Case Study: Model to Simulate Regional Flow in South Florida

South Florida has a complex regional hydrologic system that consists of thousands of miles of networked canals, sloughs, highly pervious aquifers, open areas subjected to overland flow and sheet flow, agricultural areas and rapidly growing urban areas. This region faces equally complex problems related to water supply, flood control, and water quality management. Advanced computational methods and super fast computers alone have limited success in solving modern day problems such as these because the challenge is to model the complexity of the hydrologic system, while maintaining computational efficiency and acceptable levels of numerical errors. A new, physically based hydrologic model for South Florida called the regional simulation model (RSM) is presented here. The RSM is based on object oriented design methods, advanced computational techniques, extensible markup language, and geographic information system. The RSM uses a finite volume method to simulate two-dimensional (2D) surface and groundwater flow. It is capable of working with unstructured triangular and rectangular mesh discretizations. The discretized control volumes for 2D flow, canal flow and lake flow are treated as abstract “water bodies” that are connected by abstract “water movers.” The numerical procedure is designed to work with these and many other abstractions. An object oriented code design is used to provide robust and highly extensible software architecture. A weighted implicit numerical method is used to keep the model fully integrated and stable. A limited error analysis was carried out and the results were compared with analytical error estimates. The paper describes an application of the model to the L-8 basin in South Florida and the strength of this approach in developing models over complex areas.     

Probabilistic Approach for Design and Hydrologic Performance Assessment of Reconstructed Watersheds

The oil sands mining industry in Canada has made a commitment to reclaim mining areas to an equivalent capability to that which existed prior to mining. An essential requirement in the design of reclamation covers to meet this objective is that all covers must have a sufficient available water holding capacity (AWHC) in order to supply sufficient moisture for vegetation over the summer moisture deficit typical in the region. AWHC is currently based on static evaluations of wilting point and field capacity under a constant annual evapotranspiration demand. This paper presents an alternative probabilistic approach by which the hydrologic performance of these reclamation soil covers can be assessed. A field-calibrated water balance model is used along with the available historical meteorological record to estimate the maximum soil moisture deficit that a soil cover is able to sustain over the growing season. Frequency curves of the maximum annual moisture deficit are used to assess the probability that a cover is able to provide any particular threshold of moisture demand. The method also allows for a quantification of the predictive uncertainty of the model. The predictive uncertainty is used as a margin of safety to estimate a design value of moisture deficit for various alternative cover designs. This paper recommends procedures for a frequency-based assessment and design of reclamation soil covers in the oil sands industry. This method takes into account climatic variability as well as parameter uncertainty in estimating the soil moisture deficit.     

Modeling Evapotranspiration of Two Land Covers Using Integrated Hydrologic Model

Modeling evapotranspiration (ET) distribution in shallow water table environments is of great importance for understanding and reproducing other hydrologic fluxes such as runoff and recharge. Unfortunately, ET distribution can be the most difficult hydrologic process to analyze. The partitioning of ET into upper zone ET, lower zone ET, and groundwater ET is complex because it depends on land cover and subsurface characteristics. One comprehensive distributed parameter model, integrated hydrologic model (IHM), builds on an improved understanding and characterization of ET partitioning between surface storages, vadose zone storage, and saturated groundwater storage. It provides a smooth transition to satisfy ET demand between the vadose zone and the deeper saturated groundwater. In this paper, the IHM was used to analyze ET contribution from different regions of the vadose zone and saturated zone. Rigorous testing was done on two distinct land covers, grass land and forest land, at a study site in West-Central Florida. Sensitivity analysis on the key parameters was investigated and influence of parameters on ET behavior was also discussed. Statistics with the root mean square error and mean bias error for forest total ET were about 1.46 and 0.04 mm/day, respectively, and 1.61 and 1.07 mm/day for grass total ET. Modeling results further proved that ET distributions from the upper and lower soil and water table, while incorporating field-scale variability of soil and land cover properties, can be predicted reasonably well using IHM model.     

Retrospective Comparison of Watershed Analysis Risk Management Framework and Hydrologic Simulation Program Fortran Applications to Mica Creek Watershed

As part of an ongoing watershed model comparison program for forested watersheds, Watershed Analysis Risk Management Framework (WARMF V5.18) and Hydrologic Simulation Program Fortran (HSPF V10) were independently applied to the Mica Creek Watershed in Idaho. A comprehensive model comparison was made in terms of watershed delineation, hydrologic formulations, model parameterization, meteorological data, hydrologic calibration, and hydrologic verification. Comparison was not made for water quality, which was not simulated in the HSPF application. It was concluded that WARMF is a mechanistic model structured to simulate the hydrologic processes, whereas HSPF is an empirical water budget model. The WARMF is suitable for application to forested watersheds. It successfully predicted stream flows comparable to measured values. The HSPF results were also good, if one ignores an unrealistic amount of water loss to inactive groundwater and an empirical treatment of rain-on-snow events.     

Bayesian Model Calibration Using Geotechnical Centrifuge Tests

The predicted performance using a geotechnical prediction model is expected to deviate from reality. A practical approach to assess the model error is through calibration with observed performances in physical model tests. In this paper, a Bayesian framework of model calibration using centrifuge modeling tests is proposed and the procedure of model calibration is illustrated. Two centrifuge tests conducted to investigate the performance of soil slopes under rainfall conditions are used to calibrate a coupled hydromechanical analysis model. It is found that for centrifuge tests with different levels of soil variability, the test with a smaller variability of soil properties is more efficient for model calibration. According to the concept of random field, a centrifuge model with a larger model size and accelerated to a lower acceleration is better for model calibration. When the discrepancy between the performance interpreted from the centrifuge model and the field performance is small, the improvement of the reliability estimation for a new slope is significant. However, when there is little information about the discrepancy, the reliability estimation cannot be significantly improved by the information from centrifuge modeling. The proposed procedure is shown to be able to quantify the calibration effects of centrifuge tests and may be used to achieve a more reliable calibration.     

Groundwater Modeling for the Analysis of Active Slow-Moving Landslides

Active slow-moving landslides in clayey soils exhibit continuous movements generally controlled both in the accelerating and decelerating phases by the pore-water pressure regime that, in turn, is strictly correlated to the net rainfall regime. The paper stresses the importance of a reliable groundwater model to predict these types of movements. To this aim a procedure is proposed to define the transient groundwater regime in the slope on the basis of recorded rainfall and monitoring data; the model is then used to derive the time-dependent shear strength along the main slip surfaces. The displacements at selected points along the slip surface are computed using a phenomenological (i.e., empirical) relationship between the local factor of safety and the displacement rate at those points. The procedure is employed for the analysis of a well-documented case history: the Porta Cassia landslide (central Italy).     

Image-Based Machine Learning for Reduction of User Fatigue in an Interactive Model Calibration System

The interactive multiobjective genetic algorithm (IMOGA) is a promising new approach to calibrate models. The IMOGA combines traditional optimization with an interactive framework, thus allowing both quantitative calibration criteria as well as the subjective knowledge of experts to drive the search for model parameters. One of the major challenges in using such interactive systems is the burden they impose on the experts that interact with the system. This paper proposes the use of a novel image-based machine-learning (IBML) approach to reduce the number of user interactions required to identify promising calibration solutions involving spatially distributed parameter fields (e.g., hydraulic conductivity parameters in a groundwater model). The first step in the IBML approach involves selecting a few highly representative solutions for expert ranking. The selection is performed using unsupervised clustering approaches from the field of image processing, which group potential parameter fields based on their spatial similarities. The expert then ranks these representative solutions, after which a machine-learning model (augmented with the spatial information of the selected fields) is trained to learn user preferences and predict rankings for solutions not ranked by the expert. To better mimic the “visual” information processing of human experts, algorithms from the field of image processing are used to mine information about the spatial characteristics of parameter fields, thus improving the performance of the clustering and machine-learning algorithms. The IBML approach is tested and demonstrated on a groundwater calibration problem and is shown to lead to significant improvements, reducing the amount of user interaction by as much as half without compromising the solution quality of the IMOGA.     

Calibration and Verification of a 2D Hydrodynamic Model for Simulating Flow around Emergent Bendway Weir Structures

The predictive capability of a two-dimensional (2D)-hydrodynamic model, the finite-element surface water modeling system (FESWMS), to describe adequately the flow characteristics around emergent bendway weir structures was evaluated. To examine FESWMS predictive capability, a sensitivity analysis was performed to identify the flow conditions and locations within the modeled reach, where FESWMS inputs for Manning’s n and eddy viscosity must be spatially distributed for to better represent the river bed flow roughness characteristics and regions where the flow is highly turbulent in nature. The sensitivity analysis showed that high flow conditions masked the impact of Manning’s n and eddy viscosity on the model outputs. Therefore, the model was calibrated under low flow conditions when the structures were emergent and had the largest impact on the flow pattern and model inputs. Detailed field measurements were performed under low flow conditions at the Raccoon River, Iowa for model calibration and verification. The model predictions were examined for both spatially averaged and distributed Manning’s n and eddy viscosity model input values within the study reach for an array of emergent structures. Spatially averaged model inputs for Manning’s n and eddy viscosity provided satisfactory flow depth predictions but poor velocity predictions. Estimated errors in the predicted values were less than 10% for flow depth and about 60% for flow velocity. Distributed Manning’s n and eddy viscosity model inputs, on the contrary, provided both satisfactory flow depth and velocity predictions. Further, distributed inputs were able to mimic closely the recirculation flow pattern in the wake region behind the bendway weir structures. Estimated errors in the predicted values were less than 10 and 25% for flow depth and velocity, respectively. Overall, in the case of distributed model inputs, FESWMS provided satisfactory results and allowed a closed depiction of the flow patterns around the emergent bendway weirs. These findings suggest that 2D models with spatially distributed values for Manning’s n and eddy viscosity can adequately replicate the velocity vector field around emergent structures and can be valuable tools to river managers, except in cases when detailed three-dimensional flow patterns are needed. The study was limited to the examined low flow conditions, and more field data, especially under high flow conditions, are necessary to generalize the findings of this study regarding the model prediction capabilities.     

Modifications to SCS-CN Method for Long-Term Hydrologic Simulation

The original soil conservation service curve number (SCS-CN) technique is primarily used to transform daily rainfall into surface runoff by assuming the proportionality between retention and surface runoff based on a parameter referred to as curve number (CN). The conventional method does not take into account the temporal and spatial variability of curve number. In this paper, an attempt has been made to modify the existing SCS-CN model in two ways by varying the CN using antecedent moisture condition (designated as Model I), and by using antecedent moisture amount (designated as Model II). The daily moisture storage is updated based on varying the curve number and other hydrologic abstractions. These two different models are constructed to compute streamflow components: Direct surface runoff, base flow, and hydrological abstractions. These methodologies were successfully applied to daily data of catchments of Cauvery, Narmada, Ganga, and Ulhas Rivers, lying in different climatic regions of India, and the results were analyzed. Application of Model I to Hemavati (a tributary of River Cauvery, Karnataka State) data yielded maximum efficiency of 84% in calibration, and minimum efficiency of 54% with Ramganga (a tributary of River Ganga, Uttaranchal State) data, whereas Model II showed maximum efficiency of 85% in Hemavati catchment and minimum efficiency of 64% in Kalu catchment (a tributary of River Ulhas, Maharashtra State). Model II performed better than Model I on all four catchments. It is found that the proposed models reasonably simulate the catchment response and these SCS-CN-based models are applicable to complex natured watersheds.     

Parameter Estimation for the Nonlinear Muskingum Model Using the BFGS Technique

In the past, various methods have been used to estimate the parameters in the nonlinear three-parameter Muskingum model to allow the model to more closely approximate a nonlinear relation compared to the original two-parameter Muskingum model. In this study, the Broyden–Fletcher–Goldfarb–Shanno (BFGS) technique, which searches the solution area based on mathematical gradients, is introduced. The technique found the best parameter values compared to previous results in terms of the sum of the square deviation between the observed and routed outflows, using the smallest number of computational iterations. A sensitivity analysis showed that the initial values of certain parameters were critical when finding the optimal solution. Although this gradient-based technique makes use of initial value assumptions and involves complicated calculus, different initial values reach the same optimal or near-optimal solution within less time. Moreover, this mathematical technique does not require the algorithm parameters that are essential factors in meta-heuristics such as genetic algorithm or harmony search. The technique also considers the hydrologic parameters to be continuous rather than discrete variables for pure structures.     

Modeling Nonpoint Source Pollutant Losses from a Small Watershed Using HSPF Model

Hydrologic models play an important role in the assessment of nonpoint source (NPS) pollution, which is essential for the environmental management of water resources. The present study has been undertaken to evaluate the applicability of a physically based continuous time scale, hydrological, and water quality computer model—Hydrologic Simulation Program-Fortran (HSPF)—in simulating runoff and sediment associated NPS pollutant losses from a small mixed type (land under agriculture, shrubs and forest, rocks, grasses) watershed of the Damodar Valley Corporation, Hazaribagh, India. Water soluble NO3–N, NH4–N, and P were considered as pollutants and their concentrations in the runoff were measured at the outlet of the watershed, randomly for 15 dates during the monsoon season (June–October) of 2000 and 2001. The model calibration and validation results reveal that the seasonal trend of HSPF simulated runoff, sediment yield, and NPS pollutants compared reasonably with their measured counterparts. Although the concentrations of pollutants were generally overpredicted for NO3–N and underpredicted for NH4–N and water-soluble P in the month of June when fertilizers releasing NH4–N and P are applied in rice fields, the differences in the mean concentration were not significantly different at a 95% level of confidence. Variation in the simulated losses of water soluble N and P species between the years occurred largely due to differences in the amount and distribution of rainfall. These results indicate that the HSPF model can be used as a tool for simulating runoff and sediment associated NPS pollution losses from the study area.     

Parameter Estimation in Groundwater Hydrology Using Artificial Neural Networks

The capability of artificial neural networks to act as universal function approximators has been traditionally used to model problems in which the relation between dependent and independent variables is poorly understood. In this paper, the capability of an artificial neural network to provide a data-driven approximation of the explicit relation between transmissivity and hydraulic head as described by the groundwater flow equation is demonstrated. Techniques are applied to determine the optimal number of nodes and training patterns needed for a neural network to approximate groundwater parameters for a simulated groundwater modeling case study. Furthermore, the paper explains how such an approximation can be used for the purpose of parameter estimation in groundwater hydrology.     

Watershed-Scale Impacts of Nitrogen from On-Site Wastewater Systems: Parameter Sensitivity and Model Calibration

A numerical watershed model was used to evaluate the potential influence of various point and nonpoint sources including on-site wastewater systems (OWS) on stream nitrate concentration in Turkey Creek Watershed, Colorado. A watershed analysis risk management framework model was used for this study, and was calibrated to observed stream nitrate concentrations using an automatic calibration tool. Parameter sensitivity analysis was done to select critical parameters for calibration and to reduce uncertainty in the simulated results. Sensitivity analysis of nitrate transport and transformation parameters showed that stream nitrate concentration is highly sensitive to cation exchange capacity, nitrification rate, base saturation of ammonium, initial concentration of ammonium in the soil, and some of the crop growth related parameters. The calibrated model was used to evaluate scenarios related to OWS including the impacts of population growth and new development and impacts of conversion of OWS to conventional sewers. The results showed that there would be a significant increase in stream nitrate concentration with increasing population. Conversion of OWS to sewers increased stream nitrate concentration but decreased nitrate concentration in the bottom soil layer indicating that OWS are beneficial with respect to stream nitrate concentration but may increase nitrate concentrations in groundwater.     

Calibration of a Continuous Simulation Fecal Coliform Model Based on Historical Data Analysis

Since 1984, the major water reclamation plants discharging to the Chicago Waterway System (CWS) have not disinfected their effluents. The possible addition of disinfection at these plants is the subject of an ongoing use attainability analysis (UAA). For the UAA, Escherichia coli (E. coli) is used as the indicator of bacterial contamination. However, only a few years of E. coli data are available for the CWS and the treatment plants discharging to the CWS. Thus, it was decided to develop a model based on fecal coliforms for which more data are available and to develop a relation between fecal coliform and E. coli counts for the CWS. A 1:1 relation was found between fecal coliform and E. coli counts in the CWS by Limnotech (2004, written communication) as part of the UAA. In order to evaluate the effects of possible disinfection measures on fecal coliform and related E. coli counts in the CWS, a simple first-order fecal coliform decay model was added to the continuous-simulation flow-water quality model DUFLOW applied to the CWS system. Due to the limited amount (monthly samples) of measured fecal coliform concentration data for the CWS, a reasonable calibration of the model would have been difficult to achieve based on the traditional trial and error method. In this paper, a new concept of model parameter estimation based on historical data analysis and its application to model calibration is presented. The fecal coliform decay rate k was estimated for every reach of the CWS based on analysis of historical data (1990–2003) between each two consecutive sampling locations and the related travel time between these stations. The fecal coliform decay rate then was determined on the basis of many years (14 years, in this case) of monthly fecal coliform samples rather than the few monthly samples taken in a typical calibration period. The results obtained indicate that the calibration process was successful, and a good match between measured and simulated fecal coliform concentrations at almost all locations along the CWS is achieved with one model run for several multiple month periods in 1998, 1999, 2001, and 2002.     

Optimal Sampling Design Methodologies for Water Distribution Model Calibration

Sampling design (SD) for water distribution systems (WDS) is an important issue, previously addressed by various researchers and practitioners. Generally, SD has one of several purposes. The aim of the methodologies developed and presented here is to find the optimal set of network locations for pressure loggers, which will be used to collect data for the calibration of a WDS model. First, existing SD approaches for WDS are reviewed. Then SD is formulated as a multiobjective optimization problem. Two SD models are developed to solve this problem, both using genetic algorithms (GA) as search engines. The first model is based on a single-objective GA (SOGA) approach in which two objectives are combined into one using appropriate weights. The second model uses a multiobjective GA (MOGA) approach based on Pareto ranking. Both SD models are applied to two case studies (literature and real-life problems). The results show several advantages and one disadvantage of the MOGA model when compared to SOGA. A comparison of the MOGA SD model solution to the results of several published SD models shows that the Pareto optimal front obtained using MOGA acts as an envelope to the Pareto fronts obtained using previously published SD models.     

Daily Grass and Alfalfa-Reference Evapotranspiration Estimates and Alfalfa-to-Grass Evapotranspiration Ratios in Florida

Efficient use of natural water resources in agriculture is becoming an important issue in Florida because of the rapid depletion of freshwater resources due to the increasing trend of industrial development and population. Reliable and consistent estimates of evapotranspiration (ET) are a key element of managing water resources efficiently. Since the 1940s numerous grass- and alfalfa-reference evapotranspiration (ETo and ETr, respectively) equations have been developed and used by researchers and decision makers, resulting in confusion as to which equation to select as the most accurate reference ET estimates. Twenty-one ETo and ETr methods were evaluated based on their daily performance in a humid climate. The Food and Agriculture Organization Penman-Monteith (FAO56-PM) equation was used as the basis for comparison for the other methods. Measured and carefully screened daily climate data during a 23-year period (1978–2000) were used for method performance analyses, in which the methods were ranked based on the standard error of estimate (SEE) on a daily basis. In addition, the performance of the four alfalfa-based ET (ETr) equations and the ratio of alfalfa ET to grass ET (Kr values) were evaluated, which have not been studied before in Florida’s humid climatic conditions. The peak month ETo estimates by each method were also evaluated. All methods produced significantly different ETo estimates than the FAO56-PM method. The 1948 Penman method estimates were closest to the FAO56-PM method on a daily basis throughout the year, with the daily SEE averaging 0.11 mm?d?1; thus this method was ranked the second best overall. Although 1963 Penman (with the original wind function) slightly overestimated ET, especially at high ETo rates, it provided remarkably good estimates as well and ranked as the third best method, with a daily average SEE value of 0.14 mm?d?1. Both methods produced peak month ETo estimates closest to the FAO56-PM method among all methods evaluated, with daily peak month SEEs averaging 0.07 and 0.09 mm?d?1, respectively. Significant variations were observed in terms of the performance of the various forms of Penman’s equations. For example, the original Penman-Monteith method produced the poorest ETo estimates among the combination equations, with a daily SEE for all months and peak month averaging 0.50 and 0.35 mm?d?1, respectively and ranked 11th. An average value of 1.18 was used to convert ETr estimates to ETo values for alfalfa-reference methods. The Kr value of 1.18 resulted in reasonable estimates of ETo throughout the year by the Kimberley forms of the Penman equations. Another ETr-based equation, Jensen-Haise, gave consistently poor estimates. The Stephens-Stewart radiation method was the highest-ranked (10th) noncombination method overall. The temperature-based McCloud method (ranked 19th) produced the poorest ETo estimates among all methods with a daily SEE for all months and for the peak month averaging 1.93 and 1.22 mm?d?1, respectively. In general, the results obtained from the temperature methods suggest that all of the temperature methods, with the possible exception of the Turc method, can only be applicable for these climatic conditions after they are calibrated or modified locally or regionally. The FAO and Christiansen pan evaporation methods (ranked 17th and 18th, respectively) produced poor ETo estimates and had the largest amount of point scatter in daily ETo estimates relative to the FAO56-PM ETo. Both methods resulted in the highest daily SEE of 1.18 and 1.19 mm?d?1 for all months, after the McCloud method (1.93 mm?d?1), and with the highest SEE of 1.30 and 1.24 mm?d?1 for the peak month of all methods evaluated. The FAO56-PM method uses solar radiation, wind speed, relative humidity, and minimum and maximum air temperature to estimate ETo. It has been recommended that the FAO56-PM be used for estimating ETo when all the necessary input parameters are available. However, all these input variables may not be available, or some of them may not be reliable for a given location if the FAO56-PM equation is used, and one may need to choose other temperature, radiation, or pan evaporation methods based on the availability of data for estimating ETo. The results of this study can be used as a reference tool to provide practical information on which method to select based on the availability of data for reliable and consistent estimates of daily ETo relative to the FAO56-PM method in a humid climate.     

Single- and Multievent Optimization in Combined Sewer Flow and Water Quality Model Calibration

Over the last three decades, storm-water quality modeling has been used increasingly commonly to describe the general system behavior and assess the pollution loads transferred in and spilled out of combined sewer systems. The calibration of quality models is, in most cases, based on conventionally obtained calibration data, e.g., by automated sampling. Long-term high-resolution online measurement data are available for the Graz West catchment (Graz, Austria), allowing an assessment of the full dynamics of discharge and pollution concentrations. This paper focuses on the application and comparison of single-event and two different multievent optimization schemes for sewer-water quality model calibration. While both single- and multievent optimization lead to satisfying results for the calibration events in discharge calibration, it is shown that validation events are better reproduced by using multievent calibration. Single- and multievent autocalibration of pollution concentration is based on the best dataset obtained from the discharge calibration. As for discharge, the pollutographs are reproduced satisfactorily, and multievent calibration is more stable. In all cases, the two multievent approaches performed equally well.     

Automatic Calibration of the U.S. EPA SWMM Model for a Large Urban Catchment

The Storm Water Management Model was adapted and calibrated to the Ballona Creek Watershed, a large urban catchment in Southern California. A geographic information system (GIS) was used to process the input data and generate the spatial distribution of precipitation. An optimization procedure using the complex method was incorporated to estimate runoff parameters, and ten storms were used for calibration and validation. The calibrated model predicted the observed outputs with reasonable accuracy. A sensitivity analysis showed the impact of the model parameters, and results were most sensitive to imperviousness and impervious depression storage and least sensitive to Manning roughness for surface flow. Optimized imperviousness was greater than imperviousness predicted from land-use information. The results demonstrate that this methodology of integrating GIS and stormwater model with a constrained optimization technique can be applied to large watersheds.     

Load Modeling Approach for Evaluating Selenium Stream Standards Compliance

Elevated selenium concentrations have been observed in several rivers in the western United States and are a concern for healthy aquatic systems. Water quality modeling is a valuable tool for quantifying the importance of sources and assessing management alternatives for stream standards compliance in basins with impaired water quality. This modeling study uses a relatively simple approach to describe a water distribution system and processes that drive loading of dissolved selenium and salts from the naturally occurring Cretaceous shale soils in the extensively irrigated Uncompahgre and Gunnison River Basins in Western Colorado. Calibrated model output characterizes processes that load an average total of 4,130?kg (9,100?lb) of selenium and 2.9×108?kg (316,000?t) of salts per year from agricultural subbasins to the Uncompahgre River. Simulations of best management practices predict that extensive implementation of methods to minimize seepage and constituent loading would be required to reduce selenium concentrations to comply with the current water quality standards.     

SIMGRO, a GIS-Supported Regional Hydrologic Model in Irrigated Areas: Case Study in Mendoza, Argentina

The SIMGRO hydrologic simulation model was extended to include irrigation practice. It could then be used to evaluate the effect of hydrologic changes in an irrigated area in the province of Mendoza, Argentina where, given an average annual rainfall of approximately 200?mm, irrigation is crucial for agriculture. A storage dam was recently constructed in the Mendoza River to control the fluctuating river flow and to guarantee that the demand for water is met throughout the year. The dam will impact on parts of the irrigation system where groundwater levels are already high and salinization occurs. To evaluate these changes and possible mitigation measures, two performance indicators that consider groundwater and surface water were used: Relative evapotranspiration and the depleted fraction. Scenario runs revealed that the irrigation water losses from the canals affect the groundwater levels in the downstream part of the irrigated area; an increase in salinity was also revealed.