements for forecasting PDSI compared to the traditional regression model and popular nonlinear neural networks. The figures below show the performance of the proposed model compared with other forecasting models. The climatological average was used as baseline. Several forecasting skill scores were used to measure the performance of the models. As shown in the figures, the coupled model has a better performance than the other models. One of the reasons is the weakness of other forecasting models to adequately represent the innate long-term memory of the PDSI, which is highly dependent on antecedent soil and atmospheric moisture conditions. Regarding the development of DSS, which will be based in a dynamic simulation model, an initial model representing the main characteristics of the Conchos basin is expected to be finished by September 2002.

Plans

In the first three years, the drought characteristics and predictability in the Conchos River basin have been examined. In Year 4 we expect to finalize the characterization and predictability issues of droughts in semi-arid and arid regions. We will focus on the implications of droughts in the region and on the development of a drought preparedness plan. During Year 4, we plan to determine the frequency of major droughts in the historic record. This will be accomplished by including paleoclimatologic data in the analysis. Using system dynamics, the representation of the basin will be completed and connected with the model developed by IMTA for the Mexican portion of the Rio Grande/Rio Bravo. The next logical step is to continue working on the definition of the principal characteristics and the creation of the DSS. These activities are to be held together with the activities mentioned in sub-area 2 (see above).
(return to top)

Extending URGWOM to include dissolved oxygen, using DSM
K. Lansey, K. Varvel (UA-ENG)

Water resources management decisions affect a broad community and tools to assist all constituents in that community will be valuable in transmitting the rationale for, and understanding the potential impact of, decisions. Dynamic simulation modeling is a tool that can provide a link between technical and scientific information and decision-makers. To demonstrate the utility of this approach, a lumped-sum, mass balance model for simulating surface water flow and dissolved oxygen concentrations on the upper Rio Grande River has been created for technical decision-makers.

Activities and Results

The Upper Rio Grande (URG) water operations group has a total of 8 objectives that are being considered by different subgroups. Water quality in terms of arsenic, algae, and total sediment load is an issue in northern New Mexico, and salinity concerns have been voiced in the southern portion of the state. We have developed a dissolved oxygen simulation model using a dynamic simulation model. A water budget from Cochiti reservoir through Elephant Butte reservoir has been developed and calibrated for a fifteen-year record on a monthly time step. Calibration of the dissolved oxygen component is nearing completion. Kyle Varvel, CEEM MS student, is finalizing his thesis and will defend in August 2002.

The components of the river model (stream segments and reservoirs) have been developed as generic components and can be applied to any system. Mr. Varvel's responsibilities were expanded to develop a first level model of the Rio Conchos for use in a management model with the drought management effort in SAHRA. Data availability is a major difficulty in Mexico, but a simple model has been programmed and will be calibrated for local gains and losses.

The primary result of this year has been the near completion of the Upper Rio Grande Water Quality model. The range of the model is the Cochiti reservoir to Elephant Butte reservoir. Screen shots from the model's mass balance component are shown in Figure 1. Inputs can be easily modified using the slide bars and dials as shown in the figure on the left, while output can be displayed graphically or in tables (right figure). This model allows operators to consider the impacts of water quality when making release and diversion decisions.

As discussed above, the components of this model are written in general form in order to be easily applied to other locations, e.g., Rio Conchos. The water quality model mass balance component will serve as the basis for salinity modeling in the URG. It is anticipated that the water quality model can also be transformed to represent salinity. If desired, the basis model can be extended to the entire Rio Grande basin. Such large-scale modeling may be helpful to understand mechanisms.

The major contribution of this research direction has been the adoption of dynamic simulation (DS) as an integral component of SAHRA. Several investigators have been involved in this area and, as noted, DS should provide a mechanism to link SAHRA science and water resources decision making. The current emphasis is on assisting decision makers in planning for a sustainable water supply. This ability is particularly critical for the State of Arizona.

Plans

The general use of dynamic simulation (DS) can link SAHRA science with stakeholders. In the short term, we propose that the primary application area for DS modeling/water resources management will be the Upper San Pedro (USP) watershed in southeast Arizona. This application will complement the Sandia URG work in New Mexico.

During Year 4 a dynamic simulation (DS) model will be developed that includes potential conservation measures. Modules will be written for water balance and costs. Appropriate interfaces will be developed in collaboration with the USPP report. During Year 5, the DS model will be extended to include response function relationships for USPP most preferred groundwater models and riparian zone integrity measures. Interfaces will be extended as needed and reported. Future efforts will be to extend the model to include alternative groundwater model relationships. This option is desired by USPP due to lack of agreement on best model. Sensitivity and uncertainty analyses also are desired by USPP members.
(return to top)

A surface-groundwater mass balance model for the Verde River basin using dynamic simulation modeling (DSM)
K. Lansey, S. Bowen (UA-ENG)

Water resources management decisions affect a broad community, and tools to assist all constituents in that community will be valuable in transmitting the rationale for, and understanding the potential impact of decisions. Dynamic simulation modeling is a tool that can provide a link between technical and scientific information and decision makers. The DSM model approach is being applied for policy level settings in the Verde basin. The Verde basin model is slow in developing due to time commitments of an unfunded graduate student.
Activities and Results

Population growth in the Prescott area has stressed the limited available water resources. Several groups have formed with the goal of water resources management. They are receiving additional support from the Arizona Department of Water Resources (ADWR) to become better organized. We have discussed the development of a decision support system based upon a water balance model developed in dynamic simulation. They are receptive and willing to collaborate. The initial effort is focusing in the Prescott Active Management Area. After initial discussions, this project has proceeded very slowly due to limited availability of Ms. Bowen. The major effort has been collection of water availability and supply data for the region.

The major contribution of this research direction has been the adoption of dynamic simulation (DS) as an integral component of SAHRA. Several investigators have been involved in this area and, as noted, DS should provide a mechanism to link SAHRA science and water resources decision making. The current emphasis is on assisting decision makers plan for sustainable water supply. This ability is particularly critical for the State of Arizona.

Plans

Work within the Verde basin will continue as student time and availability permits. A basic water balance model that can identify critical components and possible solutions continues to be the first objective of this work.

In the original proposal, a main objective of SAHRA was to assist communities with water resources planning. Unbiased science to improve these decisions is clearly a major contribution and we are working toward that end. A long-term concern is the relative emphasis on science versus tools for decision makers and stakeholder involvement. If water resources remain a priority, water resources planning concerns exist in other regions within Arizona and can provide long-term connections in the state. The Verde River basin is water-limited in certain sectors, and water allocation and impacts of groundwater pumping on riparian zones and endangered species is of interest to many constituents. The Phoenix and Tucson regions and their local utilities must also make planning decisions about how best to use their resources and maintain a sustainable water supply. See for example David Modeer's article in H20 Tucson (Tucson Water Newsletter). As an acceptable understanding (from a decision maker perspective) is achieved in the San Pedro basin, these other locations offer opportunities for SAHRA to have an impact within Arizona.
(return to top)

Evaluating Sierra Vista conservation alternatives using dynamic simulation modeling (DSM)
K. Lansey (UA-ENG), J. Roach (UA-HWR)

Water resources management decisions affect a broad community and tools to assist all constituents in that community will be valuable in transmitting the rationale for, and understanding the potential impact of decisions. Dynamic simulation modeling is a tool that can provide a link between technical and scientific information and decision makers. The DSM approach is being applied for policy level settings in the San Pedro river basin.
Activities and Results

Significant time has been spent building links with the Upper San Pedro Partnership and developing their buy-in to the modeling approach. Several proposals will be submitted in early July to extend the San Pedro work. The large-scale San Pedro model can serve as a link between the science elements within SAHRA and with the science results and decision-makers.

The Upper San Pedro Partnership (USPP) is a consortium of 20 agencies and interest groups that are resolved to managing the limited water resources in their watershed. Water demands are dominated by domestic, industrial, and environmental uses. Decisions for short- and long-term allocations are pressing. SAHRA and others are improving knowledge of water demands and distribution. With Dr. David Goodrich's assistance, we have begun a dialogue on their needs for a general decision support system for regional water management decisions. The USPP is very receptive to the dynamic simulation modeling and contributing to the development of a model for their region. The first question that they are interested in answering is what conservation measures should be instituted to move toward safe yield. Jesse Roach (a UA graduate student) developed a prototype model for demonstration purposes. He has shifted from this project and a new student has not been identified for summer/fall 2002.

The major contribution of this research direction has been the adoption of dynamic simulation (DS) as an integral component of SAHRA. Several investigators have been involved in this area and, as noted, DS should provide a mechanism to link SAHRA science and water resources decision making. The current emphasis is on assisting decision makers plan for sustainable water supply. This ability is particularly critical for the state of Arizona.

Plans

The overall DSS development will be completed in the five tasks described below. Each task requires two major efforts: data collection and computer programming. Valid input information and model calibration is critical for DSS acceptance and use. The USGS, USPP, ADWR, University researchers, and local communities have collected water availability and consumption data for over 30 years. Within each task detailed below, appropriate data will be documented by source and accuracy in the DSS database. Most information will be static and stored in spreadsheets. If necessary, more complex databases that are compatible with the dynamic simulation model will be used.

The first four tasks are related to the development of the DSS that links water consumption with environmental demands. The primary water use (outside of riparian ecosystem demands) is municipal use. The USPP has developed a set of alternative conservation measures that are being examined to bring the community to a safe yield without groundwater mining. The first major effort is to provide a tool that will allow evaluation of these alternatives including their costs (Tasks 1 and 2). The dynamic simulation model will allow selection of different management options, such as reducing residential demand, water reuse, and allowable population growth.

Using a dynamic simulation modeling platform, a graphical interface will be developed to modify input and key evaluation criteria can be displayed in several formats. Simple user interfaces such as slider bars can be moved to define a desired input (see Part I). The model is then executed by a single mouse click and the results are displayed in graphical form for this example. Total consumption and aquifer overdraft can be shown as a function of time. Based on results, inputs can be quickly modified to examine alternative decisions and re-executed. These interfaces can be tailored with graphics, photographs, etc. to be more visually appealing. As noted, dynamic simulation is not simply representing a water balance; other information can be presented in alternative forms such as tables or single indices (e.g., total cost, water reduction/dollar of investment, riparian zone condition for various locations).

The remainder of the DSS component of this project is to enhance the basic model through improved spatial detail and incorporation of uncertainty. Task 3 will focus on spatial disaggregation of groundwater withdrawal and recharge. This effort will provide answers to the question of the impact of municipal use on the riparian corridor. The interaction between groundwater inflow/outflow and groundwater levels in the riparian zones will be determined using available groundwater models and imbedded in the DSS (Task 3). Based upon the groundwater levels, riparian zone condition will be evaluated using the relationships developed in first component of this study (Task 4). This task will then provide a direct link between the water use and riparian zone sustainability.

Task 5 is listed as a final separate task but it will be completed throughout the project. The data input/output interface will also be developed in conjunction with the USPP to best present the desired model results.

This work will link with ongoing work in the San Pedro basin including work on riparian integrity, data collection, and groundwater modeling. As noted above, this type of model can serve as the integration tool to bring the various science projects together. In addition to the clear connection by including models (or simplifications of them) in the DS model, other aspects of coordination will be the variability of annual groundwater recharge and generating of time series of future basin inflows.

During Year 4, a dynamic simulation (DS) model will be developed that includes potential conservation measures. Modules will be written for water balance and costs. Appropriate interfaces will be developed in collaboration with a USPP report. During Year 5, the DS model will be extended to include response function relationships for USPP's most preferred groundwater models and riparian zone integrity measures. Interfaces will be extended as needed and reported. Future efforts will be to extend the model to include alternative groundwater model relationships. This option is desired by USPP due to lack of agreement on the best model. Sensitivity and uncertainty analyses are also desired by USPP members.
(return to top)

Dynamic simulation model for the Albuquerque basin
V. Tidwell (Sandia), E. Webb (Sandia), G. Woodard (UA-HWR)

There is an urgent need to assure sustainable water supplies to meet demands, which are increasing with population, affluence, and environmental concerns. Structural, technological solutions, although necessary, are increasingly costly and take time to develop and implement. On a more immediate basis, efficient water management practices must be pursued. This is made more difficult by the highly dispersed responsibility for resource management, resulting in a piecemeal approach. Additionally, management practices must fully appreciate the interconnections, feedbacks, and time delays among watershed subsystems (e.g., precipitation, runoff, groundwater discharge, evapotranspiration, pumping, recharge) that operate over a range of spatial and temporal scales. To better address these problems, adopting a holistic, systems-level approach is being adopted. Specifically, systems dynamics is used as a decision support framework in which we integrate the natural system with other systems including economics, demographics, and ecology. The resulting decision models are couched in the context of the legal, political, and social constraints that limit the decision process. Currently, project participants are developing a prototype decision support model for the Middle Rio Grande Basin. This model incorporates the basic components of water supply (basin inflow, recharge, evaporative losses, etc.) and water demand (municipal, agricultural, and industrial).

Activities and Results

Enhancements of the prototype decision support model for the Middle Rio Grande Basin that have been completed or are well under way include:

a) Improving the evapotranspiration functions
i) Establishing cooperative relationship with the US Bureau of Reclamation, US Army Corps of Engineers, New Mexico Interstate Stream Commission, City of Albuquerque, and other stakeholders and decision makers.
ii) Calibrating the model against historical data for basin outflows and groundwater depletion.
b) Developing integrated user-friendly interfaces to allow real-time analysis of alternative water management practices in the basin.
c) Converting the model to the current software version being used by other SAHRA researchers.

Specific results cannot yet be reported, as the research is still ongoing. However, perhaps the most important achievement of this project is that the model generated considerable interest among SAHRA researchers as an integrating approach and a DSS tool. It catalyzed half a dozen other DSM efforts involving modeling of various water parameters in the Upper and Middle Rio Grande, the Upper San Pedro, the Conchos, and the Upper Verde.

Plans

The model is by definition an integrative device. It is generating connections with the NMT work on salinity in the Middle Rio Grande (Phillips, Hogan, et al), and with water demand work at UNM (Brookshire et al).

In the forthcoming year, we intend to increase the spatial area covered by the model, including downstream to Elephant Butte Reservoir. We will also change the time step of the model from annual to monthly to incorporate seasonal effects. The model will be improved by adding economic relationships and by adding at least one water quality parameter (salt). The model will be published on the Web, and we will host a workshop in fall 2002 for researchers throughout the western U.S. who are using DSM to model natural resources, including water.
(return to top)

Multi-objective water resources management optimization model for the San Pedro river basin
W. Yeh, J. McPhee (UCLA)

This research has two main goals. The first goal is to develop a Decision Support System (DSS) that merges knowledge from diverse disciplines and simulation tools in a practical, meaningful decision-making environment. The second goal is to assess the reliability of management decisions made using DSS and to evaluate data sufficiency for management purposes. Our research approach combines groundwater simulation with multi-objective optimization to generate trade-off curves among competing objectives in groundwater management. Multi-objective decision-making techniques can be used to select the most desirable pumping policies. Our research aims to developing a method to evaluate uncertainty in the decision space with respect to parameter uncertainty, which in turn is related to the quantity and quality of data used to calibrate the model. Key results for the reporting period include the establishment of a working relationship with the Upper San Pedro Partnership and the formulation of the multi-objective optimization problem to be used in the management model.

Activities and Results

In the first three years of our research, we have selected the San Pedro River Basin, in the U.S. Southwest, as the candidate basin for the development of a DSS. The overall framework for the DSS has been built, linking a simulation model with an optimization model. The linkage uses the response matrix to replace the groundwater flow model in the optimization problem. A multiobjective management model with three conflicting objectives has been proposed and submitted to the main stakeholder organization in the Upper San Pedro River Basin, the Upper San Pedro Partnership (USPP). Potential for integration with other SAHRA researchers has been identified, and research activities are being coordinated between the USPP, UA and UCLA. Our research goals and the way in which they are being carried out aim at bridging the gap between sciences, decision-makers and planning. The DSS will integrate knowledge from diverse disciplines such as hydrology, numerical modeling, social science and operations research.

Activities during Year 3 have focused mostly on communicating our research plans to Basin stakeholders and decision managers, and in formulating a multi-objective optimization problem that reflects the needs and priorities of these actors, for inclusion in the DSS environment. To formulate a multi-objective function that incorporates hydrologic, economic and ecological values, we need to identify desirable states of the hydrologic system, and express them in mathematical terms and as a function of the output of the simulation model. In the initial stage of management model formulation, feedback from stakeholders is fundamental to ensure that research results will be meaningful and useful. A field trip to Arizona and Sonora was conducted in November 2001. On that trip, we attended a meeting of the advisory committee of USPP, where initial contacts were made. The SAHRA Annual meeting in February 2002 provided a forum to enhance those contacts and further acquaint USPP representatives to the nature of our research. This led to a one-hour presentation made to the USPP Technical Committee on April 2002. The presentation allowed us to describe our research, in detail, to representatives of most of the 18 member agencies of the USPP.

Potential for integration with other SAHRA researchers was identified. Based on system dynamics, Kevin Lansey at UA is working on a DSS for the San Pedro Basin. The UA model is intended to evaluate broad water demand and supply decisions using lumped functional relations at a regional scale to evaluate fluxes and storage changes. Both approaches may be coupled to perform fast, interactive analysis with the UA model, and detailed pumping policy optimization with the UCLA model.

The outcome of the above meetings and presentations is a management model proposal that was submitted to the USPP in May 2002. The proposed management model includes: 1) an economic objective that minimizes costs of supply, treatment and conservation; 2) an ecological objective that maximizes groundwater heads in specified locations so that riparian vegetation is preserved; and 3) a maximum pumping objective that seeks to evaluate explicitly the trade-off of pumping against the two other objectives.

Although initially groundwater quality objectives were considered in the analysis, our interaction with stakeholders suggests that water quality is not a relevant issue in the U.S. portion of the basin. Therefore, our initial research approach has been updated and will include only groundwater flow simulation.

Plans

Once the management model is fully developed and linked with the simulation model, it will be possible to obtain a set of non-inferior solutions and find desirable pumping policies, taking into account the multiple conflicting objectives that have been identified. Since decisions made with DSS are based on results obtained using a simulation model, the next logical step is to assess the reliability of those decisions by evaluating the parameter uncertainty that affects the simulation results.

The concept of Management Equivalent Identifiability - MEI - (Sun and Yeh, 1990) will be extended to the San Pedro Basin, and to the case in which model parameters are correlated. MEI purports determining sets in the parameter and decision spaces, so that all acceptable parameter vectors lead to acceptable decision vectors. Acceptable parameter vectors are those that provide a good fit of the simulation model to observed data. Acceptable decision vectors need to be defined in terms of the optimal policy uncertainty that decision-makers are willing to allow.

Parameter uncertainty is closely linked to the quantity and quality of data used for model calibration; model reliability analysis leads to conclusions about the sufficiency of data for management purposes. If model parameter uncertainty leads to unacceptable dispersion in the decision space, and on the other hand the calibrated model adequately fits the historical data, then we can conclude that data is insufficient for calibration from a management point of view.

We will develop a method for experimental design that considers management objectives. Experimental design for parameter estimation traditionally seeks to minimize some type of norm of the parameter covariance matrix, which in its approximation is a function of the Jacobian matrix of state variables with respect to model parameters. First-order approximation will be used to derive the covariance matrix of management decisions, as a function of the sampling design. Therefore, it is possible to specify sampling locations so as to minimize uncertainty in the management decisions obtained with the DSS.

This research should continue to be supported by SAHRA because it aims to provide a powerful tool for information management, specifically developed for the sustainable water resources management in the southwest U.S. and other semi-arid areas. The DSS is capable of merging ongoing research on riparian vegetation demands, hydrology, human demand scenarios, etc., and provides an opportunity for narrowing the gap between practitioners and scientists. Uncertainty analysis is a major concern in engineering and science research, since models are broadly used in decision making in areas such as water right allocations, zoning, risk analysis, etc. Therefore, understanding the implications of data quantity and quality over water resources management decisions has important consequences from the point of view of resource allocation and conflict resolution.

The project integrates with at least three major areas of research within SAHRA. First, research on water demands of riparian vegetation at the San Pedro River Riparian Corridor can be inmput as constraints or objectives to the management model. If functional relations are developed linking water table levels, vegetation areal coverage, etc., objective functions involving the extent of the San Pedro Riparian National Conservation Area could be developed. Second, current research on DSS using system dynamics simulation (Kevin Lansey at UA) can be successfully coupled with our proposed DSS, since each of them deals with the problem of water management at different spatial scales, with different levels of aggregation. Third, demand-side water management studies being carried out by SAHRA researchers will provide potential demand scenarios, within which our DSS can be run, so as to optimize particular features regarding pumping and recharge policies.