Misgana Muleta

Model Performance Sensitivity to Objective Function during Automated Calibrations

Misgana Muleta — Thu, 26 Jul 2012 15:12:51 PDT

Previous studies have reported limitations of the efficiency criteria commonly used in hydrology to describe goodness of model simulations. This study examined sensitivity of model performance to the objective function used during automated calibrations. Nine widely used efficiency criteria were evaluated for their effectiveness as objective function, and goodness of the model predictions were examined using 13 criteria. Two cases (Case I: Using observed streamflow data and Case II: Using simulated streamflow) were considered to accomplish objectives of the study using a widely used watershed model (SWAT) and good-quality field data from a well-monitored experimental watershed. Major findings of the study include (1) automated calibration results are sensitive to the objective function group—group that work based on minimization of the absolute deviations (Group I), group that work based on minimization of square of the residuals (Group II), and groups that use log of the observed and simulated streamflow values (Group III)—but not to objective functions within the group; (2) efficiency criteria that belong to Group I were the most effective when used as objective function for accurate simulation of both low flows and high flows; (3) Group I and Group II objective functions complement each other’s performance; (4) with regard to the capability to describe goodness of model simulations, efficiency criteria that belong to Group I showed superior robustness; (5) for the study watershed, use of the long-term interannual calendar day mean as baseline model did not improve capability of an efficiency criterion to describe model performance; and (6) even for ideal conditions where uncertainty in input data and model structure are fully accounted for, identifying the so-called global parameters values through calibration could be daunting as parameter values that were significantly divergent from predetermined values produced model simulations that can be considered near perfect even when judged using multiple efficiency criteria.

Improving Model Performance Using Season-Based Evaluation

Misgana Muleta — Thu, 26 Jul 2012 15:12:45 PDT

Computer models have become vital decision-making tools in many areas of science and engineering including water resources. However, models should be properly evaluated before use to improve the likelihood of making sound decisions based on their results. The model evaluation technique practiced today in hydrology assumes that model parameters are season insensitive and attempts to identify “optimal” values that would describe watershed behavior during dry and wet seasons. This assumption could compromise accuracy of model predictions. This study demonstrates performance improvement that would be achieved when a season-based model evaluation approach is pursued. A global sensitivity analysis (SA) model has been used to investigate seasonal sensitivity of streamflow parameters of a watershed simulation model on the headwaters of the Little River Watershed, one of the United States Department of Agriculture’s experimental watersheds. Two separate analyses have been performed: the conventional approach in which model parameters are assumed to be season insensitive; and a season-based evaluation in which the influential parameters may vary for months with a low runoff coefficient and months with a high runoff coefficient. The sensitivity analysis helped to identify dominant model and watershed behaviors for the conventional annual approach and for the wet and dry seasons. The SA results show that the influential parameters exhibited modest seasonal sensitivity for the experimental watershed. Model calibration was then performed by using the dynamically dimensioned search (DDS) algorithm for the conventional and season-based approaches using the principal parameters identified by the global SA model. Performance of the calibration attempts have been verified with the traditional split-sampling technique and also by assessing effectiveness of the model in predicting internal watershed behaviors through comparison of simulated streamflow with observations at multiple internal sites not used for calibration. Several efficiency measures have been used to test goodness of the model simulations. The season-based model evaluation technique showed superior performance compared with the traditional method of assuming constant model parameters across seasons. The watershed simulation model exhibited reasonable accuracy in simulating streamflow at the internal sites and for the verification periods when parameter values are allowed to vary from dry to wet season. The “optimal” parameter values identified by the calibration attempts showed significant seasonal sensitivity.

Sensitivity of a Distributed Watershed Simulation Model to Spatial Scale

Misgana K. Muleta et al. — Mon, 21 Nov 2011 12:14:33 PST

The results of distributed watershed models could be sensitive to spatial and temporal scales at which inputs and model parameters are aggregated. This paper reports findings of a detailed sensitivity analysis conducted on the U.S. Department of Agriculture’s distributed watershed simulation model, known as the Soil and Water Assessment Tool (SWAT). The Big Creek Watershed, located in southern Illinois, is used for the study. The model is calibrated to improve accuracy of its streamflow and sediment concentration predictions using observed data at two locations in the study watershed. Streamflow and sediment concentrations that are simulated by the calibrated model at various spatial scales of discritization are extracted and compared, and inputs and model parameters responsible for sensitivity of model responses are identified. Several indices that could be used as indicators of model behavior are also derived. In addition, feasibility analysis of SWAT is conducted to see if the watershed simulation model could be used as a component in future decision support models developed to assist in identifying integrative watershed management practices that control agricultural nonpoint source pollutions from watersheds. The major findings of the study are: (1) accuracy of the raw model output (streamflow and sediment yield) is very poor for all delineations indicating the need for careful model calibration; (2) streamflow is relatively insensitive to spatial scale; and (3) sediment generated and sediment that leaves the watershed decreases as spatial scale gets coarser. Unlike the findings of previous studies, sediment yield significantly varies, even when properties of the outlet channel remain practically the same. (4) SWAT’s estimate of sediment yield is sensitive to human activities conducted in subbasins of the watershed, thus indicating the capability of SWAT to evaluate consequences of alternative watershed management practices.

Multi-objective Design of Transient Network Models

Bong Seog Jung et al. — Mon, 21 Nov 2011 12:14:29 PST

The optimal design of a water distribution system under transient conditions is formulated as a two-objective optimization problem. The objectives are minimization of the total pipe costs and maximization of the hydraulic reliability for the transient network design model. Unlike most optimization models in which demands are set to their end-of-life levels, this approach assumes that the demand loadings vary throughout the design life of the system. Evolutionary algorithms are applied to support efficient search for Pareto optimal solutions to the dual-objective optimization problem. An example application is presented and relevant conclusions are stated.

Using Genetic Algorithms and Particle Swarm Optimization for Optimal Design and Calibration of Large and Complex Urban Stormwater Management Models

Misgana K. Muleta et al. — Mon, 21 Nov 2011 12:14:23 PST

Computer models are vital for the evaluation and management of urban drainage systems. Usefulness of these models, however, depends on how well they are calibrated. Properly calibrated models can be used to conceive, evaluate and compare various design improvement alternatives. Unfortunately, calibration and design of urban stormwater models, especially with the commonly used trial-and-error approach, are an expensive, time-consuming process and normally represent the most painful step of a modeling exercise. Their success depends mainly on the engineering expertise of the modeler and budget availability. The effort is complicated by the fact that these models normally necessitate the evaluation of a large number of variables and parameters in order to adequately describe the complex relationships that exist between rainfall, runoff, watershed characteristics, and system hydraulics in an urban setting. The trial-and-error evaluation of all calibration and design/improvement options is therefore unlikely to be practically feasible or manageable, and even knowledgeable modelers often fail to obtain good results. In this paper, a rigorous optimal calibration and design methodology is presented, which eliminates the need of the traditional trial-and-error technique. The optimal calibration and design problems are cast as nonlinear optimization problems and solved using genetic algorithms (GA) optimization and particle swarm optimization (PSO). The EPA storm water management model (SWMM5) is employed to perform hydrologic and hydraulic analyses. The optimal calibration model determines the set of calibration parameters that best matches field observations of flow, depth or velocity to accurately mirror actual system performance. The optimal design model determines the set of design parameters that best meets desired system performance requirements at minimum cost. Design parameters can include any combination of pipe slope and size, storage, pumping, and new piping. System performance criteria include explicit constraints on the maximum allowable depth to diameter ratio, minimum and maximum pipe velocities, and maximum head loss for force mains. The proposed optimal calibration and design models are demonstrated by application to an example urban stormwater collection system. Enhancement of urban drainage system planning, design, operation and management is a principal benefit of the proposed methodology.

Comparison of Model Evaluation Methods to Develop a Comprehensive Watershed Simulation Model

Misgana K. Muleta — Mon, 21 Nov 2011 12:14:19 PST

Comprehensive environmental models such as the Soil and Water Assessment Tool (SWAT) are becoming an integral part of decision making processes for effective planning and management of natural resources. Before their use as decision making aid, however, models must be properly evaluated to improve their prediction accuracy and reduce the likelihood of making decisions that could lead to undesirable policy outcomes. Model evaluation refers to practices such as quality analysis of input data, sensitivity analysis, calibration and verification, and uncertainty analysis. Many methodologies have been developed for model evaluations over the years. One of the major limitations of the existing model evaluation methods, in particular model calibration methods, is their computational inefficiency, especially when used to calibrate comprehensive watershed simulation models. It may take weeks to months of CPU time, depending on the problem size, to successfully calibrate a comprehensive watershed simulation model on a standard PC. In this study, two sensitivity analysis methods and four calibration methods are used to evaluate SWAT to improve its streamflow prediction accuracy for the Morro Bay watershed located on the central coast of California. Parameter sensitivity analysis was performed using step-wise-regression analysis and the one-factor-at-a time screening method. Calibration was performed using PEST, Genetic Algorithms, the Shuffled Complex Evolution, and the Dynamically Dimensioned Search using observed data from multiple sites in the watershed. The model evaluation methods are compared in terms of their computational efficiency as well as effectiveness to determine “accurate” results. The developed SWAT model can be used to evaluate effectiveness of the Best Management Practices installed in the Morro Bay watershed, and to also prioritize sites where BMPs may be implemented in the future to further improve ecological integrity of the Morro Bay Estuary, which is one of the most important wetlands in California as it supports wide variety of habitats including numerous sensitive and endangered plant and animal species.

Improving Model Performance Using Dynamic Evaluation and Proper Objective Function

Misgana K. Muleta — Mon, 21 Nov 2011 12:14:13 PST

Models have become important decision making aids. Model evaluation (i.e., global sensitivity analysis, calibration and uncertainty analysis), is crucial to improve their prediction accuracy and reduce the likelihood of making decisions that could lead to undesirable policy outcomes. The conventional approach assumes that model parameters are insensitive to season irrespective of the temporal variability of input forcings such as rainfall. This assumption could significantly compromise model performance for low flow seasons and/or high flow seasons depending on the calibration method pursued. This study will demonstrate the advantage of dynamic (seasonal) model evaluation in improving performance compared to the traditional approach. In addition, the impact of the goodnessof- fit criteria (e.g., mean of sum of square of residuals, Nash-Sutcliffe efficiency criteria, volume based efficiency criteria, etc) used as an objective function during automatic calibration on model performance has been examined. Objective functions that would improve the accuracy of simulating high flows as well as low flows were identified. The added values of using multiobjective calibration, over the more widely used single objective calibration, has also been explored. The Little River Experimental Watershed, one of the U.S. Department of Agriculture’s experimental watersheds, has been used to illustrate the approaches tested in the study. Soil and Water Assessment Tool is the watershed simulation model used for the work. Results show that the season based model calibration approach significantly improved model performance, and calibration is sensitive to the efficiency measure used as object function. As such, multiple efficiency criteria should be used to report model performance as no single efficiency measure performed consistently well in describing goodness of model results. Another important finding is that parameter values that are significantly divergent from their “true” values may lead to model performance that may be considered near perfect even when judged using multiple efficiency measures underlining the challenge of parameter identifiability.

Decision Support for Watershed Management Using Evolutionary Algorithms

Misgana K. Muleta et al. — Mon, 21 Nov 2011 12:14:08 PST

An integrative computational methodology is developed for the management of nonpoint source pollution from watersheds. The associated decision support system is based on an interface between evolutionary algorithms (EAs) and a comprehensive watershed simulation model, and is capable of identifying optimal or near-optimal land use patterns to satisfy objectives. Specifically, a genetic algorithm (GA) is linked with the U.S. Department of Agriculture’s Soil and Water Assessment Tool (SWAT) for single objective evaluations, and a Strength Pareto Evolutionary Algorithm has been integrated with SWAT for multiobjective optimization. The model can be operated at a small spatial scale, such as a farm field, or on a larger watershed scale. A secondary model that also uses a GA is developed for calibration of the simulation model. Sensitivity analysis and parameterization are carried out in a preliminary step to identify model parameters that need to be calibrated. Application to a demonstration watershed located in Southern Illinois reveals the capability of the model in achieving its intended goals. However, the model is found to be computationally demanding as a direct consequence of repeated SWAT simulations during the search for favorable solutions. An artificial neural network (ANN) has been developed to mimic SWAT outputs and ultimately replace it during the search process. Replacement of SWAT by the ANN results in an 84% reduction in computational time required to identify final land use patterns. The ANN model is trained using a hybrid of evolutionary programming (EP) and the back propagation (BP) algorithms. The hybrid algorithm was found to be more effective and efficient than either EP or BP alone. Overall, this study demonstrates the powerful and multifaceted role that EAs and artificial intelligence techniques could play in solving the complex and realistic problems of environmental and water resources systems.

Analysis and Calibration of RDII and Design of Sewer Collection Systems

Misgana K. Muleta et al. — Mon, 21 Nov 2011 12:14:03 PST

Excessive wet weather flow resulting from rainfall-derived inflow and infiltration (RDII) is a major source of sanitary sewer overflows (SSOs). SSOs pose serious problem to the public and the environment by causing back up into basements and sewer overflows to streets and rivers. Control of sewer overflows is, therefore, vital to reducing risks to public health and protecting the environment from water pollution. Computer modeling of sewer collection systems plays an important role in determining sound and economical remedial solutions that reduce RDII, improve system integrity, reliability and performance, and avoid overflows. This paper presents a rigorous and efficient three-step optimization methodology for use in solving the sewer overflow problem. The first step analyzes measured sewer flow and rainfall data and decomposes the flow data into dryweather flow and wet-weather flow components. The second step computes the optimal RTK parameters of the tri-triangular unit hydrograph that is commonly used to model RDII into the sewer collection system. The optimal RTK parameters are calibrated with genetic algorithm so that the simulated RDII flows closely match the RDII time series generated by decomposing the measured flow data. In the final step, the calibrated model is then used with genetic algorithm to design cost-effective solutions for existing SSO problems. Design parameters can include any combinations of pipe size, storage, slope, and pumping. The proposed wet-weather flow decomposition, optimal calibration, and optimal design models are demonstrated using an example sewer collection system. The methodology seems a good alternative to other methods proposed in the literature and should prove useful for engineers and planners that are involved in mitigating complex SSO problems.

An Innovative Geocentric Decision Support Solution to Comprehensive Planning, Design, Operation, and Management of Urban Drainage Systems

Paul F. Boulos et al. — Mon, 21 Nov 2011 12:13:58 PST

Geographic Information System (GIS) is quickly becoming a critical component to develop and sustain asset management for today’s wastewater utilities as most of their data is geographically referenced. This technology offers sophisticated data management and spatial analysis capabilities that can greatly improve and facilitate urban drainage infrastructure modeling and analysis applications. This paper presents a comprehensive GIS-based decision support system that integrates several technologies for use in the effective management of urban stormwater collection systems. It explicitly integrates ESRI ArcGIS geospatial model with advanced hydrologic, hydraulic, and water quality simulation algorithms, nature-based global optimization techniques including genetic algorithms for design and calibration of stormwater management models, automated dry weather flow generation and allocation, and automated subcatchment delineation and parameter extraction tools to address every facet of urban drainage infrastructure management. The geocentric interface allows seamless communication among the various modules. The resulting decision support system effortlessly reads GIS datasets, extracts necessary modeling information, and automatically constructs, loads, designs, calibrates, analyzes and optimizes a representative urban drainage model considering hydrologic and hydraulic performance requirements. It also makes it easy to run, simulate and compare various modeling scenarios, identify system deficiencies, and determine cost-effective physical and operational improvements to achieve optimum performance and regulatory compliance. These combined capabilities provide favorable geospatial environment to assist wastewater utilities in planning, designing, and operating reliable systems and in optimizing their capital improvement programs.

Evolutionary algorithms for multiobjective evaluation of watershed management decisions

Misgana K. Muleta et al. — Mon, 21 Nov 2011 12:13:53 PST

The comprehensive and systematic management of watersheds is essential for reducing the adverse environmental impacts arising from anthropogenically caused erosion and subsequent sedimentation. This paper describes a computational methodology that is designed to serve as a watershed decision support system and is capable of controlling environmental impacts of non-point source pollution resulting from erosion. In the decision process, the methodology also accounts for other inseparable objectives such as economics and social dynamics of the watershed. This decision support tool was developed by integrating a comprehensive hydrologic model known as SWAT and state-of-the-art multiobjective optimization technique within the framework of a discrete-time optimal-control model. Strength Pareto Evolutionary Algorithm (SPEA), a multiobjective optimizer based on evolutionary algorithms, has been used to generate Pareto optimal sets. For demonstration purposes, the tool was applied to the Big Creek watershed located in Southern Illinois. Results indicate that the methodology is highly effective and has the potential to improve comprehensive watershed management.

Multiobjective Optimization for Optimal Design of Urban Drainage Systems

Misgana K. Muleta et al. — Mon, 21 Nov 2011 12:13:49 PST

Control of sewer overflows, the leading cause of water pollution in the nation’s water bodies, is vital to reducing risks to public health and protecting the environment. The most common solutions for mitigating sewer overflows include adding storage volume, increasing conduit capacity, expanding pumping capacity, and implementation of real time operational controls to more effectively utilize existing system storage. Obviously, comprehensive modeling and analysis of these sewer systems becomes necessary for developing sound cost-effective and reliable solutions for enhancing system integrity and performance to convey sewer flows without causing overflows. However, identification of the optimal remedial solution that effectively circumvents overflow problems with the least expenditure is a daunting task. The current practice involves a tedious trial-and-error evaluation procedure that seldom leads to the most effective or most economical solutions. Another emerging design approach utilizes single objective optimization that identifies the solution that best satisfies a predefined criterion. The performance criterion used with single objective optimization subjectively lumps the economics objective with metrics that measure effectiveness of the remedial solution from the perspective of avoiding overflows (e.g., minimizing the number of flooding events or reducing the flooding volume). Consequently, the design solution identified using single objective optimization depends on the weights subjectively placed on the two incommensurable and conflicting objectives, and may not represent the global optimal solution. A preferable approach is to seek tradeoff solutions commonly referred to as non-dominated solutions or Pareto-optimal solutions. The methodology proposed here links an extended version of the EPA SWMM 5 model, a comprehensive drainage network simulator, with NSGA-II, an evolutionary multiobjective optimization method with a proven history of identifying Pareto-optimal solutions for a wide range of engineering problems. The method should prove useful to any wastewater utility attempting to improve system integrity, reliability and performance and optimize its capital improvement program.

Modeling Erosion and Sedimentation Processes in the Chorro Creek Subwatershed to Evaluate and Develop Effective Watershed Management Approaches

Michael Randall et al. — Mon, 21 Nov 2011 12:13:45 PST

The Morro Bay Watershed, located in San Luis Obispo County, California, covers more than 48,000 acres of land and discharges into Morro Bay through the Morro Bay National Estuary (MBNE). The Chorro Creek Subwatershed consists of approximately 30,000 acres of the overall watershed. The MBNE provides an ecosystem that supports a variety of wildlife, from the common sea gull to the endangered sea otter. The estuary is also home to over 200 species of birds. The operational waterfront of the Morro Bay Harbor was and continues to be a strong supporter to the local economy of the City of Morro Bay. Numerous studies were conducted since the 1990s throughout the watershed to study the sedimentation of the estuary and bay and identified accelerated erosion and subsequent sedimentation as a major threat to sustainability of the bay. As a result, various Best Management Practices (BMPs) have been implemented in the watershed to reduce sediment loading and transport to the bay. Localized evaluations of various BMPs have been performed to investigate effectiveness of individual BMPs. This paper consolidates this information and develops a comprehensive spatially distributed watershed simulation model (1) for detailed understanding of the erosion and sedimentation processes in the watershed; (2) to evaluate a watershed scale effectiveness of the conservation practices that have been installed in the watershed; (3) to identify optimal BMP types and sites that may be used in the future to further reduce sedimentation of the bay at minimal cost; (4) to organize and document the various sources of data and studies that have been performed to date in the Chorro Creek subwatershed. Soil and Water Assessment Tool (SWAT) was used to develop the model and to evaluate the pre and post BMP implementation characteristics in the subwatershed. Combining the data and efforts of past BMP evaluations, land use, soil type, climate data, and streamflow data, statistical evaluations, and model sensitivity analysis will help build and calibrate a robust SWAT model that can be used to track BMP evaluation efforts, as well as other watershed management tasks. Through the evaluation of BMPs in the watershed, efforts can be made to implement the more successful BMPs in the watershed or in other similar watersheds. Sensitivity analysis was performed using a global sensitivity analysis method and streamflow and sediment yield was calibrated using the Shuffled Complex Evolution-University of Arizona.

A Multiobjective SDSS for Management of Urbanizing Watersheds: The Case of the Lower Kaskaskia Basin, Illinois

Kyle O. Allred et al. — Mon, 21 Nov 2011 12:13:41 PST

The conversion of natural and agriculturally dominated watersheds to industrial, commercial and residential developments leads to a cascade of adjustments in runoff quantity and stream quality at locations further downstream. The use of sophisticated hydrologic simulation models and Geographic Information Systems (GIS) has become the standard for evaluating these impacts of urban sprawl on water resources systems. Simulation and GIS models alone, however, are incapable of directly revealing optimal land development patterns that meet specified objectives. This paper describes the development of a multi-objective Spatial Decision Support System (SDSS) designed to overcome this limitation. The SDSS is created by integrating the U.S. Department of Agriculture’s Soil and Water Assessment Tool (SWAT) for comprehensive hydrologic simulation, a GIS for generating input and visualizing output, and a genetic algorithm (GA) for identifying weighted, optimal land use patterns. In addition to the GA, future research will involve the integration of a second search mechanism, the artificial life algorithm, to verify optimal results. The optimal landscape is that which minimizes sediment yield in subsequent streams, while simultaneously maximizing approximate anticipated profit from urban development. The SDSS could be a useful visualization tool for land use managers and watershed management institutions in planning new developments. The SDSS has been tested on the Lower Kaskaskia watershed, located in the Metro East area of southwestern Illinois. Evidenced by a historical survey of population growth and hydrologic and water quality variability, this basin is an example of a watershed that is undergoing extensive water resources changes as a result of urbanization. An investigation of watershed planning activities and stakeholder groups in the watershed has also been undertaken. Meetings with these individuals have allowed direct dissemination of the research to affected groups and have been useful for generating feedback on future work and model modifications.

Watershed Management Technique to Control Sediment Yield in Agriculturally Dominated Areas

John W. Nicklow et al. — Mon, 21 Nov 2011 12:13:36 PST

Non-point source pollution is recognized internationally as a critical environmental problem. In Illinois, soil erosion from agricultural lands is the major source of such pollution. The erosion process, which has been accelerated by human activity, tends to reduce crop productivity and leads to subsequent problems from deposition on farmlands and in water bodies. Comprehensive watershed management, however, can be used to protect these natural resources. In this study, a discrete time optimal control methodology and computational model are developed for determining land use and management alternatives that minimize sediment yield from agriculturally-dominated watersheds. The solution methodology is based on an interface between a genetic algorithm and the US. Department of Agriculture s Soil and Water Assessment Tool. Model analyses are performed on a farm field basis to allow capture of different, local stakeholder perspectives, and crop management alternatives are based on a three-year rotation pattern. The decision support tool is applied to the Big Creek watershed located in the Cache River basin of Southern Illinois. The application demonstrates that the methodology is a valuable tool in advancing comprehensive watershed management. The study represents part of an ongoing research effort to develop an even more comprehensive decision support tool that uses multicriteria evaluation to address social, economic, and hydrologic issues for integrative watershed management.

Evaluating Effectiveness of Best Management Practices to Control Accelerated Sedimentation of the Morro Bay Estuary

Misgana K. Muleta — Mon, 21 Nov 2011 12:13:32 PST

The Morro Bay estuary, located on the central Coast of California approximately half way between Los Angeles and San Francisco, is one of the most important wetlands on the west Coast as it supports wide variety of habitats including numerous sensitive and endangered plant and animal species. Various studies have identified accelerated erosion and subsequent sedimentation as a major threat for sustainability of the bay. Watershed disturbances caused by agricultural activities are believed to be one of the major causes of the accelerated erosion and sedimentation. More than 200 conservation practices have been installed in the watershed since the mid-1990 to reduce erosion and sedimentation. This paper will review the implemented BMPs and will evaluate effectiveness of the BMPs using observations and modeling exercise. Streamflow and sediment concentration, measured mainly during the rainy seasons, are available for multiple locations in the watershed. However, the observations are not sufficient in terms of spatial density and data length to evaluate effectiveness of the mitigation measures at various locations in the watershed. It would be daunting in terms of cost to develop an intensive network of monitoring sites that would be needed for reliable management of NPS pollutants. As a result, comprehensive watershed simulation models that integrate watershed and climate characteristics and can estimate pollutant quantity at various locations, and that can also identify source of the contaminants, is emerging as a key component of watershed management. In this regard, a comprehensive watershed simulation model for the Morro Bay watershed has been developed using Soil and Water Assessment Tool (SWAT) to simulate both streamflow and sediment concentration. The observed data was used to improve prediction accuracy of the SWAT model through parameter sensitivity analysis and calibration steps. Parameter sensitivity analysis was performed using step-wise-regression analysis and Morris’s one-at-a time (OAT) method. Calibration was performed using four different optimization methods: PEST, Genetic Algorithms, the Shuffled Complex Evolution Algorithm, and Dynamically Dimensioned Search. Relative performance of the sensitivity analysis methods and the calibration algorithms will be discussed in terms of effectiveness and computational efficiency. The developed model was used to evaluate effectiveness of the BMPs implemented in the Morro Bay watershed, and can also be used to prioritize sites where BMPs may be implemented in the future to further improve ecological integrity of the estuary.

Using Genetic Algorithms and SWAT to Minimize Sediment Yield From an Agriculturally Dominated Watershed

Misgana K. Muleta et al. — Mon, 21 Nov 2011 12:13:28 PST

Non-point source pollution is well recognized as one of the most critical environmental hazards of modern times. In Illinois, non-point source pollution is the major cause of water quality problems, and soil erosion from agricultural lands is the major source of such pollution. Accelerated by anthropogenic activities, soil erosion reduces crop productivity and leads to subsequent problems from deposition on farmlands and in water bodies. Watershed management, however, promotes protection and restoration of these natural resources while allowing for sustainable economic growth and development. In this study a discrete time optimal control methodology and computational model are developed for determining land use and management alternatives that minimize sediment yield from agriculturally dominated watersheds. The methodology is based on an interface between a genetic algorithm and a U.S. Department of Agriculture watershed model known as Soil and Water Assessment Tool (SWAT). The original structure of the SWAT model is preserved and modifications are embedded for computational efficiency. The analysis is based on a farm field level to capture the perspectives of different stakeholders. The model thus supports Illinois EPA’s plan of developing a program based on enabling and empowering local stakeholders to take charge of the fate of their watershed. Management alternatives available for all land uses modeled by SWAT are developed considering rotation patterns of three years. The decision support tool is applied to Big Creek sub-watershed in the Cache River watershed, located in Southern Illinois. Big Creek subwatershed has been sighted by the Illinois EPA for excessive sediment and nutrient loadings and has been targeted by the Illinois Pilot Watershed Program. This research is part of an ongoing effort to develop a comprehensive decision support tool that uses multi-criteria evaluation to address social, economic and hydrologic issues for integrative watershed management.

Joint Application of Artificial Neural Networks and Evolutionary Algorithms to Watershed Management

Misgana K. Muleta et al. — Mon, 21 Nov 2011 12:13:24 PST

Artificial neural networks (ANNs) have become common data driven tools for modeling complex, nonlinear problems in science and engineering. Many previous applications have relied on gradient-based search techniques, such as the back propagation (BP) algorithm, for ANN training. Such techniques, however, are highly susceptible to premature convergence to local optima and require a trial-and-error process for effective design of ANN architecture and connection weights. This paper investigates the use of evolutionary programming (EP), a robust search technique, and a hybrid EP–BP training algorithm for improved ANN design. Application results indicate that the EP–BP algorithm may limit the drawbacks of using local search algorithms alone and that the hybrid performs better than EP from the perspective of both training accuracy and efficiency. In addition, the resulting ANN is used to replace the hydrologic simulation component of a previously developed multiobjective decision support model for watershed management. Due to the efficiency of the trained ANN with respect to the traditional simulation model, the replacement reduced the overall computational time required to generate preferred watershed management policies by 75%. The reduction is likely to improve the practical utility of the management model from a typical user perspective. Moreover, the results reveal the potential role of properly trained ANNs in addressing computational demands of various problems without sacrificing the accuracy of solutions.

Sensitivity and uncertainty analysis coupled with automatic calibration for a distributed watershed model

Misgana K. Muleta et al. — Mon, 21 Nov 2011 12:13:20 PST

Distributed watershed models should pass through a careful calibration procedure before they are utilized as a decision making aid in the planning and management of water resources. Although manual approaches are still frequently used for calibration, they are tedious, time consuming, and require experienced personnel. This paper describes an automatic approach for calibrating daily streamflow and daily sediment concentration values estimated by the US Department of Agriculture’s distributed watershed simulation model, Soil and Water Assessment Tool (SWAT). The automatic calibration methodology applies a hierarchy of three techniques, namely screening, parameterization, and parameter sensitivity analysis, at the parameter identification stage of model calibration. The global parameter sensitivity analysis is conducted using a stepwise regression analysis on rank-transformed input–output data pairs. Latin hypercube sampling is used to generate input data from the assigned distributions and ranges, and parameter estimation is performed using genetic algorithm. The Generalized Likelihood Uncertainty Estimation methodology is subsequently implemented to investigate uncertainty of model estimates, accounting for errors due to model structure, input data and model parameters. To demonstrate their effectiveness, the parameter identification, parameter estimation, model verification, and uncertainty analysis techniques are applied to a watershed located in southern Illinois.