Hydroinformatics

The following sections are included:

• Preface

• Acknowledgments

• Contents

With the advent of increased computational power, the capabalities of computer-based modelling have expanded considerably. Numerical simulation methods based on well-established physical conservation principles are at the heart of present day research and consultancy activities. But emerging paradigms involving alternative computational methods are rapidly gaining recognition as well. In particular, in application areas related to environmental issues (notably water quality and ecology), techniques based on cellular automata, fuzzy logic, rule-based (expert) systems and artificial neural networks are proving quite valuable, also in practice. More and more data is becoming available from in situ monitoring systems as well as from remote sensing satellites, requiring advanced data mining techniques for interpretation and knowledge discovery. The integration of numerical modelling systems and data-driven modelling techniques is at the heart of any next generation modelling system presently under development. Recent advances of integrated modelling in the field of ecohydraulics are presented in this paper, taking harmful algal bloom prediction as a practical and relevant example.

The modelling of processes and parameters pertaining to the quality of Singapore marine waters, including the hydrodynamics, sediment transport and water quality, has been conducted sporadically in the past on a localized project basis. The studies were at times rather fragmented and not sufficiently integrated. At the Tropical Marine Science Institute (TMSI), a multidisciplinary and integrated approach has been adopted and this has led to the development of an enhanced prediction capability that ultimately would mature into a nowcasting and forecasting system.

The establishment of a regular and reliable field monitoring network is a first step towards the successful modelling and forecast of the marine environment, and is one of the cornerstones of the system. Field measurements help to characterize the baselines and trends in the changing environment. Such knowledge provides the necessary benchmarks for the deterministic models. Through data assimilation techniques, real time monitoring would provide the necessary boundary inputs, help to "steer" the prediction and "correct" imperfections in the data driving the models. Equally important is a dynamic model that is able to adequately describe the dominant processes and interactions in the marine system.

The Singapore seas, although a relatively small domain, is not a straightforward domain to model due mainly to the complex boundary forcing. For completeness, any modeling effort would require a regional model covering the South-China Sea, the Malacca Straits, and the Indonesian waters. This is to resolve the larger scale circulation and would mean a grid resolution larger than kilometers for practical purposes. On the other hand, typical description of the hydrodynamics and transport processes in the Singapore Strait would require a finer grid resolution. At the TMSI, a hybrid multidomain approach, comprising of two-way nesting techniques and parallel computing methodologies, has been implemented. This approach allows efficient forecasting tools to be developed using the research type models and low-cost computational power. The presentation will describe the key features of such a system, leading to the concept of a virtual Singapore Seas.

The use of evolutionary computation (EC) in the solution of optimization problems of hydroinformatics is not new. Many such problems having single or multiple objectives are now routinely solved using different EC methodologies. Through this invited paper for the proceedings of the Hydroinformatics 2004 conference, I am delighted to share some recent advances in the area of evolutionary computing with the researchers and practitioners to this field. Some critical issues, such as design of an efficient evolutionary algorithm (EA), an efficient constraint handling procedure, scalability issue of EAs, and multi-objective optimization are dealt in this paper. The discussion of the topics and subsequent engineering and numerical case studies presented in this paper should be useful to EA and non-EA hydroinformaticians alike.

No abstract received.

Nonlinear advection-dispersion-reactive equations govern numerous transport problems in the environment. Robust, accurate, and efficient algorithms to solve these equations hold the key to the success of applying numerical models to field problems. This paper presents the development and verification of a computational algorithm to approximate the nonlinear transport equations of reactive chemical transport, fluid dynamics, and multiphase flow. The algorithm was developed based on the Lagrangian-Eulerian decoupling method with an adaptive ZOOMing and Peak/valley Capture (LEZOOMPC) scheme. It consisted of both backward and forward node tracking, rough element determination, peak/valley capturing, and adaptive local grid refinement. A second-order tracking was implemented to accurately and efficiently track all fictitious particles. Shanks' method was introduced to deal with slowly converging cases. The accuracy and efficiency of the algorithm were verified with the Burger's equation for a variety of cases. The robustness of the algorithm to achieve convergent solutions was demonstrated by highly nonlinear reactive contaminant transport and multiphase flow problems.

Rectangular and curvilinear meshes are used for numerical wave simulations based on finite differences calculation schemes. In order to improve the approximation quality of rectangular meshes an adapted b-spline technique is developed to generate curvilinear meshes. The numerical results and the computational efforts for both kinds of meshes are discussed.

A coupled numerical model is presented for the simulation of wave transformation in porous structures. The model is developed combining porous flow model and two-phase flow model. The porous flow model is based on the spatially averaged Navier-Stokes equations. In two-phase flow model, air zone of finite thickness above the water surface is incorporated and a unique solution domain is established with the special treatment of interface boundary between water and air. Free surface advection is modeled by the volume of fluid method with newly developed fluid advection algorithm. The parameters representing the resistance to flow in porous medium due to the presence of solid skeleton are calibrated for different wave conditions. The model is then applied to simulate wave attack on porous structure for several wave and structural conditions. There exist optimum structure width and porosity that can maximize the hydraulic performances.

A quasi-2D model of aerated flow over stepped slopes has been developed, taking into account the non-uniform velocity profile and the air transport processes. The paper presents the mathematical approach, the numerical model and its validation.

The fluid loads on some hydraulic structures and the free surface profiles of the flow need to be determined for design purposes. This is a difficult task because the governing equations have non-linear boundary conditions. The goal of the present work is to develop a suitable and accurate numerical procedure based on moving least square method (MLS) that is used in element-free Galerkin method (EFG) for the computation of free surface profile, velocity and pressure distributions and flow rate in a two-dimensional gravity fluid flow through a sluice gate. The results of the calculations show good agreement with previous works done by finite element method.

In two dimensional hydraulic free surface problems, when a mesh is required to be representative of some topography, as the number of cells used to create the mesh increases, the discrete representation of the real problem improves, so the accuracy of the results is enhanced, but the computing time grows and this can be a cumbersome difficulty. On the other hand, the experience says that, when the topography is smooth, larger cells can be used to simulate a flooding event but, if the topography is highly irregular, the number of cells must be increased in order to allow for a correct flow representation. One solution is to generate a mesh with local refinement, in order to reduce the calculation effort and, at the same time, be able to achieve the accuracy in the flow simulation results that a globally fine mesh would provide. When cuadrilateral structured meshes are used, the refinement must be done in all cells contained in the rows and columns connected with the area whose cell-density we want to increase. Unstructured triangular meshes with local mesh refinement can lead to directional cell deformation depending on how they are created, and the grid refinement procedure is more complicated. The solution proposed here consists of using a structured triangular mesh of variable density following the variation of the bed slope. This produces a grid where the local refinement is introduced in the irregular topography zones, with a refinement depending on how irregular the topography is. Furthermore, the new cells are generated following a simple algorithm, without distortions and not affecting the size of the cells in the same row or column far from the area of interest. The performance and efficiency of a finite volume upwind method on different grids will be presented and compared for a river flow simulation problem over dry bed.

Smoothed Particle Hydrodynamics (SPH) is a Lagrangian numerical method which proved its ability to reproduce a large variety of flows. However, turbulence modeling is still poorly developed in the context of this method. Two traditional turbulence models are presented here for SPH, involving additional equations regarding turbulent kinetic energy and dissipation rate. Both are applied to test cases, showing fairly good results.

This paper concerns the risk of contaminant intrusion into water distribution systems (WDS) under extreme conditions (e.g. pipe break, low or negative pressure) which is one of the main reasons for water quality degradation. This is a particular concern in developing countries where intermittent supplies are norm. To investigate the risk of contaminant intrusion, a numerical model to simulate two-dimensional water flow and contaminant transport through homogeneous, partially saturated porous media is developed using the method of lines. The Richard equation and advection-diffusion equation governing water flow and contaminant transport respectively are solved numerically subject to prescribed initial and boundary conditions using local uniform grid refinement algorithms. The developed model is validated by comparing the models results with those available in the literature. The sensitivities of soil-water hydraulic properties are studied by varying the soil type.

Lattice Boltzmann Method (LBM) was developed into a flow-field simulating scheme in the middle of 1980s. With many outstanding advantages it will become a powerful competitor to the conventional CFD approaches. In this paper, according to the character of flow field with complex boundary, a new decomposition-couple algorithm of the LBM is put forward. This algorithm makes good use of excellence of the LBM to implement data exchange between decomposed regions, sufficiently utilizes the computational resource to increase efficiency. In the end, flow in fork pipelines with "T" and "+" geometries are simulated respectively, depicting the process in details and merits of this decomposition-couple algorithm. Compared with results obtained from the FDM and LBM with a set of integrative mesh, this new algorithm is proved to be correct and effective, as well as its perspective in application.

The environmental impact of the construction of the new terminal of the Port of Muuga, Estonia has been investigated. As the outer breakwater of the terminal is the first step of the generation of the main outer breakwater surrounding the whole port, special emphasis has been put on investigating change of current patterns in the Bay of Muuga and on sediment transport balance during the construction phases of reclaiming land from the sea.

Scarcity of streamflow gages represents one of the most common hydrologic worldwide situations. Therefore, several predictive hydrologic models have tried to incorporate parameter regionalization in order to estimate streamflow data in locations where gages have not been placed. In this light, this work proposes a new model for streamflow regionalisation by using both Artificial Neural Network (ANN) parameters and geomorphological basin parameters. Additionally, the employed ANN's are calibrated such that some flow data can be temporal forecasted. Therefore, this model allows users not only to estimate spatial data, but also to carry out predictions of streamflow data. The model is applied to a Colombian Region, and the results suggest that it may be considered as a new alternative for streamflow hydrologic modelers.

In this study the integrated hydrological model MIKE SHE is coupled with the river management model MIKE BASIN with use of the OpenMI system. The OpenMI is an Open Modeling Interface, which provides a framework, which simplifies linking of existing model codes and requires a minimal reengineering of these involved numerical models. The OpenMI system is under development and described in a number of reports and conference papers see for example Gijsbers [1].

The accurate numerical solution of the pure advection equation is of great interest for modeling the advection-dominated contaminant transport problems. This paper presents the comparative study of a series of numerical schemes, such as the first-order upwind scheme, central difference scheme, Lax-Wendroff scheme, double-step implicit and explicit scheme, QUICK scheme, QUICKEST scheme, the Holly-Preissmann semi-Lagrangian scheme and the total variation diminishing (TVD) algorithm, for solving the advection equation. The von Neumann method is employed to study the numerical errors of different schemes. The numerical tests are carried out and the comparison between the numerical results with the analytical solutions has been made. Among these schemes, the combined QUICKEST with the flux limiter algorithm is found to achieve the most accurate results without numerical oscillations near the sharp gradient of the variables.

Traditional "hard" drainage solutions are found to create significant problems of aggravated flood peaks and excessive sediment discharge in the downstream parts of urban catchments. To mitigate these, low impact drainage design techniques are sought. These involve multiple ponding areas with outflows controlled by conditions downstream, so successful low impact design requires accurate analysis of such elements. St Venant analysis as conventionally implemented in many commercial software packages is shown to overpredict such outflows by 10-20%, giving a "False Pump" effect compared with the results of an energy analysis. Reasons for this are examined, and the consequences for design are assessed in the light of laboratory results confirming the correctness of energy analysis for a test configuration.

The TGP's 3rd-Stage Open Channel Closure is characterized by large closure discharge, large closure drop, and high gap velocity. In order to carry out the construction of open channel closure successfully, 1D and 2D flow mathematical models are developed and closely integrated in this paper. The overflowing discharge of the open channel and water drops shared by the upstream and downstream cofferdams can be calculated through using 1D flow model, and the detailed hydraulic conditions in the open channel can be simulated through using 2D flow model. Then the water-drop distribution relationships between two closure cofferdams are analyzed under different gap widths by using these models. Calculated results indicate that the current advancing scheme is inapplicable with the closure discharge of 12200 m³/s, and it is very favorable for constructing with the closure discharge of 9010m³/s.

Hierarchical quadtree grids are a widely used and powerful approach to model complex 2D domains. In this paper, an object-oriented hydroinformatics simulation system toolkit is presented, allowing to set-up and process adaptive quadtree based finite volume models for shallow water equations (SWE) in river and coastal engineering problems. The developed toolkit provides a number of standardized components (packages) of different generalization level. Basic aim was to enhance adaptability and reusability of hydroinformatics simulation systems in combining expertise from modern software technology and hydrodynamic modeling.

Numerical models of overland flow have been applied to a number of practical problems of interest in Engineering. Basin irrigation is a surface irrigation system characterized by its potential to use water very efficiently. It consists on flooding from a point source, a squarish, relatively large field, leveled to zero average slope and fully surrounded by a dike to prevent runoff. The idea behind basin irrigation is very simple, although its real world applications have been limited by the difficulty to level the basin with the accuracy required to flood the field quickly and uniformly. A 2D first order upwind scheme is used to build a basin irrigation simulation model considering differences in bottom level. The implementation of upwind methods is not straightforward when source terms are relevant so that an upwind approach is adopted for the discretization of bed slope terms. Due to the small initial values (zero at initial conditions) for the variables and the high value of the Manning coefficient required to take into account the effects of the vegetation in the field, the friction terms become dominant and the numerical solution is affected. The infiltration parameters included in the source terms do not need any special treatment. The model is applied to reproduce two field experiments illustrating the relevance of land leveling on basin irrigation performance. Simulation results show a clear improvement over previous simulation efforts and are in a close agreement with experimental data.

A computer model is introduced for simulating mixed sub and super critical flow. The model numerically solves shallow water equations of continuity and motions in order to compute water depth (and free surface elevation) and depth average two-dimensional velocity patterns. The model can consider bed and wall geometric complexities and resistances. The governing equations are discretized using cell vertex finite volume method for triangular unstructured meshes. In order to stabilize the explicit solution procedure, artificial viscosity formulations is adopted for the unstructured meshes in such a way that preserves the accuracy of numerical results. The model can automatically distinguish flow regime at inflow and outflow boundaries, and therefore, imposes proper boundary conditions concerning flow regime. The accuracy of results is assed by simulating flow in a challenging case of mixed flow in a Parshall Flume and comparison of computed results with available measured data. Application of the developed model is presented to simulate flow from a dam reservoir on a chute spillway.

Two-dimensional model solving advection-dispersion equation was constructed with modified governing equation using finite element method. This model was testified by the comparison with the experimental results of the mixing in the laboratory S-curved channel and also RMA4 was run at the same condition and compared. These results show the improved simulation result for the meandering channel.

Shaft spillway, a new energy dissippator discharging whirling current with free surface, is applied in the reconstruction of diversion tunnel to spillway tunnel. This paper gives the numerical results of the fluid dynamics model, in which the RNS equations are solved with both the standard k – ε and Renormalization Group k – ε turbulence models, and VOF method are adopted to trace the free surface. The numerical result is in good agreement with the experimental data, and deepens the understanding of flow behavior. Energy dissipation ratio and discharge of the spillway can be calculated.

The application of catchment modelling systems is now a common approach for management of catchments. Fundamental to the application of a catchment modelling system is the calibration and validation of the many control parameters used to ensure that the simulated catchment response adequately reproduces the actual catchment response. Recorded information about the actual catchment response is needed if this necessary calibration and validation process is to be undertaken. However, the recorded information typically used for operation of a catchment modelling system rarely, if ever, is error free. These errors have the potential to influence the calibration of the control parameters for a catchment modelling system. Traditional calibration approaches focus only on achieving the parameter values that result in the best "curve-fitting" between simulated and recorded data. Introduced herein is an "Early Stopping Technique" which is aimed at avoiding "curve-fitting" through monitoring improvements in the objective function used for determining the set of optimal control parameters. Application of this approach to the calibration of SWMM (Storm Water Management Model) on the Centennial Park catchment in Sydney is outlined.

The paper is dealing with applications of detailed mathematical models focused on designation of flood zoning at particular reaches and provides an analysis of general impacts on flood protection measures in the whole river basin. The authors want to contribute to the unsolved problems of continuing urbanization and economic growth in the areas exposed to floods, associated with further increase of population and related danger of potential economic and cultural damages. It is necessary to designate these areas as flood plain areas or areas exposed to special floods and to regulate adequately their use. This task is impossible to fulfill without strong and well tested Hydroinformatics tools as DEM, hydrodynamic models and GIS tools.

For making strategic decisions concerning the implementation and extent of flood protection measures, it is necessary to know areas exposed to flood danger (define active and passive zones in flood plain areas) and characteristics of natural and special floods (flood hydrographs, duration of floods, flow velocity in flood plain areas, etc.).

This paper demonstrates the urgent necessity for a combined use of many powerful hydroinformatic methods and techniques to investigate and analyze complex processes in complex large-scale domains. This is explained for the quantification of methane-migration processes stemming from abandoned coal mines with the modeling system MUFTE-UG. The model set-up is carried out with a CAD system, a database and a mesh generator for elements of different dimensions. Uncertainties in the geology can be accounted for with geostatistical tools. Different model concepts, efficient discretization methods, fast solvers, parallel computers as well as advanced visualization are needed to understand the relevant processes and to prepare tools for future prognoses. Furthermore, optimization methods are helpful means for decision support, for example in optimizing methane-extraction measures.

In Dutch coastal waters, spring blooms of harmful algae, in particular Phaeocystis globosa occur almost every year. The blooms incur great adverse impacts on shellfish farming and recreation. In this paper, a one-dimensional vertical advection-diffusion model was developed to investigate the effects of sinking (buoyancy) and turbulence mixing on the vertical profile of algal concentration. The numerical model is then integrated with decision tree to predict possible P. globosa blooms by using the available sparse data. The results of the case study at Noordwijk transact showed that integrated model has good perspective because (1) the numerical module can provide understandings to the fundamental mechanism; (2) decision tree model has advantage to explore spare water quality and biological data.

Optimization of work of electric station with reservoirs is even more difficult task. The main criterion of optimization of energy mode of reservoirs is to maximize energy production in that period of year while there is a shortage of water. For instance, Kayrakum hydroelectric station, which is the only hydroelectric station of the most developed northern region of the republic this period, continues 6-7 months.

This paper addresses an important issue on basin hydrology, namely hydrological nonlinear system approach. A Time Variant Gain Model (TVGM), that is a simple relation of runoff generation and a formulation of nonlinear Volterra functional series in rainfall-runoff system, is extended as distributed hydrological modeling in basin scale, i.e., Distributed Time Varian Gain Model (DTVGM). A case study of Chaobaihe Basin in North China was presented, that shown this preliminary application with a good simulation results and a workable tool to study impact of land use and cover change on runoff.

A theoretically- and computationally-robust mathematical approach for decoding movement patterns of individual fish responding to biotic and physicochemical stimuli is described. The modeling approach, coupled Eulerian-Lagrangian agent individual-based modeling (CEL Agent IBM), is intuitive and based on well-established principles in computer science, fluid and water quality dynamics, computational fluid dynamics (CFD) modeling, neuroscience, and game and foraging theories. A CEL Agent IBM couples a 3-D Lagrangian particle-tracker supplemented with behavioral rules to a Eulerian CFD model. Mathematical structure of the behavioral rules is derived from an agent-based, event-driven foraging model. Stimuli are queried from information provided by a CFD or water quality model or a priori field data. Back-casting simulation analysis results in a mechanistic mathematical formulation of behavior amenable to forecast simulation. In this paper, we describe the theoretical concepts of a CEL Agent IBM used to decode observed 3-D movement and passage patterns of downstream migrating juvenile salmon (migrants) at Lower Granite Dam on the Snake River, Washington, USA. The prototype CEL Agent IBM (the Numerical Fish Surrogate) is presently used by the US Army Corps of Engineers to quantitatively forecast and assess the response of migrants to virtual designs of alternative bypass systems at federal hydropower dams. As this specific example illustrates, CEL Agent IBMs are applicable to many aquatic systems and provide the theoretical and computational facility for improving existing individual-based modeling and water resource decision-support.

The estimation of conveyance is fundamental to river maintenance and to one-dimensional (1-D) river modelling. Current software packages are generally based on methods derived from research completed more than 50 years ago. Subsequent advances in knowledge and understanding have had little impact on industry practice. In 2002, the Environment Agency in England & Wales commissioned a targeted programme of research to produce an improved conveyance estimation system, building on recent advances and industry best practice. The programme focuses particularly on the effects and management of riverine vegetation, the interaction between river channel and floodplain flows, and the behaviour of natural shaped channels. The paper describes the advances in roughness and conveyance estimation. The application of the new techniques is described as a simple application for river maintenance engineers and as an improvement to complex, 1-D river modelling.

In this study, a micro-model was used as the porous media to conduct Kr-S-P experiments. The micro-model consisting of randomly distributed pores and throats of five different sizes were made by carving two acrylic plates with laser tool. Throughout the experiment, the distribution of liquids in the micro-model was recorded and analyzed by a digital image recording system and image analysis software. The NAPLs used in the experiments were diesel fuel (LNAPL), and trichloroethane (DNAPL). The experimental results show that the pressure-saturation relationships of every two-fluid pairs are consistent with the P-S curves derived from any other fluid pairs using the scaling rule and the Laplace's equation. Furthermore, the Kr-S curves for every two-fluid pair are in agreement with those predicted by the Van Genuchten model.

The following sections are included:

• INTRODUCTION

• WOLF 2D NUMERICAL MODEL
  • Equations
  • Spatial and temporal discretisation
  • Other main features of WOLF 2D

• DAM BREAK SIMULATION
  • Malpasset dam break
  • Simulation data and initial conditions
  • Results

• CONCLUSIONS

• REFERENCES

In this study, a new numerical model has been developed for predicting flow properties in wetlands, where vegetated surface flow interacts with the saturated groundwater flow. To do this, a quasi-three dimensional numerical surface water model able of simulating flow through vegetation is integrated with the saturated groundwater model. The quasi three-dimensional model is based on the solution of 2D shallow water and part of Navier stokes equations. The solution is achieved using both finite volume and finite difference methods. The groundwater model is based on the solution of 2D saturated groundwater equations that are solved by the finite volume method. The integrated numerical model is applied to several artificial test cases and the model ability to deal with vegetated surface and saturated groundwater flow problems has been demonstrated.

This paper deals with the application of Godunov type numerical method of two dimensional shallow water equations for the analysis of flows in fishways with stones embedded in regular and staggered patterns. The shallow water equations are discretised on quadtree grids and solved using finite volume method. Comparing the flow properties thus obtained, regular stones patterned fishways are found to be practically more suitable than the staggered stones patterned fishways.

Towards the realization of appropriate modeling of vegetation roughness for river management purposes, a sensitivity analysis was carried out in order to identify the dominant parameters on predicted water levels. Based on 1D theoretical considerations and 1D numerical calculations, hydraulic properties of a simple prismatic channel were examined in relation to changing floodplain roughness, relative floodplain width and the trajectory length of roughened floodplain. It is shown that water level rise is most sensitive to changes in floodplain width if the original floodplain width is comparable in size to the width of the main channel. For changes in the trajectory length of the roughened floodplain, a rough absolute range was identified where sensitivity on resultant water levels is significant.

The hydraulic design of the Yuanshanzi diversion works in the Keelung river is considerably important for flood mitigation. This flood diversion works is one of the major tasks in the Keelung River Flood Mitigation Project. The main purpose of the diversion works is to discharge 1,310 m³/s diverted through a 2.48km-long tunnel into ocean under 200-year return period flood. A physical model of the Yuanshanzi diversion works was constructed to investigate the flow and sediment movement characteristics to ensure the proper design of diverting discharge. The major features contain a side-overflow weir, a ogee-shaped diversion weir and a river weir with two orifice and two sluice gates. Flow fields were measured at various locations in the physical model. A two-dimensional depth-averaged numerical model will be employed to simulate the flow fields during diverting water under the project design flood.

A 2-D hydrodynamic model of settling tank including density flow and solid settling was experimentally verified and applied to study the effect of feedwell configuration on the development of suspended solid (SS) concentration profile in secondary settling tank. Effect of feed well configurations having different depths and opening sizes were investigated. The simulation results suggest that the concentration profile of SS in the settled sludge bed depend significantly on the depth of the feed-well. With deeper feed-well, the solid-liquid interface (sludge blanket) was lower and the average solid concentration within the sludge bed was higher than that observed for the shallow feed-well. While the feed-well depth showed significant differences in the settled sludge profile, the opening size of feedwell resulted in only marginal differences. Transient simulations at peak loading conditions revealed the dynamics of settled sludge bed and sludge blanket, differences in which also suggest a relation to the depth of feed well. Contrary to these differences in SS profile, both under steady and dynamic conditions, the SS concentration in effluent only differed slightly in different configurations. Physical interpretation of these results and its potential implications on biological reactions in settling tank are discussed.

A new type of grid consisting of polyhedral cells derived from a two-dimensional Voronoi decomposition of the flow domain is presented. The grid allows for a more precise accounting of arbitrarily directed flows than it is the case with usual hexahedral grid types. At the same time, the computational effort for solving the governing equations is lower compared to triangle-based grids with the same number of variables in a staggered arrangement. Examples for application of the presented modelling approach to real-world cases are given at the end of the paper.

Based on the fundamental equations of fluid mechanics a one dimensional unsteady flow model for natural rivers with floodplains is developed. The friction flow, the cross sectional area and the non uniform velocity distribution coefficient are approximated by polynomials as a function of the water depth. The variational integrals of the finite element method are estimated by calculating the derivatives of the integrals analytically and using the Newton-Raphson Method. The model is applied for the tidal stream of the river Stoer in Northern Germany and for a flood at the Danube in Bavaria.

The application of catchment modelling systems is now a common approach for management of catchments. Fundamental to the application of a catchment modelling system is the calibration and validation of the many control parameters used to ensure that the simulated catchment response adequately reproduces the actual catchment response. The calibration process, in general, consists of the systematic variation of control parameter values until a set of values is obtained that results in the adequate reproduction of the recorded catchment response. While this systematic variation may be undertaken manually, there have been a number of automatic calibration techniques which are based on the minimisation of differences between the predicted and recorded catchment response. Implicit in many of these techniques is the assumption that the residuals (ie variation between the predicted and the recorded catchment response) are independent, homoscedastic and normally distributed. Presented herein are the results of an investigation into these assumptions using the Powells Creek catchment in Sydney, Australia as a test catchment. It was found that these assumptions were not achievable on this typical catchment.

Sustainability has become a central issue in river basin management. However, sustainability is at best a compromise. The use of detailed mathematical modelling can allow reliable and robust management decisions to be made. This paper describes two differing case studies that illustrate the application of modelling to develop sustainable basin plans, and also the tensions that exist. The Segara Anakan project, in southern Java, Indonesia, aims to preserve the last mangrove lagoon in Java. The Basin Management Plan for the River Bicol in southern Luzon, Philippines, looks to improve flood defence, irrigation and fisheries in a relatively undeveloped part of the nation. In both cases the modelling carried out guided the choice of viable management approaches, but also raised larger questions as to their long-term sustainability.

In the present study, the large eddy simulation approach incorporated with Smagorinsky's subgrid-scale turbulence model is adopted to simulate the 3D flow fields in a curved channel. The good agreements between the simulated and measured results demonstrate the applicability of the current 3D flow model. The influence of turbulence coefficient C_s on the simulated flow fields is investigated. It is found that the relatively large value of C_s is needed to simulate the correct secondary flow field in such a curved channel. Furthermore, from the simulated results it is shown that the bed shear stress near the outer bank of the channel is higher than that near the inner bank. The ratio of the bed shear stress around the convex bank to that at the upstream is about 1.5.

Haimchar is located along the left bank of the Lower Meghna River, one of the Largest rivers in the world, which conveys the combined discharge and sediment load of the three major rivers of Bangladesh (Brahmaputra, Ganges and Upper Meghna) into the Bay of Bengal. A mathematical model study was carried out by Institute of Water Modelling (IWM) to assess the baseline hydraulics and morphological condition and autonomous development and also to aid planning and designing of sustainable riverbank protection work at Haimchar.

A 2D morphological model was developed covering the Padma, Upper Meghna and Lower Meghna River system using MIKE 21C, a curvilinear grid modelling system of DHI Water & Environment, Denmark. Five alternative options were tested in the model. Finally one option comprising three spurs found suitable for that erosion prone area.

In this paper the numerical method for a 3-D hydrodynamic model has been outlined. The numerical scheme is mainly based on 2-DV numerical techniques and is implicit, non-iterative and unconditionally stable. A coupled 2DH-3D system has been developed to provide the possibility of positioning the 3-D area in within part of a much larger 2-DH domain and then solving the hydrodynamic part of these areas simultaneously. Therefore, nc external boundary data are needed to be included for the 3-D model area.

Using a three-dimensional finite volume code with standard k-epsilon turbulence model the hydraulic resistance of willows (Salix Alba and Salix Fragilis) are modelled. The drag force approach was used by incorporating an additional term into the Navier-Stokes equations. This method is applied to model a flood event on the Wienfluss test reach in Vienna. The test reach comprised an asymmetric compound channel with vegetated floodplain of 114 metre length. The willow development has been monitored annually and this information was used in order to define the density of the willow canopy. Two vegetation geometry models are tested, a uniform density model where the density is constant with plant height and a non-uniform model where density is a function of height. The prediction in cross-sectional distribution of longitudinal velocities shows reasonable agreement for both models.

This paper describes an integrated computerized system facilitating the computational forecast of tidal hydrodynamics in Singapore Strait. The governing core of the system is the Model INTegrator (MINT), used for manipulation of 4-D (time & space) data, launching the hydrodynamics and relevant water quality modules, and on-line visualization. MINT is the universal-platform and model-independent environment for data pre- and post-processing and model management. The developed computational and visualization technologies are being extended to internet-based nowcast and forecast services. Communications of MINT, models and model parameters utilize XML language, for which a simple Fortran parser is developed; while storage of original and computed 4-D data employs the worldwide used netCDF standard. Usage of Java technologies and open source standards, like XML and netCDF, ensures easy exchange of data with other models, and makes MINT a platform independent. Standard looking intuitive GUI allows an easy control of complex computational tasks.

To understand retention effects of natural rivers with riparian forest case studies have been carried out on two german rivers. A two dimensional flow model was used that implements physical based formulas for the flow resistance of the bottom and of vegetation. Unsteady simulations for different flood situations lead to a better understanding of the influence of riparian forest on flow velocities on the flood plains, water stages and retention effects.

In order to model bottom shear-stresses accurately many empirical and physical formulations have been derived during the past centuries. Shear-stresses can be incorporated into a numerical model as Neumann boundary conditions and are the driving parameters for the computation of mass erosion processes. Thus, it is obvious that numerical results are highly influenced by the chosen shear-stress formulation. Consequently, two different groups of wall formulations are discussed in this paper: Empirical laws based on Manning's friction coefficient and the rivers depth averaged velocity as implemented in most two-dimensional models and wall functions in connection with bottom shear-stress velocities as most three-dimensional codes take advantage of. On a first thought it appears, that three-dimensional methods should be more accurately. However, a comparison for the Heimbach reservoir in the middle west of Germany was done using the Finite Element code PASTIS-3D with its Very Large-Eddy turbulence closure on the one hand and the depth averaged Finite Element model RISMO-2D with a k-ε turbulence model implemented on the other hand. The example demonstrates, that the resulted shear-stresses are not affected by depth-integrated contemplation when a properly shear-stress formulation was chosen. Thus, for simulations of erosion processes, a depth-integrated view is sufficient.

Contamination of water supplies intended for human consumption by pathogenic microorganisms is a concern for water managers in developed and developing countries alike. Typically, pathogens are associated with disturbed landscapes such as those used for human settlements or agricultural practices, and they progress from the catchment to the river or stream during periods of significant rainfall. Ultimately, they reach drinking water reservoirs and are potentially distributed to consumers. There is therefore a need for tools that can be applied to assess pathogen fate and transport as they move through a reservoir. This paper documents the development of a suite of freely available tools that range in complexity from a simple web-based intrusion model (INFLOW), to a one-dimensional hydrodynamic and pathogen model (DYRESM-CAEDYM), to a full three-dimensional model of hydrodynamics and pathogen fate and transport (ELCOM-CAEDYM). Results from the models are presented and assessed against data collected during a comprehensive field campaign in Australia that tracked pathogen concentrations throughout two reservoirs subjected to inflow forcing from rivers with high pathogen loads. All three models proved themselves as useful tools for investigating pathogen dynamics and are able to estimate dilution rates and timescales for risk reduction through inactivation and settling. The information provided by the models can be used to recommend a simple monitoring program and adaptive risk management strategies. The benefits and limitations of each of the models are also discussed.

For a semi-enclosed shallow water where reclaiming and dredging works are scheduled, the flow and seawater exchange rates have been predicted using a two-dimensional numerical model and a Lagrangian method. In particular, in order to evaluate wind effects, the mean wind speed and direction have been considered in the calculation of the flow with values of being averaged over thirty years. The flow after reclaiming and dredging showed a similar pattern as a whole, compared with the ones before development. However, there were variations of 20 to 100% in their velocities in the vicinity of development area. When a southerly summer wind, a seawater exchange rate at present appeared to be 71.6% while 82.9% after reclaiming and dredging, in terms of a particle tracking method. On the contrary, when a northerly winter wind, a seawater exchange rate at present appeared to be 97.1% while 93.2% after reclaiming and dredging. As a result, the seawater exchange rates turned out to be variable in accordance with a wind forcing since a construction area is located, leaning towards the north of the bay. However, the seawater exchange rate turned out to increase by 15% more when a dredging is simultaneously carried out with a reclaiming, compared with when only a reclaiming work is done. This suggests that a dredging can be an effective way to mitigate the variation of flow environment. In particular, it is meaningful because a seabed near the development area is badly contaminated by not only weak flow but also partially treated sewage.

A pressurized underground pipe system is investigated in the design of Beijing section project in middle route of North-South Water Transfer. An optimal model is used to find out the best plan of water conveyance. One of the prominences in the final plan is that in 80-kilometer route is built only one pumping station, located at the inlet of the project. The operations will be that, if the volumetric discharges are less than 20m³/s, water is conveyed by gravity; otherwise, pumps are put into use. Moreover, air valves are applied to preventing the phenomenon of liquid-column separations in power failures and it is suggested to regulate both the pump speed and check valve opening reasonably to prevent dangerous high pressures in the startup of the pumps.

Agriculture is listed as a major contributor of NPS pollution, which causes degradation of surface and ground water resources through soil erosion, chemical runoff, and leaching. Quantification and control of NPS pollutants is a real challenge. However, this can be achieved if the processes of transport (on surface and downward) of pollutants can be properly understood and modeled. Hardly any of the existing models accurately simulates both the surface transport and leaching of NPS pollutants. The processes of transport of agrochemicals being related to surface and subsurface hydrology, watershed should be taken as the unit for this type of study. Keeping these facts in mind, the current study was taken up with the major goal of developing a Decision Support System (DSS) for eco-friendly management of water resources and agrochemicals. The DSS takes into account both surface transport and leaching of pollutants. It uses modified curve number method, suitable for the region, to simulate the runoff volume. The modified Universal Soil Loss Equation is used to simulate the sediment yield. Modified Richards's equation and diffusion dispersion equations are used to simulate the downward movement of water and solutes through the soil respectively. Multiple-objective linear programming technique was used for optimization. The developed DSS was calibrated with the observed data from a 951 ha monitored watershed and controlled field experiments conducted under a sponsored research project undertaken by the authors. The DSS is to be tested in other larger watersheds so that it could be generalized and used as an effective tool for decision making related to eco-friendly management of agrochemicals, without sacrificing the agro-production.

The use of numerical models to assess the performance of the water disinfection process has become more frequent since the past decades, although the number of works in this subject is still small with respect to the demand. The ideal flow regime in contact tanks is plug flow and baffled tank is the most recommended tank geometry for the achievement of this type of flow with optimisation of space for construction. The resulting flow pattern in such tanks can approach a 2D flow if the tank inlet is well designed. In this case, the use of two dimensional numerical models is appropriate for simulations of flow patterns and solute transport along the tanks. However, some problems related to the monitoring of velocity in this type of units (full or model scales), such as low and reversing velocities, have made the calibration and validation of the numerical models a difficult task. With the objective of overcoming this problem, a new proposal of calibration and validation of numerical models applied to contact tanks is introduced and presented in this paper, which is based on the fitting of simulated and monitored flow through curves (FTC) along the whole extension of the tank. The following hypothesis is considered: "if the matching between all monitored FTC and their corresponding counterpart is satisfactory then not only the numerical module of solute transport of a conservative constituent is calibrated but also the depth averaged velocity numerical module". The aforementioned proposal is applied to a case study where a 2D numerical model (DIVAST; Roger Falconer; Cardiff University; UK) and a set of FTC obtained by the authors' research group are used. The results of this study indicate that the proposed methodology is very promising.

Computational Fluid Dynamics (CFD) is becoming more and more accessible to non-specialists thanks to the development of sophisticated Graphical User Interfaces (GUIs) and also because of the ever increasing computer power found on common PCs. However, as access to the models and data is made easier, the underlying complexity of CFD and the breadth of knowledge required from potential users should not be ignored; yet it should not detract them from making a good use of this resource either. To answer the latter the authors have developed tools around commercial CFD codes that allow non-experts to use CFD. Several scenarios arise in practice where it is perfectly feasible to customize a general code to make it more accessible. For example should a model be used in a repeated manner with different inputs it is then possible to create a generic version of the model, and provide a data-entry system that format instructions given by the user into proper CFD keywords recognized by the CAD, the mesher and/or the solver. Should some of this information be available from a measuring device – say a wind or water gauge – recording digital data files, the latter can even be used directly and the model run automatically. The present paper shows some examples from the water industry where PERL has been used to manage files and run the commercial code CFX in batch mode. Some of these examples are fully automated; others, more flexible, and simpler to build, require some input from the user. The conclusion of this work is that CFD is now accessible to non-specialists, or to persons who are not regular users. The facilitator and provider of these tools can be a specialist employee or a consultant external to the firm.

The expert system presented in this paper includes modules for the databank management and the combined simulation of surface flow (two dimensional) and groundwater flow (three dimensional) in floodplains.

This paper presents the development and application of a three-dimensional finite element model to simulate hydrodynamic and transport phenomena in bays, estuaries, and coastal waters. The hydrodynamic module of the numerical model solves three-dimensional Navier-Stokes equations with or without the hydrostatic assumptions. It also solves the energy and salinity transport equations, respectively, for the temperature and salinity distributions, respectively. The transport module solves a system of M transformed equations governing the transport, fate, and reaction of M biogeochemical constituents and sediment transport equations. The water quality module employs a general paradigm to transform the set of M biochemical constituents-transport equations into two subsets: kinetic-variable transport and mass action equations. The interaction between transport and reaction are dealt with three approaches: fully-implicit, predictor-corrector, and operator-splitting methods. Three examples are employed to illustrate the application of the model: one is the three-dimensional hydrodynamic simulation in Loxahatchee Estuaries and the other is the simulation of Eutrophication in surface water.

Before it reaches its final destination the water must travel through a distribution system and, thus, the water quality could vary in space and in time across the network, with deterioration of its chemical, physical and microbiological contents.

The water quality analysis and control in a water distribution system can be carried out through mathematical models, that simulate hydrodynamic characteristics of the flows in the pipes and transport and transformation of substances in the water. The model calibration requires the availability of an adequate number of measures in some control sections of the network. In this work in order to analyze the water quality of the Oreto-Stazione sub-network of Palermo city (Italy), the EPANET code is applied. The subnetwork is provided with a remote control system that allows to record both network's hydraulic variables and water quality parameters. The collected field data have been used to calibrate the chlorine decay law of the numerical code.

A concept to automate mesh generation for hydrodynamic finite element simulations is presented within a general framework for quality assessment, facing challenges for floodplain modeling based on high-resolution LiDAR measurements.

The bed and/or suspended sediment or pollutant movements in open channel flows can be described by means of Random Process Theory, through Eulerian and Lagrangean Descriptions. The Eulerian Description allows the analysis of the passage times of a cloud or plume of contaminant particles or sediments, through the cross flow sections. With the Lagrangean Description, the particles longitudinal positions are analyzed as a function of time. A software named PAICON (Processos Aleatórios com Injeções Instantânea e CONtínua), which means Stochastic Processes with Instantaneous and Continuous Injections, was developed in DELPHI language, possessing flexible and friend user data acquisition interface, graphical and file outputs, for the study of these movements. Analytical expressions of the models are presented and the main steps, tools and performances of the software described. The data outputs of the models are also illustrated, and compared to results obtained in laboratory channel and natural flows. The authors used data obtained in experiments carry out: in Brazil in a stretch of 39.6 km long of the Paraiba do Sul River with dyes; in the Arrudas River (6.5 km) with water labeled with radioisotope (Br⁸²); and in France in the Loire River (22.8 km) with simultaneous labels of water with dye, and fine sediments with Au¹⁹⁸. Bed-load experiments data were obtained in laboratory channel and in the Horácio Creek, Brazil, with sediment labeled with Au¹⁹⁸ and simulated with glass contained Ir¹⁹². The models considered in this paper are Homogeneous One-dimensional Poissonian Models: HDPM.

In Dutch coastal waters, spring algae blooms, in particular Phaeocystis globosa occur almost every year. The blooms incur great adverse impacts on shellfish farming and recreation. In this paper, an integrated numerical and fuzzy cellular automata model was developed to predict possible blooms basing on irradiance, nutrients and neighbourhood conditions. The numerical module used Delft3D-WAQ to compute abiotic conditions, and fuzzy cellular automata approach is applied to predict algal biomass. The simulation results of year 1995 are in good agreement with observations, and the modelled spatial patterns are close to the satellite images. Through this study, it is seen that integrated modelling is a promising approach because (1) the hydrodynamic processes and nutrients concentrations can be simulated in detail by numerical module; (2) the irregular and sparse water quality and biological data, and the empirical knowledge from experts can be explored by fuzzy logic; (3) the spatial heterogeneity, local interactions and the emerge of patchiness are captured through cellular automata.

Calibration is an issue that has been dealt with through different data assimilation techniques. An Extended Kalman filter (EKF), a well knowndata assimilation technique, has already beenimplemented in Sobek River that models the hydrodynamics of river and estuary systems. However, since the Ensemble Kalman filter (EnKF) is a more generic data assimilation technique and more suitable for highly non linear models, it is an attractive technique to be implemented as well. In this paper, a comparison of the EKF and the EnKF in Sobek River is carried out for the test case Maxau-IJssel (FEWS Rijn). The comparison is based on a set of criteria such as accuracy of the estimated state and its estimated uncertainity, sufficient number of ensembles and computational time. The comparison shows that the results of the EnKF for 100 and 30 members are varying closely around those of the EKF. Even the results for 10 members are still reasonable, considering the small number of the ensemble used. For fewer members than 10, results are unacceptable since the EnKF is underestimating the uncertainties in its estimate. The prediction capability of the EnKF for flood forecasting is also discussed and results show that both approaches are comparable. The computational time being a very important practical issue in real time application is assessed. For 10 members and without any improvement in the code to decrease the computational time, the time is 0.8 of that of the EKF. It is feasible to use the EnKF for real time applications, however, for applications where high accuracy is essential, improvement of the code or parallel processing are then required.

Relationship between critical conditions of rill erosion and the slope gradient/size of plots were investigated using numerical simulation and laboratory experiments, which featured large variations in slope gradients (10°- 65°). Factors considered include plot length, slope gradient, and net rainfall intensity. Shear velocity was used as a key parameter for establishing the critical conditions, and two cases were studied: 1) the critical shear velocity did not change with slope gradient, and 2) the critical shear velocity decreased with the increase in slope gradient. Results indicate that for any given plot size, net rainfall intensity and original topography, there exists a slope gradient (<15° in all cases) for which shear velocity in a rill will achieve the critical value. For each slope gradient, the plot length needed to achieve the critical shear stress was found.

Simulation models are increasingly becoming essential tools in the today's world of water management. They play an important role to fulfill the tasks of planning, policy preparations, operational management, and research. Nowadays the models are developed so that they provide water managers with an integrated system in whole. Integrated system where not only the different models can be coupled together, such as water flow model coupled with rainfall runoff model and water quality model, but the whole set of tools to input, validate and calibrate the data and presenting and analyzing the results. In other words, an integrated system provides the user with not only a simple modeling system but also a complete problem solving system. Such a complex system should then support a clear and step–by-step method to carry out a modeling project to avoid the risk of inexpert usage. Here, the concept of Good Modeling Practice (GMP) comes handy, not only for the user of the system but also for the developers of the system.

This paper presents Sobek-Rural and Sobek-Urban models, both developed by WL|DELFT HYDRAULICS. Sobek is an integrated numerical hydraulic modeling package to simulate hydrodynamics of one-dimensional (1D) river/channel/sewerage network and two-dimensional (2D) overland/street flow coupled with rainfall-runoff model and water quality model. Sobek has its exclusive Geographical Information System (GIS), Case Management Tool (CMT), Input data check, model logging, and Case Analysis Tool (CAT) which fully supports and streamlines its users into a practice of good modeling. The paper not only describes in details how to model a system in a best possible way using the various tools available under Sobek, but also how Sobek focus its user's attention and make them follows the good modeling practice.

The objective of this study is to investigate an integrated short-term precipitation-streamflow method for operational river flow forecast. Proposed system is based on the RDAPS model for precipitation and atmospheric variable simulations and the SFM model for streamflow simulations. The selected study area is the 2,703-km² Soyang River basin with outlet at Soyang dam site. The rainfall-runoff event from 18 to 24 July 2003 is selected for the performance test of predicted precipitations and streamflows. It can be seen that the simulated basin-scale precipitation from the RDAPS can be useable as an input for SFM hydrologic model. Short-term hydrometeorological simulations using the RDAPS and SFM model were well captured the important hydrometeorological characteristics in this study area. It is concluded that atmospheric precipitation forecast would be useful for streamflow forecast.

The paper presents the approach and methodology for water quality mo deling for the Nam Pasak II Channel; one of the recently rectified channels of Vientiane, the capital of Lao PDR. It also evaluates options to improve the deteriorating water quality observed in the channel. Two of the priority pollutants, total-P and NH₄-N, are modeled. The study is carried out in three steps, dry weather flow simulation, wet weather flow simulation and nutrient modeling using MOUSE. The dry weather flow simulations are carried out to calibrate the model for hydraulic roughness coefficient, dispersion coefficient and travel time. The wet weather flow simulations analyze the effect on flooding of two channel states, un-vegetated and vegetated conditions. The nutrient modeling evaluates the nutrient removal efficiency by the vegetation. The model results are compared with the observed data and recommendations are made with respect to the predicted effects of the water quality improvement schemes studied. The modeling approach presented herein can be applied for the performance analyses of urban channels in the developing part of the World – where data are often limited.

Despite their potential in so-called 'Third World' societies, mass-customised advice-serving systems have not so far been accepted in practice due to institutional constraints. Although facing similar constraints in 'First World' societies, there is in this case the possibility that public opinion will force applications of such systems upon a recalcitrant industry. Some potentially important applications in 'First World' societies are accordingly described here. In order to economise on space, all illustrations have been suppressed here and reference must then be made to the cited publications.

The Flyland research programme was developed in 1999 by the Government of the Netherlands to study the feasibility and impact of a new airport on an artificial island in the Dutch coastal zone. Within the research theme Marine Ecology and Morphology, WL | Delft Hydraulics performed extensive modelling of the hydrodynamics, morphology and ecology of the southern North Sea in order to determine the extent of spreading of sediment fines from sand mining operations and to determine the subsequent effect of increased suspended particulate matter (SPM) concentrations on the primary production. The results show that under worst-case conditions, significant impacts can be expected on the ecosystem. It is then demonstrated that the environmental impacts of sand mining can be reduced to an acceptable level by the proper selection of the location relative to the coast and by using a variety of (technical) mitigating measures.

The possibility is being considered for runoff modeling of water resources assessment and forecast using the mathematical model of river runoff formation, which was developed in the Central Asian Research Hydrometeorological Institute (SANIGMI). There are three main blocks in this model: calculation of water equivalent of snow and incomes of rain and melted snow water, calculation of runoff from glaciers, transformation of all incomes into runoff hydrograph at the outlet. To use data of mathematical model in practice an automated information system of hydrological forecasting (AISGP) was designed. The general scheme of data assimilation aimed at improvement of hydrological forecast quality was designed in AISGP. The results are adduced of improved technology of forecasting for particular rivers in the Syrdarya river basin. The results obtained show an opportunity to apply successfully assimilation data in AISGP for water resources assessment and forecast.

This study describes and evaluates a set of hydroinformatic tools with respect to the needs of the implementation of the EU Water Framework Directive (WFD). The tools that were applied in the study were; Sobek to handle river processes, MyLake for the lake processes, and ENSIS as a database and presentation system. They were all applied on the Vansjø-Hobøl River Basin, located in the south-eastern part of Norway, where eutrophication was the main environmental problem. The hydroinformatic tools (Sobek–MyLake-ENSIS) seemed to be useful for water management according to important needs given by the WFD. However, a need to extend the system with one or more models to simulate emission/runoff model of nutrients was identified. The given set of tools was applied to analyze the earlier proposed measures for pollution abatement, and concluded that the present definition of 'good ecological status' will be reached only 4-5 times in the simulated 30 years period.

In numerical modelling of coastal ocean processes, two largely separate modelling communities exist, originating from original decisions to implement either finite element or finite difference concepts. The development of improved schemes and new functionalities for both approaches, notably in the last decade and a half, has now led to very comparable functionality for practical applications. The paper argues the historic key role of the grid approach in this. It elaborates the considerations of grid selection in finite difference modelling, depending on problem scales. It presents the novel idea of shallow water flow finite difference modelling on a spherical curvilinear grid. This combines the flexibility of a curvilinear orthogonal boundary-fitted grid with exact mapping on the sphere, and permitting for very large model areas. The new approach is evaluated for a simple nearly β-plane test case with different grids, while its advantages are illustrated on the real-life application of typhoon-induced surge modelling and forecasting in the South China Sea.

Paper deals with estimation of groundwater parameters specifically Storage Co-efficient (S) and Transmissibility Co-efficient (T) of an aquifer. The values of these co-efficients are obtained by various methods namely Thies' method; Jacob's method and Chow's method and by using developed computer program based on numerical solution method. Pumping tests are carried out to study the groundwater potential in Amravati region of India. The spacing of wells in this region is recommended.

The usefulness of ECMWF climate model simulation information is examined for Korean water resources management. The objective of this study is to check the applicability of ECMWF information for the Korea peninsular. The methods are based on probabilistic measures of the effectiveness of GCM simulations of an indicator variable for discriminating high versus low regional observations of a target variable. The formulation uses the significance probability of the Kolmogorov-Smirnov test for detecting differences between two variables. AMIP-II (Atmospheric Model Intercomparison Project-II) type GCM simulation done by ECMWF (European Centre for Medium-Range Weather Forecasts) for the period 1/1979-12/1996 was used for indicator variable and observed mean areal precipitation (MAP) and discharge values on 7 major river basins in Korea were used for target variable. Monte Carlo simulations were used to establish the significance of the estimator values. Analyses were done in various ways to detect the sensitivity of discrimination condition, various window sizes, conditions of constraint, etc. The results show that GCM simulations done by ECMWF are skillful in discriminating the high from the low observed values. This means that seasonal and monthly ECMWF simulation data can be useful information for Korean water resources management. In addition, some trends were found in accordance with the change of discrimination condition, window setting, and conditions of constraint.

Side weirs are flow diversion devices that are widely used in irrigation, land drainage and urban sewage systems. The present study focuses on the experimentally investigation of the effect of an inclined crest on the discharge coefficient of a side weir under subcritical flow condition in prismatic and non-prismatic rectangular channels. In this study 750 laboratory tests were conducted and a discharge coefficient related to the flow through the elementary strip along the side weir length was obtained. Experimental results confirm that the proposed model well explains the behavior of distributed flow over side weir.

Urban flooding management is a very sensitive subject for municipal authorities and the need for analysis and models is increasing every year. However, the methodologies are facing the data integration problem. Since several years, the technology used for the production of High Resolution Digital Elevation Models (HRDEM) has known a very important evolution especially with availability of airborne the optoelectronic scanners, including the High-resolution Stereo Camera (HRSC) from DLR (Deutsches Zentrum für Luft und Raumfahrt) and the ADS40 from Leica Geosystems. Some tools are capable of processing such imagery to yield 3D digital models and true 25 cm orthoimages on a large scale. These data and new products as the HRDEM offer some new possibilities for the application of modeling activities to the urban processes. The possibility to obtain easily an accurate HRDEM – 25 cm to 1 m grid size – and integrate it directly in a 2D hydrodynamic modeling software is a key issue. This paper present the interests and the possibilities offered with this new technology combined with the classical modeling tools as Mike 21 applied for urban flooding purposes.

Comprehensive and regularly updated computerized hydrological databases are the keys to efficient water management. Hydrological Information Systems in many countries lack reliability, accessibility and timeliness. A major initiative to revitalize and improve the existing systems in nine peninsular states of India, covering a geographic area of about 1.7 million sq. km., has been taken by the Central and State governments through a World Bank supported Hydrology Project. This project, spanning eight years (1996 – 2003), sought to upgrade, strengthen and standardize all aspects of HIS viz. data acquisition, processing, management and dissemination mechanisms. As one of the elements of the revitalized HIS, a dedicated hydrological data processing system HYMOS, has been implemented at about 60 data processing centres of the Central Water Commission and water resources/irrigation departments of these nine States. The software is being effectively utilized at the middle two levels of the four-tier distributed HIS for the purpose of validation, in-filling, compilation, analysis, visualization and reporting of the surface water hydrological data. The introduction of the software in a multiple users environment at so many locations entailed a well-laid out training program. The paper highlights the institutional set-up of HIS in which the software has been employed, as also the various user-friendly features of data validation, analysis, visualization and reporting that are being advantageously used as uniform and standard tools by all the agencies.

Many countries are experiencing increases in both the demand for river flow data and the financial pressures which constrain monitoring. This paper describes the implementation of data management system designed to maximise accessibility to user-focused river flow data of an appropriate quality for meeting strategic information needs. The system is based on a Service Level Agreement (SLA) which governs the transfer of data from regional data providers to the national database, and the subsequent quality control of data. The SLA employs scoring mechanisms to monitor performance of providers in relation to the completeness and quality of data along with the timeliness of provision. The SLA mechanisms and the associated data management infrastructure are discussed and results for the first year of implementation are presented. The potential of the system for improving the utility of flow data is discussed from a strategic perspective.

The paper deals with development of an advanced and inexpensive pressure logger to be used for efficient dynamic monitoring of water supply distribution network. The device will store both the slow, steady state pressure data and the pressure transient using high sampling rates. Acquired records of pressure transients are valuable source of information in attempt to verify extended period simulation models and to manage the leakage in water distribution networks using Inverse Transient Methods.

Risk estimations associated with dam failures based on statistical studies were difficult to carry out, because either the information of different data base were contradictory or no data were available. Dams are complex systems, their behaviour in the moment of a failure is dependent on the dam type, to their dimensions, to the volume of stored water, to the geology of their foundations and to the construction material of their dam bodies. The comparison of different failure rates also faces difficulties, because some failures listings define "failure" as an accident that destroys a dam and renders it useless, while other means a catastrophic accident, which releases most or all of the impounded water. The Data Station for Dam Failures DSDF-VIENNA collects since more than 20 years data of dam failures and presents now an event based harmonized accident data collection, which allows the evaluation of an operational experience of 150 years of dam engineering, including the basic research information which are necessary for risk assessments and probabilistic studies of structural reliability. The data base documents more than 900 dam failures and 133 of those of tailing dams.

The following sections are included:

• INTRODUCTION

• THE SOFTWARE PACKAGE
  • Data acquisition and integration
  • Data storage
  • Data processing
  • Rating curves
  • Derived and aggregated values
  • Data analysis
  • Data dissemination

• THE USE OF WISKI AT THE ENVIRONMENT AGENCY OF ENGLAND AND WALES
  • Implementation of WISKI

• CONCLUSION

• DISCLAIMER

Quality and adequacy of hydrological data has a major role in design and operation of water resources projects. Hydrological Information System (HIS) in many countries lack reliability, accessibility and timeliness. A considerable improvement has been brought about in the HIS in India by implementing the Hydrology Project (1996–2003). Benefits due to organization of huge amounts of paper-based historical records in well-defined databases and introducing computerized data processing, management and dissemination are beginning to be recognized. There is considerable positive outlook among most of the concerned agencies with respect to sharing of data. Concerted efforts in the project area of 9 peninsular states have provided a unique experience. This paper discusses several issues that require attention for further growth of the HIS.

The calculation of the exact amount of water needed to bring the entire field precisely to field capacity is complicated by spatial variations in the soil characteristics such as soil texture. The objective of this study is to investigate the feasibility of using a global calibration Soil Electrical Resistance (SER) to determine the moisture content of a wide range of soil in a sugarcane plantation, in term of supporting computerized irrigation management and to determine the feasibility of applying the data acquisition system on variability of soils. Development of data acquisition system consists of 4 main hardware components: sensor, amplifier, data converter and personal computer. However, in practical application, the soil moisture reading from the sensor can be encountered by using a digital AVO-Meter. The system was calibrated by comparing between voltage measurement and soil moisture data obtained from the analysis using gravimetric methods. The calibration was conducted on 3 different soil types: clay, clay sandy and sand. Regression coefficients from the calibration is 0.99, 0.98 and 0.86 for clay, clay sandy and sandy soils, respectively. The system can provide initial soil moisture for supporting the computerized irrigation management in a sugarcane plantation. The Takalar Sugarcane Plantation expects the developed sensors can be applied for improving irrigation management at the plantation.

Water quality assessment requires processing of large volumes of data collected at numerous sites of a monitoring system. Particular data organization on secondary storage predetermines the efficiency of data processing. The comparison of two schemas built based on relational and multidimensional data models has demonstrated the possibility of application and the advantage of data warehousing technology for water quality assessment.

A comprehensive analysis of rainfall estimation by weather data in Southern Switzerland and Northern Italy is carried out. Main rationale of the study is the construction of a reliable rainfall field as input for a flood forecasting hydrological model in the Lake Maggiore watershed.

Validation of the MeteoSwiss weather radar vs. rainfall gauges is conducted by using a data set composed of 365 days (Year 2002) per 24 hourly observations at 35 different rain gauges: these have been homogenized by classifying gauge measurements as the radar estimates and further subdivided in two different seasonal datasets.

Preliminary analysis proves good compatibility between gauges and radar measurements in terms of rainfall detection, although radar generally underestimates the amount of precipitation. The functional dependence between the two variables displays a complicated behavior and therefore the radar adjustment can not be undertaken by using a simple linear model. Results of the statistical analyses are shown and a prototype model is presented.

The authors have proposed a new parameter, "time-related concentration of precipitation", for estimating probable unit time precipitation. The parameter can be estimated by examining intensities of unit time and representative time precipitation. As part of this research, we used the AMeDAS^[4] hourly precipitation data in Japan and obtained an equation: log (_ξR₁)= (m+1)log (R_T) + n, where _ξR₁ and R_T are amounts of probable unit time precipitation and representative time precipitation, respectively. _ξR₁ can be calculated with (m+1) and n that are a function of R_T and the normal variable ξ. Using this equation, it is possible to simulate probability distribution for amounts of the annual maximum unit time precipitation for Bangladesh^[5] data. Bangladesh has one rainy season a year and different climate from Japan, which has two rainy seasons. However, the method is sufficiently applicable and seems to be all-purpose. In addition, the study focuses on continuous measures of probability distribution for amounts of a given unit time precipitation by preparing coefficients.

The population of the EU has increased by more than 72 millions since 1960 with growth rates being positive in nearly all countries. Changes in population, population distribution and density are key factors influencing the demand for water. There are many pressures on water resources including those arising from agriculture, industry, urban areas, households and tourism. The problems of over-exploitation of water resources are extremely complex, not only from a hydrological point of view but also regarding the socioeconomic and political circumstances. Solutions have to be environmentally sound as well as socially and politically feasible. Upgraded wastewater reuse is a good solution to confront this pressure, but a thorough investigation must take place before reusing schemes could be introduced. Because of the long term deterioration of the quality and quantity of water, the European Council supported the AQUAREC research project, contract number: EVK1-CT-2002-00130, with the general objective to provide knowledge for developing rational wastewater reuse strategies. Project will provide research, national monitoring and data-gathering concentrate on the improvement of the present state of information, trying to establish reliable records on a European scale and provide meaningful information to decision-makers. The main tool to implement formulation of this project for the analysis of the European water Market, The Geographical Information Systems (GIS) was selected. GIS provides a means of integrating, handling and visualising water supply and demand data at various scales. Visualizing the different water related and geographical data will help to identify regions, where the increased use of upgraded wastewater should to be promoted. GIS will be used to reveal relations between different data layers using overlaying techniques and similar standard built-in GIS tools.

Using economic and ecological indicators and collected data, the boundary conditions for water reuse are being estimated. Next phase will be selection some several regions that will be analysed in a more detailed way assessing their water reuse potential. The findings of the work carried out will provide a basis for the formulation of policy guidelines and quality standards for water reuse in Europe.

This paper describes a new method of identifying flooded areas by using a differential image obtained from before- and after-disaster images. The method effectively uses three satellite images: (i) high resolution image after disaster, (ii) high resolution image before disaster, and (iii) low resolution image before disaster. Using (ii) and (iii) above, the method simulates a high resolution NDVI image for the identification. The method was applied to the central part of South Korea near Muju City, which was much damaged by Typhoon Rusa in August and September 2002. In the proposed method NOAA/AVHRR (1.1-km resolution) and Landsat-7/ETM⁺ (30-m resolution) images are used as before-disaster ones, while a Landsat-7/ETM⁺ image acquired on Sep. 3, 2002 is used as an after-disaster one. Comparing the proposed method with conventional methods, this paper concluded as follows. The conventional differential image method does not perform very well, which is affected by seasonal change in land surface. The modified conventional method performs better but still has some difficulty in upstream and urbanized areas. On the contrary, the new identification method with a simulated high resolution image can exclude the effects of seasonal change and has a good capability to identify only flood disaster area without misinterpretation.

In recent years, several typhoons had seriously hit the northern part of Taiwan in the Tamshui river basin, especially in the urban areas along the Keelung river basin. To protect 6.5 million people lives and their properties, the Keelung River Flood Mitigation Project is now undergoing. It consists of three main components: Yuanshanzi flood diversion, 11 flood protected zones and flood forecasting/warning system. It will be completed in May, 2005. Meanwhile, lots of public hearings and presentations will be held to demonstrate the progress of the project as well as to communicate with the public. The powerful 3D/VR techniques are used to generate interactive scenes, which create delicate 3D models such as buildings, levees, river flow, topography, etc. The virtual environment created by 3D/VR increases the efficiency of communication among designers, decision makers and the public.

A GIS methodology for the identification of 22 potential landslide dam locations for Westland, South Island, New Zealand is given. Five factors are used to identify sites: geology, elevation, slope, slope form, and slope orientation. Landslide dam geometry parameters are developed based on empirical methods. Model results compare favourably against an historical landslide dam site.

The increasing human population and industries in urban areas are posing serious problems to the parent watershed. Construction of roads, bridges, parking zone, high buildings etc. decreases the porosity of the land resulting in the increase of runoff during rainy season. Construction of cement roads and flooring of footpath drastically reduces the ground water recharge in the region, which results in depletion of ground water level. The present study attempts to identify the different parameters having impacts on the recharge process and which play important role in controlling the groundwater conditions of the Pioli Watershed. Remote sensing and Geographic Information System has been successfully employed in the present study as most important tools in analyzing watershed conditions. Thematic maps pertaining to various parameters like geology, geomorphology, hydrology, Land use/Land cover have been generated using IRS-LISS-III image and have been utilized to identify the hot spots within the urban watershed.

The paper deals with partial problems of water quality management. Attention is paid especially to application of mathematical modelling in combination with geographic information system for the operative solution in case of accidental pollution. There are two main possible causes of accidental pollution. It could be either technological failure or consequence of flooding. The paper discusses the problems related to the first case, the technological failure. In that case pollution is transported entirely in the stream. Therefore 1D numerical model (MIKE 11) is applied for the simulation of advection-diffusion processes. GIS (ArcEditor) is used for the data management (geo-database set up, pre-processing and post-processing), graphical presentation in form of maps representing progress of pollution in the stream. Furthermore, GIS is used as a tool of the risk analysis of potentially effected areas. Case study was made for a local river in an inhabited and partially industrialized area. The work is part of the grant project 103/03/P152 of Czech Science Foundation.

The coastal ocean includes estuaries and the region between shorelines and the beginnings of deep ocean; thus numerical ocean models dealing with the coastal oceans need to consider a large range of phenomena and scales in both time and space, which should handle with a variety of physical processes of different scales. For these complex coastal ocean models, marine geographical information systems will be a challenging problem. The integration Geographic Information Systems (GIS) with coastal ocean circulation model is conducted to revitalize the use of geographical information and to help the understanding of ocean circulation pattern in coastal waters. A two-dimensional finite difference circulation model is used to simulate the tidal circulation in the Suyoung Bay in Busan, Korea. GIS, especially using ArcView S/W, is used to make input data of numerical model and also used for the visualization of model outputs on the ground of loosely coupling method. In this study, an electronic navigational chart (ENC), which can gives more accurate information in the ocean and coastal area than other digital information, is used as a base map of this integration. The integration can support the understanding of coastal ocean circulation with the help of GIS and can be applied in coastal management.

Evaluation of impacts of urbanization in terms of land use/land cover patterns on the ground water quality of Zone-V of a part of Municipal Corporation of Hyderabad area is carried out using Remote sensing and Geographic Information System. Various thematic maps like base map, drainage and road network maps were prepared using SOI toposheets on 1:50,000 scale using AutoCAD and ARC/INFO software. Land use/land cover map of the study area is prepared from the linearly enhanced fused data of IRS-ID PAN and LISS-III imagery using visual interpretation technique. The ground water samples collected were analyzed physico-chemically for the generation of attribute database. Based on the results of the analysis, maps showing spatial distribution of water of selected water quality parameters such as pH, alkalinity, chlorides, Sulphate, nitrates, hardness, TDS fluorides and sodium are prepared using curve fitting method in Arc View GIS software. The Water Quality Index (WQI) in the study area was determined to find out the usability of water.

The results showed a high concentration of nitrates, hardness, alkalinity and TDS in areas like Yousufguda, Jubilee Hills, Banjara Hills, Erragadda and Shaikpet of Zone-V which could be due to the present landform, dumping sites, seepage of waste and effluents and also the weathering of mineral bearing rocks and leaching. The overall view of the water quality index of the present study zone showed a satisfactory result with most of the study area with < 50 standard rating of water quality index in Banjara hills, Jubilee Hills, Mehdipatnam etc except places like Yousufguda, Erragadda, Shaikpet and Hakimpet. Appropriate methods for improving the water quality in these areas have been suggested.

This paper is intended as an investigation on influence of changes for land use on flooding potential. Southern Taiwan Science Park (STSP), which was agricultural land in Yen-Shui river basin, is adopted as study region. There is a lot of changes in physiographic features of land use after the development of STSP. To compare flooding potential in Yen-Shui river basin, the hydrologic and physiographic parameters can be analyzed and dealt with by GIS, which linked the physiographic inundation model to estimate the influence of land use on the runoff, inundated depth and flooding potential. According to the Simulating results, the artificial development in STSP cause the increasing discharge and stage of peak flow and the change of flooding potential as well as the increasing inundated depth for local area in Yen-Shui river basin.

The groundwater resources in a 265 ha of highly diversified and intensive rice-based environment was endangered to NO₃-N contamination with spatial degree of influence and temporal vulnerability risks as affected by intensive cropping systems with application of high N-fertilizer and judicious use of groundwater for irrigation. Such nitrate contamination levels are above the World Health Organization's maximum contamination level of 10 ppm for drinking water. Tree-joining, complete cluster analysis of monthly groundwater depths from 1994-2002 on observation wells revealed three distinct groups of wells differentiated by groundwater depths. Planting of nitrate catch crops such as legumes to reduce groundwater contamination and vigorous information dissemination on ill-effects of high NO₃-N, as well as groundwater recharging were considered to reduce NO₃-N levels. The combined-use of GIS and GPS proved useful for spatial and temporal risk assessment on groundwater nitrate vulnerability among other geo-referenced attributes of groundwater at the study site. Such systems analysis tools can be used by planners, researchers, extensionists, research students and farmers for thematic mapping, assessment, extrapolation analysis and strategic planning.

This paper describes a new hybrid regression method that combines the best features of conventional numerical regression techniques with the genetic programming/symbolic regression technique. The key idea is to employ an evolutionary computing methodology to search for a model of the system/process being modeled and to employ parameter estimation to obtain coefficients using least squares. To illustrate applications of the new Evolutionary Polynomial Regression technique (EPR) to system identification a simple polynomial formula with progressively increasing level of noise was used in this paper. The results of the EPR approach show a very good agreement with the formula used.

Evolutionary Algorithms (EA) have been applied to a number of Water Distribution System design and operation optimization problems with great success. Many different EAs have been developed over the past 25 years, the most popular being the Genetic Algorithm (GA), and most of these algorithms have displayed the ability to be effective in the solution of a range of optimization problems. However, every case study is different, and just because a certain set of parameters leads to the best solution on one problem does not mean that it will work efficiently on another. To provide an insight into the difficulty of solving an optimization problem, and which parameters might work the best, a method of measuring the correlation structure of a given fitness function is proposed by the use of spatial autocorrelation statistics. Two scenarios of a simple distribution network have been optimized using a GA. The best parameters for the GA in each case were found from extensive sensitivity analyses. The results show there is a significant relationship between the correlation structure of a fitness function and the best parameters for the GA. It is expected that the proposed methodology will allow the most appropriate EA parameters to be selected for a given problem, allowing very good solutions to WDS optimization problems to be found in the smallest possible computational time.

The Dutch Waterboards are mandatory to establish the 'Legger', a register describing the design dimensions as well as the maintenance and management of the main canals. As a consequence of changes in the water system, such as subsiding, urbanization and increased discharge capacity, the criteria for the design of the canals become more demanding. The current register is therefore often out of date. In order to conform the register to today's demanding criteria the Dutch Waterboard of Rijnland decided to update the design dimensions of the canals. Because of the complexity of the water system, it is very difficult and timeconsuming to translate these criteria to the required optimal dimensions using traditional methods. However, it appears to be feasible to determine the required optimal dimensions in an effective way by using mathematical optimization techniques, supported by the increase of processing power of computers and the availability of advanced hydraulic computer programs.

Transient fluid analysis is a complicated problem and thus so is the optimization of its control system for water distribution systems. The purpose of this paper is to obtain the optimal design of the pipeline network considering both steady and transient state flows. Two global optimization methods, genetic algorithms (GA) and particle swarm optimization (PSO), are incorporated to obtain optimal pipe diameters in the pipeline system with allowance for water hammer conditions. The case study shows that the integration of GA or PSO with a transient analysis technique can improve the search for hydraulic protection strategies. This study also shows that not only the selection of pipe diameters is crucially sensitive for the surge protection strategies but also that more global systematic approaches should be involved in water distribution system design.

The leakage constitutes the largest portion of the unaccounted-for-water, which can easily reach 40% of the total water production. Thus the scarcity of new water resources forces authorities to consider more on the leakage control. Although there are various devices to detect leakage in water distribution networks, the use of these devices requires considerable investment, operation time and labor. In this study a systematic procedure to detect link leakage in conjunction with district metering method, which aims at reducing the time needed for the fieldwork, is proposed. Extensive numerical experiments are then conducted to illustrate and validate the efficiency of the proposed method. The method is shown to be useful and efficient in detecting leakage if the measurement of heads and flow rates and the estimation of roughness coefficients are accurate.

The optimisation of water resources systems using computer simulation models is characterised by two aspects. The used models are complex and a large number of parameters have to be considered. In the past it has been shown by several authors that evolutionary algorithms are a robust and effective technique to optimise water resources systems.

The optimisation of complex systems requires the combination of several penalty functions which may be of different type (economic related, ecological related, etc.). In addition, the conversion from emission to immission oriented types of decision variables often requires a series of models.

For these demands it is advantageous to use a modular software structure. In the developed software the simulation models, the properly graphical user interface (GUI) and the optimisation algorithm are strictly separated. This structure allows the inclusion of different simulation models and to perform a multi criteria decision.

The functionality and capability of the software could be shown by means of a complex urban drainage system including the receiving water.

This paper proposes a least cost design methodology for the design of water distribution systems based on the principles of particle swarm optimization. The objective of the optimisation is to minimize the capital cost, subject to ensuring adequate pressures at all nodes. Particle swarm optimization is a newly developed method and is based on the social behavior of bird flocking and combine local search methods (through self experience) with global search methods (through neighboring experience). The developed PSO-based method is compared with a genetic algorithm approach that has been widely applied to the design of water distribution systems. The results are promising and show the effectiveness and robustness of the proposed approach. In addition the method has been tested on several water distribution systems including systems used for benchmark testing least cost design algorithms, and has been shown to be very efficient and robust.

It is widely mentioned in hydroinformatics that hydrologic models serve as a decision support tool for the planning and management of water resources. One key step for the adequate implementation of these models, is their calibration, that is, finding the right parameters that correspond to the natural system that is to be studied. In this work, two meta-heuristic optimization methods were used for the calibration of the Thomas rainfall-runoff model applied to several catchments located near Bogotá, Colombia. The aim is to compare a simple genetic algorithm (with elitism) versus an ant system (adapted for continuous functions) in terms of their efficacy and efficiency. Knowing that for each method, different (and more effective) versions exist and that, for each version, different input values (algorithm's parameters) may be used, the aim is not to finally conclude which is the best, but to give a preliminary step and suggestions of what could be a rigorous comparison when applying these two methods to the Thomas model.

The paper focuses on the performance of three elitist multi-objective non-dominated sorting GAs (NSGAs) for the optimization of water distribution systems. Numerical analyses have been carried out for a two-objective problem of a simple two-loop network, in order to compute its true Pareto front as a basis for NSGAs evaluation. Two performance metrics have been taken into account, namely a convergence and a sparsity index, for a direct comparison of the final populations obtained by the algorithms. Results show how the controlled NSGA-II not only has better performances with respect to the two indices, but is also characterized by good stability over different simulations.

Nearshore process models are designed to resemble the physical processes that govern hydrodynamics, sediment transport and/or bathymetric evolution in (coastal) water depths less than about 8 m. The physically realistic functions implemented in these models are governed by parameters that usually do not represent measurable attributes of the nearshore and, therefore, need to be determined trough calibration. Here, a hybrid genetic algorithm, comprising a global population-evolution-based search strategy and a local Nelder-Mead simplex search, is used to calibrate nearshore process models in an objective and automatic manner. The effectiveness of the algorithm is examined with a 3-parameter alongshore current model.

A simple genetic programming (GP) software is developed and applied to identify environmental response function for Taiwan Salmon (Oncorhynchus masou formosanus). Genetic programming is one of heuristic algorithms. GP can not only optimize parameters but also function types, which is more flexible for identifying better functions. The simple GP software is applied to identify the relationship between Taiwan Salmon's population and two key environmental factors, air temperature and extreme streamflow. The results indicate that GP successfully identify environmental response functions for Taiwan Salmon.

The successful application of a conceptual rainfall runoff model depends on how well the model is calibrated. In the quest for developing a robust and efficient optimization algorithm for rainfall-runoff model calibration, an evolutionary algorithm (EA), the shuffled complex evolution (SCE-UA) method was developed at the University of Arizona. For an EA-based model calibrating technique to be successful, two important facets of the search - exploration and exploitation of the search space need to be addressed. In the SCE-UA, the use of multiple complexes and their periodic shuffling provide a more effective exploration of the search space. This is coupled with evolution of each simplex using the downhill simplex algorithm, which provides effective exploitation.

In this study, instead of using multiple complexes and their shuffling, we propose a new scheme to improve the exploration capability of the SCE-UA in finding the global optimum. The scheme used is a systematically located initial population instead of a randomly generated one used in SCE-UA. Thus, a single complex with a systematic initial population provides the exploration and the downhill simplex provides the exploitation of the search space. The robustness and efficiency of the modified SCE-UA was compared with the original SCE-UA on a suite of commonly used test functions. It is observed that the modified SCE-UA leads to a superior balance between exploration and exploitation, resulting in a significant improvement in the robustness of the algorithm.

Satisfying conflicting objectives in the rehabilitation of pipe network for water distribution systems is still a challenging task for both researchers and water managers. Many evolutionary algorithms are available to determine the optimal rehabilitation options. However, they need large number of evaluations of the objective function and high computational time even to solve simple problems. In this paper, an efficient evolutionary algorithm, Shuffled Complex Evolution (SCE), which is based on Nelder and Mead's simplex search method has been successfully applied to the design and rehabilitation of water distribution network problem, New York City water tunnel. The efficiency of SCE in finding the decision alternative is discussed and comparison is made with results of other published techniques. The results show that SCE reaches to the optimal solution very much faster than the widely used Genetic Algorithm (GA) as well as the recently introduced algorithms Shuffled Frog Leaping Algorithm (SFLA), and Ant Colony Optimization Algorithms (ACOAs).

Harmful algal blooms (HAB) have been widely reported and have become a serious environmental problem world wide due to its negative impacts to aquatic ecosystems, fisheries, and human health. A capability to predict the occurrence of algal blooms with an acceptable accuracy and lead-time would clearly be very beneficial to fisheries and environmental management. In this study, we present the first real-time modelling and prediction of algal blooms using a data driven evolutionary algorithm, Genetic Programming (GP). The prediction of the algal blooms is carried out at Kat O station in Hong Kong using 3 years of high frequency (two-hourly) chlorophyll fluorescence and related hydro-meteorological and water quality data. The results for daily prediction of chlorophyll fluorescence, a measure of algal biomass, are within reasonable accuracy for a lead-time of up to 2 days. As compared to traditional data-driven models, GP has the advantage of evolving an equation relating input and output variables. A detailed analysis of the results of the GP models shows that GP correctly identifies the key input variables in accordance with ecological reasoning; the results generally concur with that obtained with artificial neural network. GP has been shown to be a viable alternative for real-time algal bloom prediction, with an added advantage of interpreting the evolved equations.

The water distribution system (WDS) rehabilitation problem is formulated and solved here as an optimisation problem under uncertainty. The objective is to maximise the WDS robustness subject to the rehabilitation budget available. System robustness is defined as the probability of satisfying the minimum pressure head requirements at all network nodes under uncertain demand conditions. Decision variables are the alternative rehabilitation options for each pipe in the network. All uncertain nodal demands are assigned a probability density function (PDF) in the problem formulation phase. The optimisation problem is solved using genetic algorithms (GAs) linked to a stochastic WDS hydraulic model. However, rather than using a large number of random Monte-Carlo samples for each fitness evaluation throughout the GA search process, an alternative approach based on significantly smaller number of samples is developed. This new optimisation methodology is tested on a case study. The results obtained indicate that the method is capable of identifying near optimal solutions despite significantly reduced computational effort.

The problem of the optimal design of water distribution networks with uncertain demands at nodes is considered. It is assumed that demands are statistical variables with given probability distribution functions. Two parameters are subject to minimisation – a) cost of the network design/rehabilitation; b) probability of network failure due to demand fluctuation. The stochastic formulation for the second parameter requires using Monte Carlo simulation, which means calculating the state estimate several hundreds or thousands of times even for relatively simple networks. This is unacceptable when using genetic algorithms (GA), since you need to calculate the fitness function for thousands of possible network configurations. In this paper the original stochastic formulation is replaced with a deterministic approximation, using standard deviation as the natural measure of the variability of the pressure in the nodes caused by uncertainty in demands. Such an approach allows us to use effective numerical methods to quantify the influence of the uncertainty on the robustness of the water distribution system and gives us a fast but still reliable way of estimating fitness function, suitable for using in GA's. The proposed methodology was tested on the New York tunnel problem. The optimum networks were found for different levels of reliability. The robustness of the optimal networks found was compared with known solutions for a deterministic formulation, using Monte Carlo simulation.

Booster disinfection is the addition of disinfectant at some critical locations of the network such that minimum disinfectant residuals are maintained at various demand nodes of a water distribution network. Compared to conventional methods that apply disinfectant only at source, booster disinfection can reduce the total disinfectant dose, which may lead to reduction in the formation of by-products, which are harmful to human health. This paper investigates the booster facility location and injection scheduling problem using multi-objective genetic algorithms. Previous researchers have studied either mass injection type boosters and/or flow-paced type boosters. However, most commonly used and preferred booster type is set-point booster. Therefore, in this study the problem of optimal set-point booster facility location and injection scheduling (set points for set-point type boosters) in a water distribution network is investigated. The problem is formulated as a two-objective problem. The objectives are minimization of total disinfectant dose and maximization of volume of water with residuals in specified limits. A Multi-objective genetic algorithm is used to solve the two-objective model. The model is applied to an example problem, which was previously published, and trade-off curves are prepared between total disinfectant dose and level of constraint satisfaction for different number of booster stations.

A heuristic method based on trade-off between reliability and cost is proposed for strengthening and expansion of water distribution networks to obtain a level-one redundant network, i.e., a network that can sustain single pipe-failure without affecting consumer services either in part or in full. The method is illustrated with an example of a network in Washington, D.C., USA. The results are compared with those obtained by linear-programming based algorithm.

This paper isolates and investigates pressure zones as separate components of actual water networks and seeks a solution controlled by the joint interaction operation of storage tanks with controlled pumping activities. With assumed friction factors for the pipes, the network is solved under simulated fixed and variable daily demand load conditions. Simulated deterioration of pipes is also assumed and the network is calibrated by calculating new values for the pipes frictional factors using modified version of the Genetic Algorithm technique. In the current GA scheme, two best fit selection criteria were used instead of one, one before and one after the genetic breeding process. This was found to enhance the accuracy and convergence process. The analysis found optimized values for pump operation controlled by reservoir level settings in an isolated pressure zone of a network. Comparisons were made to see the effect of variable daily demand as compared to fixed diurnal demand on the interaction between periods of pressure zone reservoir filling and emptying optimized with the rate of power consumption of the controlling pumps. Variable and constant values of diurnal water consumption at an average of 0.15 m3/s were assumed. The resulting optimized pump setting range values for both were found to be close to 30% in tank volumetric setting. However, the performance of the reservoir in keeping up with demand requirements was only evident in the constant consumption case. The tank volumetric setting mentioned contributed to reducing the total operation of the pumping station by about 20 minutes per day in both cases.

It is often necessary to have stage discharge curve extended (extrapolated) beyond the highest (and sometimes lowest) measured discharges, for river forecasting, flood control and supply for agricultural/industrial uses During the floods or high stages, the river may become inaccessible for discharge measurement. Rating curves are usually extended using log-log axes, which are reported to have a number of problems. This paper suggests the use of Support Vector Machine (SVM) in the extrapolation of rating curves, which works on the principle of linear regression on a higher dimensional feature space SVM is applied to extend the rating curves developed at two gauging stations in Washington, namely Chehalis River at Dryad and Morse Creek at Four Seasons Ranch (for extension) of high stages) and Bear Branch near Naselle (for extension of low stages). The results obtained are significantly better as compared with widely used logarithmic method and higher order polynomial fitting method.

This paper proposes a new approach for the automatic calibration of numerical simulation models using an innovative hybridisation of genetic algorithm and artificial neural network. The proposed method reduces the number of simulation runs required in the numerical model considerably thus making the automatic calibration of computationally intensive hydraulic models viable. The new approach was developed and tested in the calibration of a popular rainfall-runoff model, MIKE11/NAM, applied to a Danish catchment. The results for single objective calibration indicate that the proposed method is able to achieve the same or better calibration performance compared to traditional soft computing techniques and yet required only about 40% of the simulation runs on average.

The major difficulty concerning the use of conceptual rainfall-runoff (CRR) models in practice is related to their calibration since most of these models involve a large number of parameters. In general, the CRR model calibration could be considered as a global optimization problem since its main objective is to find a set of physically plausible model parameter values that provide the best fit between observed and estimated flow hydrographs. However, although several global optimization methods (GOM) have been proposed there is no general agreement as to which method is the most appropriate one for CRR calibration purposes. The main objective of the present paper is therefore to propose a systematic procedure for assessing the robustness, accuracy, and efficiency of seven popular GOMs. The performance of these methods were judged based on the computation of the optimal solution of a number of benchmark theoretical functions as well as on the calibration of a typical CRR TANK model using both real and synthetic runoff data. It has been found that the Multistart and the original SCE methods are the most robust and the most accurate, while the Control Random Search and Adaptive Cluster Covering methods are the most efficient. The modified SCE might provide less accurate calibration results using real data. Finally, the Simulated Annealing and Tabu techniques were found to be the least satisfactory.

This paper presents an interactive multiobjective decision making procedure to analyze a trade-off between water storage and hydroelectric energy generation in the Han-River basin in Korea. In the multiobjective multireservoir operation, it is desirable to provide a dam operator with a wide range of choices of Pareto-optimal solutions since the reservoir operation plan must be developed based on the dam operator's intuitive interpretation of storage variation of each reservoir as well as the utility values implied from the tradeoff relationships between storage and hydroelectric energy generation. Therefore, we focus on the ability of generating well-distributed Pareto-optimal solutions at each iteration and pinpointing a final solution properly. To facilitate pinpointing a final solution, the proposed procedure parameterizes the Pareto-frontier via the set of the reference point vectors instead of weight vectors within the framework of Interactive Weighted Tchebycheff Procedure. It uses a sequence of a progressively reduced subset of reference points interacting with the dam operator so that he can sample corresponding sequence of the Pareto-optimal solutions. Using well-distributed reference point vectors, we could identify widely distributed Pareto-optimal solutions and reduce the chance of losing the most preferable Pareto-optimal solution from the dam operator's utility point of view.

In this paper a new multi-objective GA method was brought forward by which the Pareto set can be searched by calculating only once based on the variety of the plentiful solution in every generation in the evolution. The appraising function in this method ranks the gene by comparing their performance in each objective function through the sorting matrix created by the objective function. In order to increase the feasibility of this method, the parameters' calibration is improved and the Pareto solution's selection is controlled effectively. Applied in a multi-objective reservoir's long-term optimal operation for generation and water supply, it is proved that this new method is feasible.

Groundwater resources play a diverse but vital role in regional water supplies, and generally depend on a network of pumping wells to distribute water to the community. The design of the network determines its capacity, the goal of which at the time of construction is to satisfy the water demand of a community into the foreseeable future. Since water demand increases over time, the system may be over designed initially, but eventually needs to be expanded if water demands exceed the maximum capacity of the system. Therefore, this research proposes a novel optimal capacity-expansion model capable of determining an optimal schedule to expand the system capacity according to increasing water demand.

The proposed model assumes a 2-D groundwater flow for a confined and unconfined aquifer while combining the Genetic Algorithm (GA) and Differential Dynamic Programming (DDP) to solve the optimization problem. The main structure of the hybrid algorithm is GA, in which each chromosome represents a possible network design and its expansion schedule. The fixed cost of each chromosome can be computed easily. The DDP then solves the optimal pumping and injecting scheme while evaluating the optimal operating costs associated with each chromosome.

Simulation results indicate that this capacity-expansion model saves more of the present value of the total cost for the same annul interest rate and water requirement than the conventional design that determined the system capacity at the beginning. Furthermore, the benefits of using this capacity-expansion model increases with a rising interest rate. Results further demonstrate that the proposed model is highly promising for use in facilitating a cost-efficient well system design for regional groundwater supplies.

The purpose of the research described in this paper is to provide meta-level support to the process of constructing and interpreting complex composite models. Above the numerical computations that hydroinformatic models are enacting, we think of a semantic layer of evidential reasoning about models, their relationship with measured observations and the extent to which their results provide a sound basis for decision-making. In the past this evidential reasoning has been done by humans, but, given the almost bewildering complexity of modelling systems either existing or soon to emerge, it is now necessary to supplement human reasoning with formal methods. In this paper the principles upon which our approach is based are outlined briefly. A preliminary implementation of some simple examples using the FRIL Fuzzy Reasoning and Inference Language is presented.

Water is one of life's most precious elements, yet it is often managed so poorly that some networks lack a sufficient supply for many days. Even if the situation is not this bad everywhere, there are many parts of the world where almost half the water is lost in the network. One such example is Jakarta, the capital of Indonesia where significant improvements have been achieved despite the numerous technical difficulties. This was achieved by applying a systematic and logical approach to leakage location which preclude the application of the more traditional techniques.

There exists a strong need to explore the integration of simulation techniques and decision making process under a total framework that not only represents complex and dynamic nature of river aquatic environment but also assists users in designing sustainable management strategies. To fulfill this goal, the current research has applied the approach of System Dynamic (SD) and developed the decision support system (DSS) for the Love River in the city of Kaohsiung, Taiwan. The SD software STELLA, used for the DSS implementation, is not only capable of building simulation models, but also efficient in developing user interfaces. The user of the system is allowed to experiment with designate scenarios by adjusting levels of some variables through a user friendly interface. One applicable case is to assist the decision makers in identifying appropriate gate operations of the sewer intercept system in a rainfall that would create less impact to the river aquatic environment. The benefit of using such DSS is therefore obvious that managerial efficiency for the river can be enhanced toward sustainability.

In the design or rehabilitation of any water distribution systems, the designer has to decide a priori the location and water demand (quality and quantity) of different users, the required serviceability and performance of the designed system and translate these requirements into either design parameters or constraints on the system. However, the above selection process may lead to sub-optimal design due to: (1) uncertainties in the estimation of water demands at the design stage; (2) the cost is usually used as a sole criteria for the design; (3) once the serviceability and performance are decided, they are seldom changed during or after the selection process.

With the recent advances in evolutionary computing, it is becoming increasingly feasible for the selection step of the design process to be semi, if not fully, automatic. However, an automatic selection method, if left unchecked, may lead to even less desirable designs. This paper presents a novel usage of a multi-objective genetic algorithm coupled with an "expert heuristic" search method to alleviate the problem of design inflexibility. The designer will be able to change the serviceability and performance requirements during the automatic selection process to come up with a wide range of new designs not explicitly identified earlier. Through the interrogation of the Pareto front, the designer may decide that a cheaper system may be more cost effective even though the requirements are not fully met. An application example of the rehabilitation of the New York City tunnel water distribution system is presented.

A method with multi-layer and mesh-typed runoff model using Hydro-BEAM (Hydrological River Basin Environment Assessment Model) is proposed to analyse the integrated hydrological processes. The spatiotemporal simulation is calculated with the kinematic wave model for surface runoff, Richard's equation for unsaturated subsurface flow and the unconfined flow for groundwater. The initial loss of rainfall due to interception by depression storage reprocess is considered here. Moreover the basin division and land use dynamics are introduced to encounter reservoir operation and land utilization with human activities. The proposed model is calibrated for different initial conditions and parameters, and applied into the Yasu River to verify the dynamic linkage between surface and groundwater.

Water resources management involves complicated social, organizational, legal and economical issues in addition to the undoubtedly important technical matters and environmental aspects. Management decisions have potential to be controversial because the involved groups (ecologists, economists, hydrologists and sociologists) hold distinct interests and unique objectives. Therefore, the development of decision support systems (DSS) faces great difficulties not only because of the multiple objective and multiple participant decision making situation but also due to the complexity of participating an team interdisciplinary in the modelling processes. For instance, while ecologists use natural languages and qualitative reasoning for the description of ecological relationships, hydrologists communicate in the form of systems of differential equations or analytical models. The proposed concept suggests to identify specific performance indicators to quantify the multiple objectives and assess the management scenarios under investigation. The indicators are classified into ecological, economical, social and technical indicators. To overcome the problem of multiple participant, the Analytical Hierarchy Process (AHP) is implemented to assign weights to the involved decision makers with respect to their experience and background. The management scenarios will be assessed and ranked according to the predefined performance indicators with respect to the opinion of decision makers taking into consideration their assigned weights. Fuzzy logic (FL) is used to facilitate transformation of qualitative knowledge into mathematical assessment models. FL is also implemented to model and aggregate the performance indicators. Finally, the paper introduces a case study for implementing this concept for the management of irrigation schemes.

The purpose of this paper is to outline a decision-support framework for stormwater control in Taipei urban area by means of reviewing the development of Taipei's storm sewer system history and reporting a series of projects administrated by the Taipei City Government nowadays. A pilot project of the best management practice for stormwater control in Taipei is also introduced in the paper. Taipei's experiences may be valuable for other cities.

The use of nutrients is central to modern farming because of the requirement for high yields. Unfortunately this often results in nutrient rich runoff into waterways especially during storm events. Two hydroinformatics tools are presented which, in a clear, precise manner, can educate farmers and policy makers on good practice for reducing nutrient loss and persuade them to implement such measures. TopManage, a high resolution digital terrain analysis toolkit for runoff management, highlights likely flow paths, operation of critical source areas and identifies potential locations for controlling nutrient loss without compromising the economic viability of farms. By operating at a high grid resolution TopManage enables the effect of man-made features such as tramlines, land drains and hedgerows to be modelled. TopManage is designed to demonstrate to the wider community that we understand runoff, pollution and how to prioritise the locations best suited to manage farmland. The phosphorous export risk matrix is a policy tool that integrates this understanding of runoff with a number of agronomic and policy factors into a clear problem-solving framework to allow farmers and policy makers to visualise strategies for reducing nutrient loss through proactive land management. The risk of pollution is assessed by asking a series of informed questions relating to farming intensity and practice. This information is combined with the concept of runoff management to point towards simple, practical mitigation strategies which do not compromise farmers' ability to obtain sound economic returns from their crop and livestock.

Integrated river basin management is a process, in which hydrological, ecological and socio-economic conditions have to be considered. Stakeholder participation belongs into the key elements of planning and decision making. Acceptance of planned measures, social justice and equity among the different groups of interests demand public participation. This paper is focused on a Spatial Decision Support System (SDSS) which can be used interactively with a posterior specification of reasonable goals as part of the mediation process. Based on an assessment of the ecological and socio-economic changes, measures and management strategies have to be defined, which should be implemented to reach a good ecological state of the rivers within a basin after a given deadline. Within a joint research project with ecologists, engineers, IT-specialists and hydrologists a SDSS is designed by the Ruhr-University to develop management schemes for the implementation of the European Water Framework Directive. It is based on a comprehensive analysis of the efficiency and feasibility of water management strategies. Use cases and workflows of river basin managers are specified using the Unified Modeling Language (UML). Workflow oriented tools to manage, evaluate and communicate data and decisions are developed. A structured system design ensures the placing of data within a dynamic context of driving forces, pressures, status, impacts and responses. Special emphasis has to be given to the spatial distribution and the different temporal and spatial scales of socio-economic, ecological and hydrological data and information. The result will be an interactive, learning-based spatial decision support system, which allows multi-criteria trade-off in participatory group decision situations.

The WARGI optimization package aims at giving an operating tool for the mathematical optimization of complex and large-scale water resource systems with the conjunctive use of conventional and marginal water sources, taking into consideration different water demands. The problem has been faced using modeling methods appropriate to the structural complexity of the optimization problem. WARGI allows for the possibility of starting from the physical system and can also control all the intermediate phases; it is easy to update the system configuration and consider different optimizers. WARGI prevents obsolescence of optimizer codes exploiting the standard input format MPS. It is easy for the user to modify system configuration and related data to perform sensitivity analysis and to process data uncertainty using scenario optimization.

This paper presents an approach for model calibration support as an alternative to currently used trial-and-error and optimization methods. The aim is to make expert knowledge available for end-users of simulation codes to help them achieve this time-consuming task. We have developed a prototype knowledge-based system which includes both 1-D hydraulic modeling tools and knowledge about their use in a calibration context. Tests have been conducted on data from a previously elaborated model of the downstream part of Hogneau River in order to verify both consistency and efficiency of implemented reasoning processes.

This paper presents a decision support system for the multireservoir system operation of Upper Yellow River. This system has 4 subsystems including database management subsystem, operation rules deriving subsystem, operational scheme making subsystem and information service subsystem. The function of the database management subsystem is to manage all the data used in the system, including the original data, the middle-result data and the operational data, and three databases are set up to manage these data. The function of the operation rules deriving subsystem is to derive the reservoir operation rules which includes several modules, such as the deterministic optimization module, the rules analysis module and rule functions deriving module. The function of the operation scheme making subsystem is to make the operational scheme for stage, annual and any stages based on the forecasting inflows and the derived rules. The information service subsystem is to provide a query function for searching all kind of information stored in the system that include the original data, middle-result data and the operational scheme. This system is developed under the windows operation system with VB6 which has an interactive man-machine interface to facilitate the decision making process. This system has been put into operation and works well.

A variety of Multi-Criteria Decision Analysis (MCDA) tools, including (i) outranking methods, (ii) preference-based methods and (iii) the analytic hierarchy process, have previously been applied to many different types of environmental problems. However, it has been found that there is no one definitive method that is appropriate for all decision-making situations. It is therefore difficult to assess which of the many MCDA methods is best suited for each task. Some of the problems arise from the decision-maker's bias and preference, in part due to a familiarity and affinity to a favourite MCDA tool, which clearly influences the final solution.

This paper identifies three different outranking MCDA tools that are practicably applicable to a real situation and discusses their relative merits in supporting the management of the disposal of domestic sanitary waste in the most sustainable way. The paper briefly describes the options and criteria used in a case study conducted within a test catchment area. In the drive towards sustainability, the social, economic, and environmental aspects together with the technical and cost-effective elements of each option are analysed. Six scenarios and 16 primary criteria were considered, along with the opinion of professionals as decision-makers. MCDA tools were employed as Decision Support Systems (DSS) to assess which of the six alternatives provided the preferred solution in the context of the best compromise across all criteria.

The primary objective of this research is to design a computer model for paddy irrigation management in Indonesia. The computer model was written in Delphi Programming Language. It consists of five sub models. The first sub model is DBMS (Database Management System), it is used for managing input data for the system. The second sub models deals with predicting climate variable that is used as input data. The prediction method follows Weibull Plotting Position. The third sub model is SWP, this is the core of the system that is intended for simulating water demand. The water demand simulation follows soil-water balance concept. The forth sub model is developed for presenting output of the simulation. The output of the system is water delivery scheme that describe which present irrigation area and amount of water should be applied. The output of the simulation can be presented not only in tabular form but also in graphical form. The model provides the possibility to evaluate the irrigation network during and after growing seasons. Three possibilities are provided in this model, namely; relative water supply (RWS) or reduction factor (K) and cumulative water supply (CWS). The program was validated and tested at Langkemme Irrigation Network, Indonesia. From the test, it was found that the maximum of average values of RWS indicator is found to be 41.1. The minimum value of this indicators is 1.6. From the average values of RWS and CWS indicators, it can be concluded that during the growing season, most of the irrigation performances are good.

An integrated software for real-time monitoring of coastal environment such as currents, wave and water quality has been developed. The system consists of real-time modeling module, real-time monitoring module, and post-processing module of the modeling and monitoring results. In the system the model results are corrected with the real-time observed data and the corrected results are served for user group. The main purpose in developing the system is to offer the spatial distribution of environmental parameters accurately and quickly through combination of modeling and real-time field monitoring. The system will be used for monitoring the environmental changes due to construction activities in coastal waters and assessing the environmental impacts in real-time. In conclusion the system will be a good tool for finding out countermeasures to protect seawater quality.

Groundwater resources management in water supply areas is nowadays an essential task to minimize the anthropogenic interference and to ensure the sustainable use of the available water. A management system has been developed that takes the interests of several stakeholders into account, where stakeholders are Ecology, Agriculture, and Water Supply. The purpose of this project is to provide recommendations for the amount and the distribution of groundwater extractions from various wells in a water supply area on the basis of a defined evaluation system and a forecasting groundwater flow model. The evaluation system is the objective base for the optimization algorithm to find the optimized extraction rates for all stakeholders. This approach was applied to a water supply area in Germany, providing drinking water for nearly three million people. In addition, the extraction area is partially covered by a peat bog of European importance and is intensively used for agriculture. The optimization presented here was performed for steady state flow processes. The transient optimization requires transient evaluation values which have been developed for the seasonal plant growth in ecological and agricultural areas. A continuous use of the management system is planned as a decision support system for the water supplier.

The concept of an integrated approach to water resources and river basin management or planning has become commonly accepted. In general conflictive demands have to be balanced. This leads to a multi criteria optimisation problem. Possibilities for applications and the convenient structure of a generic Decision Support System are discussed. The functionalities of the main components of this framework are described and their interactions during the planning process are outlined. The presented methodology achieves improvement for current management practice. A more efficient use of available instruments is obtained. Furthermore the objectivity and transparency during the analysis and comparison of alternative management strategies is increased. The consideration of innovative measures and technology is promoted. The involvement of stakeholders and public participation during the planning process augments the acceptance of derived management plans.

A real-time decision support system incorporating data acquisition, data visualisation, data management, hydrodynamic water quality modelling and forecasting has been developed for two drinking water reservoirs in Sydney, Australia. The system supports both operational and planning related decisions over short (weekly) to long (inter-annual) timescales.

Various pressures and needs are driving demand for ever more elaborate models, thorough analysis of sensitivity and uncertainty, and rigorous quality assurance. A number of contemporary projects are looking at one or other of these issues, but the pressing need for a common meta-model or ontology of modelling – a model of models and of the modelling process – is overlooked. Meaningful tool integration can only be achieved if those tools share a schematisation of the world, or ontology. Easy tool integration can only be achieved if those tools use a shared data representation grounded in that shared ontology. This paper introduces an effort to develop such an ontology and representation, the Model Description Framework, layered on top of the Resource Description Framework (RDF) of the World Wide Web Consortium. The use of RDF provides a number of benefits, among which are its capacity to provide the basis for an Internet scale infrastructure for distributed collaboration around shared descriptions of models.

The SMURF project aims to help coordinate the urban development planning process with management of the water environment, by providing a decision support system to be used by urban planners and water managers. The project will demonstrate this coordinated approach for the highly modified and degraded catchment of the River Tame in the West Midlands, UK. The integrated software system has been designed and is now being built. It combines a GIS user-interface, with a database for water quality and ecology, and automatic running of hydrological models of the catchment. This paper describes the requirements for this system and the attributes it needs to posses.

In many countries the Internet became element of a daily life, offering new possibilities for information access and sharing. On the other end there are growing concerns that not only technical specialists and politicians should be involved in decision making associated with utilization of natural resources. Many groups within the societies express the willingness and desire not only to be barely informed about decisions made by technicians and politicians; they also voice requests to be actively involved in decision making. This creates challenges to political and legal systems, it also creates entirely new situation confronting those actively involved in development of models and tools applied during decision making. Drawing from these developments the authors are reviewing the status of current trends in the area of decision support systems (DSS) for water resources. The authors discuss capabilities of these tools to provide understandable and meaningful information for stakeholders of the decision making. The capabilities offered by Internet technologies are discussed and major challenges are presented. The discussion is enhanced by the presentation of a prototype web-based DSS allowing users to select among possible development strategies in a case system and analyze consequences of chosen strategies. The response of the case system to chosen development strategies is computed using complex spatial simulation models.

The existing GIS based Urban Water modeling environment 3DNET has been expanded to accommodate transients flow simulation. Several algorithms have been developed to enable seamless connection between Urban Water System (UWS) database and simulation models. The main idea of this approach is to provide means to perform transient analyses on daily basis, for on-line pollution risk assessments, in addition to analyses done by specialist in design phase of the system.

RAP (Rapid Assessment Program) is a user-friendly and lightweight software implementation of a methodology for rapid, integrated policy analysis. Areas of application for RAP are virtually unlimited but have so far been mainly focussed on water resources management and scenario development in environmental engineering. Using an attractive visual interface, RAP takes its user through the steps involved in systematic policy development. While the overall organisation of these steps is fairly generic, each step in itself is designed to contain project-specific information made available by the user. This paper discusses the RAP methodology, and puts particular focus on the qualitative cause-effect processing engine in a broader state-of-the-art perspective. Other RAP features are also addressed, such as functionality to incorporate stakeholder valuation, and analytical tools for backtracking, and correlation analysis.

Multicriterion Decision Making (MCDM) has emerged as an effective methodology as it can integrate quantitative and qualitative criteria for selection of the best alternative. On the other hand, fuzzy logic is gaining importance due to its flexibility in handling imprecise subjective data obtained due to uncertain environment. In the present study two fuzzy logic based MCDM methods are adopted. The two selected methods are implemented in Visual Basic environment to develop a decision support system and named as FDSS (Fuzzy Decision Support System). FDSS is applied to a case study of Sri Ram Sagar Project (SRSP), India for selecting best performing irrigation sub system. Both the methods suggested the same irrigation sub system as the best. It is concluded that integration of MCDM with fuzzy logic methodology for real world decision making is found to be effective.

RAP (Rapid Assessment Program) is a user-friendly and lightweight software implementation of a methodology for rapid, integrated policy analysis. Areas of application for RAP are virtually unlimited but have so far been mainly focussed on water resources management and scenario development in environmental engineering. RAP attempts to support the user in various ways, from structuring the policy analysis approach, via the building of a model representation of the problem-system to presenting effects of possible interventions. This paper explains the RAP methodology by going through all steps involved for a specific project, discussing all the elements, aspects, problems, solutions and pitfalls of the process as well as the lessons learned. It also describes the evolution of the methodology and the software as a result of its application in this and other projects.

Methods for improving the hydrological simulation through the use of multi-model ensembles (MME) are demonstrated. In recent years, the meteorological community has exploited several MME combination techniques as a means for improving short-term weather and seasonal climate forecasts. Within the hydrological community, little work has been carried out to explore the benefits of MMEs for streamflow simulations. This study examines the use of MMEs for improving streamflow simulation, including their potential benefits to flood simulation. Multimodel ensembles are generated using ten distinct model structures derived using a new hydrological modelling tool. These model structures were applied to a US NWS test catchment, the Blue River and evaluated using a split-sample procedure. The resulting ensemble can be used to make probabilistic simulations that characterise model structure uncertainty. Furthermore it is shown that the ensemble average of all 10 models performs better than any single model in split sample test. Using regression methods, improvements in ensemble simulations using linear combinations of the ensemble members were explored. The performance of the resulting weighted ensemble is similar to the simple ensemble average but uses a smaller ensemble. This may provide a means to identify which model structures provide significant contributions to accurate hydrological simulation.

Data assimilation in operational modeling systems is a discipline undergoing a rapid development. Despite the ever-increasing computational resources, it requires efficient as well as robust assimilation schemes to support on-line prediction products. The parameter considered for assimilation here is water levels from tide gauge stations. The approach examines the combination of the Ensemble Kalman Filter with spatial regularization techniques. A steady Kalman gain approximation of the Ensemble Kalman Filter is further adopted. The estimation skill of the assimilation scheme is assessed in a forecast experiment using an operational model covering the North Sea - Baltic Sea system. The computationally efficient filter works well. At the time of forecast, the use of distance regularization gives much improved analysis results in data sparse areas, while maintaining performance in areas with a denser distribution of tide gauges. Hence, filtering with distance regularization estimates a state closer to reality in the entire domain and thus significantly improves the forecast skill.

Provision of early flood warning to at risk properties is an important strategy in reducing flood damage and loss of life. To increase warning lead-time and mitigate impacts more efficiently, flood forecasting systems are increasingly becoming an essential step in the warning process. Development of these has traditionally been instigated by local authorities, and in the early stages these systems were often no more than a dedicated user interface around hydrological and hydraulic models. Recent advances in weather forecasting, radar data and on-line meteorological and hydrological data collection, however, require for an increasing focus on data import and processing. Together with progress in database development, hydrological and hydraulic model development and on-line data availability, the challenges for developing a modern flood forecasting system is found in the integration of large data sets, modules to process the data and integration of various existing models. In this paper an open shell flood forecasting system is presented that provides essential generic functionality for handling real-time data, data assimilation and managing forecast runs, while also allowing integration of existing forecasting modules through an open (XML based) interface. The modular structure of the system and generic forecasting functionality are shown to allow natural integration of the system in the flood warning process, without the requirement of extensive migration to a specific modelling environment.

A brief review of researches is presented on the impact of the impoundment of the Three Gorges Reservoir, began in June 2003. The impoundment leads to significant reduction of sediment transport in the downstream alluvial channel of the Yangtze River, and provides a unique opportunity to test the previous theoretical analysis and numerical predictions concerning the pattern change of the lower Jingjiang River, which is the most vulnerable section endangered by Yangtze River flood. Close monitoring of the channel pattern change in the next few decades is necessary to understand the processes and to serve as a means of early warning, in case a potentially disastrous situation does build up.

In this paper, a method for predicting low and high precipitation seasons in the Karoon River basin in Southwest of Iran is presented. The method consists of three steps. In the first step, the correlation between Sea Level Pressure (SLP) variations and the average precipitation in the study area is investigated. In the second step, the fuzzy regions and the membership functions for the SLP gradient in the selected locations are developed. In the last step, the fuzzy rules are developed to incorporate the SLP variations in precipitation prediction. The fuzzy rules are used to predict the range of total precipitation in the period of 1990 to 2002. The results have shown that the actual precipitation have been within the forecasted range of precipitation in 68 percent of the years and therefore, the developed rules can effectively be used in predicting high, normal, and low ranges of precipitation in winter and spring in the Karoon River basin.

The robust method based on a limited dataset of satellite images and capable to predict snow cover extent over alpine basin in 5–15 days was developed. The set of snow patterns acquired during one melting season is used as a collection to search the pattern–analogue in a past. The basin of Charvak reservoir (Tian–Shan Mountains) is a case study area.

A new hybrid data assimilation procedure is presented that combines a general filtering update with error forecasting at measurement points. The filtering update is based on a predefined, time invariant weighting (gain) function that is used to distribute model errors at measurement points to the entire state of the modelled system. The error forecast models are used to propagate model errors at measurement points in the forecast period. The procedure supports a general linear and non-linear formulation of the error forecast models, and fully automatic parameter estimation techniques have been implemented to estimate the parameters of the models based on the observed model errors prior to the time of forecast. The parameter estimates are automatically updated, which allows the error forecast models to adapt to the prevailing conditions at the time of forecast. The developed procedure is demonstrated in an operational flood forecasting setup.

One of the most often encountered modeling problems is that of infilling missing data, referred to as the data reconstruction problem. It is usually associated with environmental systems where intermediate data gaps exist, i.e. where data or observations before and after the missing intermediate observations are available. Reproducing the missing data inside a time series is a complex problem. Two difficult choices have to be made: 1) which methodology to use; 2) how to measure the performance of different possible realizations of in filled data. This paper introduces a new approach which performs modeling and reconstruction simultaneously. The technique used is the so-called Evolutionary Polynomial Regression (EPR). It is a regression-based algorithm, which uses a hybrid between evolutionary computing and polynomial structures to model environmental processes and/or systems. The EPR approach is tested on two case studies involving groundwater level modeling and water temperature modeling.

Derivative-free Kalman Filter has recently emerged as a very promising optimization tool for any state-space model without the need for derivatives. This paper demonstrates the use of derivative-free Kalman Filter, Unscented and Central Difference Kalman Filters, for state-space Neural Network trained first with back propagation algorithm. The results show them as a promising tool to further improve the prediction accuracy of, for example, a trained Neural Network.

Understanding the predictability of a streamfiow process is important for making decision for flood control and water management. However not much attention has been paid to the predictability study of streamfiow processes so far. Predictability may be broken down into two categories: (1) model predictability and (2) potential predictability. The widely used model performance measure, coefficient of efficiency (CE), can act as a measure of model predictability for stationary series, however, it could be misleading about the model performance when being applied to seasonal processes which have seasonal mean values. Therefore, an adjusted coefficient of efficiency (ACE) is proposed for evaluating model predictability for seasonal time series. The model predictability of two daily river flow processes of the upper and middle Yellow River are studied based on linear ARMA models and measured in terms of ACE.

By combining the multi-criteria scenario analysis model, comprehensive evaluation model and other sectoral models based on the synthetic analysis of the complex system of water-socioeconomy-ecoenvironment, this paper explores a new way to study water resources carrying capacity(WRCC). The assessment of the models indicates that the situation of WRCC of Northwest China will be improved evidently in the future and the goals of water supply and demand balance could be achieved by taking such countermeasures as renovating ecological environment, developing water-saving technology and adjusting industrial structure etc, the A1B2C2 scenario is proved to be the preferred alternative, while reducing the rate of water resources development, improving the efficiency of water use and controlling population growth will be the key factors to improve the water resources carrying capacity.

Long-lead streamflow forecasts are very important for the operation of water resources systems. In many regions, the accuracy of these forecasts results in better water utilization and less drought damages. In this study, a framework for combining two conceptual climatic and hydrologic models for long-lead streamflow forecasting is presented. For this purpose, two models based on the fuzzy inference system and artificial neural networks are used. Because of the vagueness in the predictors of climate and the relationship between predictors and the meteorological variables, the fuzzy inference system is an effective tool in climate forecasting. The capability of Artificial Neural Networks (ANNs) for mapping a set of input-output data with an acceptable error makes them useful tools for hydrologic process modeling. The link between the conceptual Fuzzy Inference System (FIS) and ANN-based models allow us to represent the streamflow variables as a range of possible values that may occur during a specific time period. Forecasted streamflow values are represented as different possible traces. By analyzing these traces, the results of streamflow forecasting could be presented as the mean fuzzy, maximum, and minimum values. The results show the efficiency of applying the proposed approach in long lead streamflow forecasting.

The present paper analyses the large-scale monsoon-induced water level fluctuations in the South China Sea and Indonesian waters. The aim is to obtain more insight in the prescription of open boundary forcing ("tilts") commonly used in limited area coastal models for monsoon-driven flows in the area, for example the update model for Hong Kong waters. TOPEX-POSEIDON altimeter data for 1992-2000 are processed to derive monthly mean sea surface anomaly (SSA) distributions for the region, which are shown to compare well with data obtained from long term observations. Multi-year simulations of 2D barotropic monsoon-driven flow are made with a reduced depth model, driven by 5 day averaged WOCE winds and SSA boundary forcing. The 1998 El Nino year is modelled separately, applying 6 hourly ECMWF winds. A sensitivity analysis to forcing and model settings is performed. The model results are in good agreement with the altimetry derived SSA data and climatological data. The results show that the water level differences prescribed in the limited area model are largely in agreement with the basin-scale modelled SSA for the same area. Recommendations are made for more generic prescription of limited area model forcing.

Flood forecasting schemes may have the most diverse structure depending on catchment size, response or concentration time and the availability of real time input data. The centre of weight of the hydrological forecasting system is often shifted from hydrological tools to the meteorological observation and forecasting systems. At lowland river sections simple flood routing techniques prevail where accuracy of discharge estimation might depend mostly on the accuracy of upstream discharge estimation. In large river basin systems both elements are present. Attempts are made enabling the use of ensemble of short and medium term meteorological forecast results for real-time flood forecasting by coupling meteorological and hydrological modelling tools. The system is designed in three parts covering the upper and central Danube including 160 forecast stations. The available meteorological analysis and forecasting tools are linked to the flood forecasting system. Meteorological forecasts include the 6.5 km resolution ALADIN/HU model - 2-day ahead temperature and precipitation forecast and ECMWF 10-day ahead temperature and quantitative precipitation forecast. The hydrological side of the system includes the data ingestion part producing semi distributed catchment wise input from gridded fields and rainfall-runoff, flood routing modules. The feasibility of the system is demonstrated by the forecast and hindcast for the extreme flood events of August 2002 on River Danube.

Spatial and temporal interpolation of observed groundwater data is an important issue for efficient groundwater management. With the advent of high frequency data collection and monitoring systems the necessity to adopt to new modelling methods which can handle large amount data is pertinent. In the present research, a methodology has been developed for spatial and temporal interpolation and prediction of groundwater level based on machine learning approaches using artificial neural networks and model trees. This methodology has been applied for spatio-temporal prediction of groundwater levels at the city of Delft (The Netherlands) based on the information available at nearby measurement points. The predictive accuracy of the data-driven models was found to be adequate and it is concluded that the use of these methodologies for spatio-temporal interpolation of groundwater level is promising.

Today several new sources of weather data are ready available for use in water management practice. Present precipitation radar information, especially when calibrated, can be of great help in obtaining well performing rainfall-runoff models. Such precipitation data is not ready for use, but has to be cleaned and converted into a hydrological load per subcatchment of a water system.

The availability of the ensemble weather forecast data permits stochastic analysis with hydrological models. Such models present probability distributions of forecasted excess water, water levels and inundation of land per unit of time. This is an important advance in application of precipitation data to day-to-day water management.

This paper presents a methodology in solving a mixed pixel problem when using low spatial resolution remote sensing data (LSRD) for agricultural applications. A numerical experiment is presented to show the strengths and limitations of the proposed approach. First, a simple mixed pixel model is developed and then an un-mixing algorithm was used to derive the sub-pixel information inside the pixel. This information include, the agricultural land uses such as rainfed and irrigated, which the latter could have two or three croppings a year, their dates of sowing and area fractions in the pixel. A soil-water-atmosphere-plant model, SWAP, simulated the dynamics of these land uses and a modified micro-Genetic Algorithm (μGA) was used to solve the mixed-pixel model using generated remote sensing (RS) data. This GA variant was chosen to minimize the curse of computational time in a conventional GA. Feature like 'time-saving-mechanism' was also included in the model. Two approaches were presented to solve the mixed pixel problem: a dynamic linkage and a look-up table (LUT) approach. The purpose of devising LUT is to improve further the computational time and the practical application of the methodology. The results showed that it is possible to derive agricultural sub-pixel information from LSRD. These are important in food security and water management studies at the regional or even at the global scale.

Majority of the popular data assimilation techniques in use today would provide an improved estimate of the system state up to the current time level based on measurements. From a forecasting viewpoint, this corresponds to an updating of the initial conditions of a numerical model. The standard forecasting procedure is then to run the model into the future, driven by predicted boundary and forcing conditions. The problem with this methodology is that the updated initial conditions quickly disappear. Thus, after a certain forecast horizon the model predictions are no better than from an initially uncorrected model. This paper considers a novel approach to wave data assimilation and demonstrates that through the measurement forecast (made using so-called local models), entire model domain can be corrected over extended forecast horizons (i.e. long after updated initial conditions have become disappeared). The distribution of error forecasts over the entire model domain were performed using steady gain vectors derived from the ensemble of spatial error covariances. The improvements in the prediction of wave characteristics are highlighted.

Periodic models are ideal for modeling hydrological time series. But when applying periodic models to daily flow series, there is a problem of how to make the periodic model "parsimonious", because it is infeasible to fit a model for each day of the year. Therefore, an approach to periodic modeling is presented which groups the neighboring days based on cluster analysis, then fits an AR model to each group. A periodic model is fitted to the daily streamflow process of the upper Yellow River, and compared with an ARMA model that is fitted to the entire deseasonalized streamflow series.

This work presents several pre-processing procedures on data used in an Artificial Neural Network model: a) the influence of catchment wetness as evaluated through the use of past runoff values averaged as an input to the ANN. b) rainfall, besides its essentially chaotic behaviour is characterised by its high frequency and abrupt changes in value between consecutive measurements that are not reflected in the runoff generation process. Hence to facilitate the hydrologic simulation, a digital filter is applied to the rainfall data series to provide an improvement to the ANN performance and c) a cross correlation routine is applied to both series, rainfall and runoff in order to determine the time steps required to shift one series in relation to the other in order to maximise the correspondence prior to input to the model.

The NNRRT (Neural Networks Rainfall Runoff Toolkit) has been developed in the Delphi language linking the Matlab Neural Toolbox to a user-friendly interface, allowing a common user to generate simulated series of river flows. Developed with Object-Oriented programming techniques, the NNRRT confirms the potency of the Delphi Language in the hydro-informatics field. Software links are implemented using Microsoft COM (Component Object Model) technology that allows programmers to drive a Windows Application, in this case Matlab, through an external application, in this case a Delphi Project, by simply loading the type library containing all objects of the Windows application. The NNRRT data management has been developed by linking an existing database of historical hydrological series. The NNRRT also considers a Principal Component Analysis module to allow the user to simultaneously manage with a great number of gauge stations. The NNRRT has been tested considering river basins in the Sardinia region (Italy) confirming the easiness in using the toolkit and the effectiveness of obtained results.

It is thought that tremendous expense may be required to install equipments related with remote control system, especially on-line sensors for monitoring organic and nutrient concentrations in the treatment processes. In this research, as a cost-effective tool for replacing expensive on-line sensor, a software sensor was developed to estimate the NOx-N and ammonia concentrations by only using on-line values of ORP, DO and pH at the SBR. The polynomial neural network (PNN) was applied to make software sensor. Developed PNN model could estimate the NOx-N and ammonia profile well. However, the error was increased at the first anoxic period of the first sub-cycle and NOx-N accumulation was occurred at the sub-cycle. Therefore, the rule-base-compensator was developed to solve these problems based on operational knowledge. It was concluded that developed PNN model with the rule-base-compensator could be an alternative for nutrients sensor.

The overall planform changes of braided rivers are the result of the realignment of their channels to the gradually changing local flow direction. Characteristic local planform changes on the channel-scale include migration, island formation, channel cut-off, abandonment, and widening and narrowing. As part of a study to predict such planform changes we have investigated the possibility to use a neural network approach for such a task when input data is limited to geometrical data obtained from satellite images (or, depending on the scale, aerial photographs).

The neural network has been applied to individual pixels of the satellite image and their schematized neighborhood. For each pixel corresponding to a dry (i.e. sand or vegetated) area, the neighborhood is parameterized using quantities such as distance and direction to nearest channel, width of that channel, and amount of water in the neighborhood. Various configurations of neural networks have been trained to predict the state (dry/wet) of the pixel one year later. The fractional output of network can be interpreted as the probability of erosion. This paper gives a short overview of the results obtained using this approach.

Artificial neural networks (ANNs) are used to address a knowledge base system for Houman Navigation Canal Deepening Project, Louisiana, USA. The concerns of this coastal estuarine system are to limit the hurricane environment enhancement and to determine the salinity front for a lock design. Better management decision requires no data gaps in a continuous salinity record. A knowledge base was constructed by including the freshwater inflow, stage, neighboring salinity, tidal forcing, precipitation, and wind stress as the inputs. The complex data recovery systems are conducted mainly due to missing values in the input system as well. The first level data recovery determines the input missing values while the second level data recovery estimates the salinity missing record. Three recovery approaches, namely self-data recovery, neighboring data recovery, and multivariate data recovery, are conducted. The reliabilities are 52 percent, 71 percent, and 87 percent respectively. The optimal training reliability is about 91 percent without any missing values in the input system. Better recovery reliability of salinity should consider the entire physical system as the driving forcing.

The problem of accurately determining river flows from rainfall, evaporation and other factors, occupies an important place in hydrology. The rainfall-runoff process is believed to be highly non-linear, time varying, spatially distributed and not easily described by simple models. Data-driven modeling approaches have been embraced enthusiastically by practitioners in water resources, as they are perceived to overcome some of the difficulties associated with physics based approaches. Such approaches have proved to be an effective and efficient way to model the rainfall-runoff process in situations where enough data on physical characteristics of catchment is not available or when it is essential to predict the flow in the shortest possible time to enable sufficient time for notification and evacuation procedures. Since, in the recent past, there has been an increasing interest to exploit the capability of Artificial Intelligence (AI) techniques in rainfall-runoff modeling, we analyze and apply two Al-based data-driven techniques - the Artificial Neural Network and Genetic Programming for modelling the runoff from a small steep-sloped mountainous catchment in Hong Kong. Although all care was taken in the application of the two models, because of the complex rainfall-runoff dynamics of the catchment and lack of suitable data, both models do not predict well. It is concluded that in such catchments, even well established data-driven models may not perform well, especially if suitable data is also not available.

Wave tranquility studies in and around and approach channel are conventionally carried out using experimental or numerical models. Although numerical schemes yield workable solutions, the underlying assumptions as well as the noticeable differences between the resulting estimations and actual measurements leave scope to employ alternative approaches. The current study is an attempt in that direction and is based on the approach of neural networks. A modular neural network was developed to estimate the wave height distribution in and around the approach channel to the harbor from seaward boundary to the harbor entrance. The trained network was found to perfectly follow the expected trend of wave height attenuation along the harbor approach channel and wave concentration away from the channel and when tested for unseen input yielded satisfactory output of wave heights, which were very close to the numerical model. The network yielded satisfactory results at other site also.

Simulation of rainfall field plays many important roles in water resources studies. River training works and design of flood warning systems are usually confronted by the fact that historical rainfall data are insufficient and sparse in spatial domain for analysis and decision-making purposes. Recent advanced in artificial intelligence and in particular those techniques aimed at converting input to output for highly nonlinear, non-convex and dimensionalized processes such as rainfall field, provide an alternative approach for development of rainfall forecasting model. Artificial Neural Networks (ANNs), which perform a nonlinear mapping between inputs and outputs, are such a technique. Current literatures on ANNs show that selection of network architecture and its efficient training are major obstacles for their daily usage. In this paper, feed-forward type networks will be developed to simulate the rainfall field and an algorithm so called Back Propagation (BP) coupled with Genetic Algorithm (GA) will be used to train the networks. The technique will be implemented to forecast rainfall for a number of lead-time using rainfall hyetograph of recording rain-gauges in the Upper Parramatta Catchment in the western suburbs of Sydney, Australia. Results of the study showed a remarkable improvement in performance indicators when the network was trained via its coupling with GA compared to similar work.

This article presents an approach to forecast the accuracy of surge predictions on the Dutch coast predicted by a hydrodynamic model of the North Sea. Information theory based techniques are used to study the relationship between selected parameters and surge prediction errors at Hook of Holland, which helped to study the dynamics of the surge prediction accuracy and identify parameters that are related to the surge prediction errors at Hook of Holland. Artificial neural network (ANN) models are trained to learn the connection between selected parameters and the surge prediction accuracy in the form of bias and 90% confidence intervals. The model forcings and past surge prediction errors are used as input to the ANN. The study showed that the ANN model forecasts the bias and confidence intervals of the surge predictions with a remarkable accuracy.

Meteorological predictions such as precipitation and temperature are commonly used to improve real-time hydrologic forecasting despite their inherent uncertainty, and their absence in the model calibration stage. In this study, we quantify the effect of meteorological prediction errors on the accuracy of daily spring reservoir inflow forecasts using weather predictions in both the model calibration and testing phases. Different modeling experiments are compared using an operational conceptual model and artificial neural network-based hydrologic models to assess the effects of using daily numerical weather predictions as compared to the use of historical observations. It is found that even with large prediction errors, meteorological forecasts can provide significant improvement of spring flow forecast up to seven-day lead-time particularly for low flows. Spring flow prediction errors associated to the type of hydrological model used are significantly larger than those related to the meteorological predictions particularly for 1-to 4-day ahead forecasts. The experiment results also indicate that multiple model-based forecasting using iterative prediction approach appears the most effective method for an adequate use of weather predictions for streamflow forecasting.

In this paper, a time-series forecasting model is developed. The model is based on the combination of the radial basis function network (RBFN) and the self-organizing map (SOM). Traditionally, the positioning of the radial basis centers is a crucial problem for RBFN. In the proposed model, the SOM is used to construct the two-dimensional feature map from which the number of clusters (i.e. the number of hidden units in the RBFN) can be figured out directly by eyes, and then the radial basis centers can be determined easily. Finally, the proposed model is applied to actual groundwater head data of a well in southern Taiwan to forecast the future groundwater head time series. According to the performance indices, the proposed model has better performance than the seasonal ARIMA model apparently. For time series forecasting, the proposed model is recommended as an alternative to the existing method, because it has a simple structure and can produce reasonable forecasts.

Infiltration is a significant process which controls the fate of water in a catchment. Over the years, many infiltration models have been developed which are either physically based, conceptual or empirical. Literature shows that a model's applicability will always be limited to its context (such as location, availability of data, etc.). Artificial neural networks (ANN) have been recently used with success for a variety of nonlinear hydrologic processes. In this paper, several topologies (Multilayer Perceptron, Radial Basis Functions and Generalized Feedforward Networks) of ANNs were evaluated and used to model infiltration based on data derived from plot scale rainfall simulator experiments conducted in catchments in Cebu, Philippines. In the analysis, the data was organized as static and dynamic time series. Sensitivity analysis determined the relative importance of the input parameters in affecting the output. The network developed was compared with the Green-Ampt (1911), Philip (1957), Horton (1949) and Kostiakov (1932) infiltration models. Results show that neural networks predicts infiltration with higher accuracy over the traditional models.

Climate-change impact studies on hydrologic regime have been relatively rare until recently, mainly because Global Circulation Models, which are widely used to simulate future climate scenarios, do not provide hourly or daily rainfall reliable enough for hydrological modeling. Nevertheless, more reliable rainfall series corresponding to future climate scenarios can be derived from GCM outputs using the so called 'downscaling techniques'. Though these conversion methods do not correct the GCM model inaccuracies, they can provide future daily rainfall scenarios relevant to impact studies on flood regime. This paper presents the results from the investigation of some promising statistical downscaling techniques and compared the results with that of artificial neural network based downscaling using precipitation and temperature data of Chute-des-Passes station located in the Saguenay watershed, in northern Québec, Canada.

To correctly model the flood regime of a catchment, continuous rainfall-runoff simulation at hourly or at least at daily time steps is necessary. Such daily rainfall series at a catchment corresponding to future climate scenarios can be derived from Global Climate Model (GCM) outputs using the so called 'downscaling techniques'. These conversion methods can provide future daily rainfall scenarios relevant to impact studies on flood regime. This paper presents the results of an investigation on the possible impact of climate change on the flow regime of Serpent River in northern Quebec, Canada. Precipitation and temperature data downscaled with three different methods are used as input to the hydrological models and the corresponding differences in the resulting stream flows are compared.

An integration of Geographic Information Systems (GIS) and Unsupervised-Supervised Artificial Neural Networks (ANNs) was applied to quantify the similarity of watershed characteristics, including geometric, meteorologic, and hydrologic parameters. While the Self-Organizing Features (SOFMs), a type of unsupervised ANNs, was used to perform the clustering process, a supervised ANNs such as Multilayer Perceptron, was used to examine the classification accuracy. The purpose of this study was to determine the best fitness "home" for an additional watershed from an existing knowledge base. The result can provide further "transplant" or "synthesis" of hydrological information between watershed systems, such as the streamflow hydrograph and meteorological forcing. An example using a knowledge base with 193 watersheds, which were described by 15 geometric characteristics and 3 hydrological parameters, was demonstrated. The classification and clustering sensitivity due to the number of involved parameters and clustering dimension are discussed.

The occurrence of harmful algal blooms (HAB) in coastal waters is a chronic problem in many parts of the world. The development of real time measurement systems and forecasting models of HAB is important for understanding the algal dynamics and for fisheries and environmental management. In the recent past, data-driven modelling techniques such as artificial neural networks have been increasingly used to model algal bloom dynamics. However the predictability of most of these models is questionable due to the use of interpolated data from low frequency measurements. We present here the first real time modelling and prediction of algal bloom dynamics using artificial neural networks. The model is based on continuous high frequency measurements of water quality and hydro-meteorological parameters at Kat O field monitoring station in Hong Kong. A multi-layer perceptron neural network model trained with back propagation algorithm is employed to model and predict the algal biomass (chlorophyll fluorescence) with different lead times. Daily values of chlorophyll fluorescence, dissolved oxygen, water temperature, solar radiation and wind speed are used as input variables. Multivariate statistical analysis of the input variables and tests with different network scenarios suggest that the use of past values of chlorophyll fluorescence alone is sufficient for algal prediction. Consistent with previous studies, this illustrates the auto-regressive nature of the bloom dynamics in the semi-enclosed coastal waters of Hong Kong. Comparison of predicted and observed chlorophyll fluorescence suggest that the neural network can make reliable real time predictions up to 2-days lead-time.

The reliability of a water-distribution system depends on many interrelated factors such as the water demand variability, the pipes dimension and their maintenance state, the dimension of the reservoirs. Each of these elements affects both the network design and the network management, although the first one represents the most important factor. Because of the uncertainty in evaluating the instantaneous water demands, adequate numerical techniques have to be developed. In this work the Artificial Neural Network technique has been applied in order to define a short-term water demand model; then a stochastic model has been implemented for long term forecasting. The models have been compared in order to define their application fields. The models calibration has been carried out by using field data collected in the Oreto-Stazione network in Palermo (Italy).

The nonlinearity and uncertainty of the flood process are such that estimating or predicting required hydrologic data is often tremendous difficult. Consequently, this study employs a Back-Propagation Network (BPN) as the main structure in flood forecasting to learn and demonstrate the sophisticated nonlinear mapping relationship. However, sophisticated natural systems and highly changeable hydrological environments require that the construction of an artificial neural network (ANN) as a forecasting model should include a risk analysis to reflect the hydrological situation or/and physical meaning of the predicted results. In this paper, a Self Organizing Map (SOM) network with classification ability was applied to the solutions and parameters of BPN model in the learning stage, to classify the network parameter rules and obtain the winning parameters. Hence, hydrologic data intervals can then be forecasted, with the outcomes from the previous stage used as the ranges of the parameters in the recall stage. Overall, this research develops a methodology for providing the decision-maker with more flexibility in forecasting floods.

The ability to simulate the dynamics of flood flow accurately is of crucial importance in flood management operations. Hydrodynamic numerical (HN) models provide a sound physical basis for this purpose and are capable of simulating flow variables at different locations in the river reach. Imprecision in the input data however can affect the performance of HN models, especially for flood forecasting purpose. Artificial neural networks (ANNs) provide a quick and flexible means of flood routing without detailed physical information and are capable of making forecasts. Hence, the combination of both approaches can be used so that the strength of one model can complement the other. This paper evaluates the combined HN-ANN approach for flood flow simulation in the river reach and depiction of the inundation areas at desired locations. A case study from the River Neckar in Germany demonstrates the application.

Several artificial neural work (ANNs) software packages are available and being under use by researchers in different fields of sciences and engineering. These types of software could be used successfully by hydraulic engineers to simulate and predict hydraulic information. In this paper, three ANNs packages are used separately to predict hydraulic data of scour and hydraulic jump phenomena. These packages are NN toolbox of Matlab, Neural Connections and Neuro-Solutions. The performance of the packages are evaluated by comparative prediction of each software and by the measured hydraulic data. The correlation coefficient (R) and the mean relative absolute error (MRE) are used in the evaluation performance of each package. Recommendations and conclusions are presented.

Uncertainty associated with fuzzy membership functions for a water quality management problem is addressed through interval grey numbers. The lower and upper bounds of the membership functions are expressed as interval grey numbers, and the membership functions are modeled as imprecise membership functions. A grey fuzzy optimization model for water quality management of a river system is developed. Application of the optimization model with imprecise membership functions is illustrated with a hypothetical river system.

Ecohydrology is a scientific area that deals with both hydrological and ecological processes and their interactions. Modeling these processes is characterized as a complex task due to associated uncertainties and nonlinearities. Many of the ecological processes are ill-defined or qualitatively described, sharp knowledge is rather exception. Heterogeneity and variability are main characteristics of hydrological parameters. Unfortunately, traditional mathematical models often fail to cope with the mentioned difficulties. According to the European Union (EU) Water Framework Directive (WFD), the actions to be taken for ecosystem improvement should enable reaching good ecological status for all waters in Europe before 2015. This adds the development time and cost of our models as additional difficulties to the ecohydrological modeling process, since the implementation of the EU WFD requires to afford many reliable ecohydrological models in a limited time space. Fuzzy logic offers potential enhancements to ecohydrological modeling as well as many other scientific areas. It facilitates describing complicated, uncertain and non-linear processes in a strict mathematical framework as well as transforming available experts' knowledge into rule-based models with less efforts and low cost. The basic principals of fuzzy ecohydrological models are illustrated in this paper. The advantages of using fuzzy-based models as tools for implementing the EU WFD are discussed. The paper introduces also a demonstration example for a fuzzy fish-habitat model. The model shows that fuzzy logic is a reliable technique for representing many ecohydrological processes and encourages researchers to carry out further similar research.

Performance risk measures are often used in environmental decision making applications as a means to enhance the understanding of a system subject to stochastic inputs and demands. These risk measures are usually probabilistic random variables describing some performance measure of a system. These risk measures are incorporated into the decision making (and optimisation) by specifying required levels of attainment, for example system reliability of .95 or 95%. Typically the specified constraint level is a crisp number even though there is imprecision or uncertainty about what level should be specified and the relative importance of the metrics themselves. Allowing a linguistic rating of the importance and incorporating fuzzy constraints better encapsulates the imprecision associated with the decision making. This paper examines the effect of imposing a fuzzy specification on a probabilistic random variable in an optimisation problem. The example chosen is a Markov decision process with probabilistic constraints. The practical example follows from formulating the operation of a reservoir as a MDP with constraints on reliability, resilience and vulnerability. The specified levels of performance are then given an imprecise specification i.e. a fuzzy representation.

This paper introduces an intelligent river temperature warning system in real time using the Dynamic Evolving Neuro-Fuzzy Inference Systems (DENFIS) to predict the downstream temperature two hours in advance due to the cooling water discharge to a distance of 1km and 5 km downstream from the thermal power station. Results shows that real time predictions using the DENFIS developed can predict the river temperature with a high level of accuracy and can effectively be used as an intelligent warning system to help engineers to take appropriate action in advance on when the potential exists for river temperature to exceed consented limits.

The neuro - fuzzy technique (NFT) called the Adaptive Neuro - Fuzzy Inference System (ANFIS) was employed to forecast daily flood flow of the Yom River Basin in Thailand. Many inputs were tried, using observed daily discharges and mean areal rainfall series from 1990 to 1999 for calibration (training) and from 2000 to 2001 for verification (testing), to obtain the most accurate results. The accuracy of flood forecast is evaluated by using statistical efficiency index (EI), root mean square error (RMSE) and mean absolute error (MAE). The results were found to be very satisfactory. The NFT model results agree very well with the results obtained by using artificial neural network (ANN) model. The results indicate that NFT perform slightly better than ANN in flood forecasting in terms of accuracy and computer time.

Prediction of the plant status is an essential technology to manage the plant stably. In the past decades, several types of modeling methods have been studied in the many types of fields but the single modeling was the main stream. This paper introduces the multi modeling approach to improve the performance of the model. To construct the multi models corresponding to the loading rate, the DO lag-time was employed because the DO lag-time has a strong relationship with the loading rate. Therefore, we could design the estimation system of the loading rate using the lag-time. By the proposed technique, the expensive sensor for the loading rate or ammonia can be replaced by the cheaper DO sensor. In the prediction states, new process data were measured and transferred to the database, and then the loading rate was estimated based upon the lag-time extracted from DO profiles. After estimation of the loading rate, one appropriate model group was selected among prepared groups that consist of the high, medium, and low loading rate group. Finally, the best model among the models of the pre-selected group was determined by the model evaluation. The proposed prediction modeling led the efficient prediction of the process under the various operation conditions. And we also achieved the process diagnosis because of the good performance of the model.

A proposed model by Portilla [1] aimed at estimation of soil erosion by means of a fuzzy expert system is revised. In this framework, this work integrates such a model with a software of Geographical Information System named "Geographic Resources Analysis Support System" (GRASS). Therefore, a new integrated model for soil erosion estimation is presented by using both fuzzy sets and fuzzy logic, and GIS tools. Thus, by means of input of both simple thematic maps and some pertinent parameters, the model capacities allows users to obtain alternate maps for soil erosion as a function of geophysical information such as soil cover, slope, climate and susceptibility for soil erosion. The model in successfully applied for some regions in Colombia (South America).

The Finnish Environmental Administration is responsible for the regulation of the most important lakes in Finland, hence we have developed a tool for river basin management, which includes river basin simulation and decision support systems. In this paper we focus on harnessing a fuzzy logic based model into the river basin management tool to give operational release suggestions using prevailing state and expert knowledge. Input variables for the fuzzy model are net inflow of the previous two weeks, prevailing water level and snow water equivalent. In terms of fuzzy inference it is possible to incorporate experience of an expert into the model. A fuzzy rule-based approach reminds human thinking and explanations of fuzzy decisions can be described to the user. The results show that with the help of twelve years training period and an existing time-dependent target water level zone the fuzzy model imitates operational release decisions of the human decision-maker well. The results indicate that the fuzzy logic based tool is usable in giving support to decisions or in simulating scenarios in a river basin. The application area is Lake Inari, which is a large and important regulated lake in the northern part of Finland and belongs to the transboundary Paatsjoki river basin. The regulation constraints are defined in a joint agreement between Finland, Russia and Norway and the Finnish Environment Institute is responsible for the operative release planning.

Monthly river flow forecasting is an essential part of water resource management. In this paper, different approaches to modelling river flow are compared for the Santa Chiara section of the Tirso Basin in Sardinia (Italy). The results indicated that no significant improvement can be obtained with different neural models for monthly data forecasting, although some pre-preprocessing techniques can improve the forecasting performances as confirmed by the literature.

Support vector machine (SVM) is one of the most elegant data mining engines developed recently. Hydrological time series usually contains a large data set and therefore it is computational difficult to deal with such large data set. It has been shown that SVM provides higher accuracy level than local model does on chaotic time series. Instead of using decomposition method to handle large scale data set, a ridge linear regression applied in a feature space is proposed in this study. The feature space of Gaussian kernel has infinite dimension. With given sample data records, the feature space can be estimated with a finite number of dimensional space effectively. The scheme can be guaranteed to be faster and stable computationally while the accuracy remains close to or better than that of the decomposition method.

Contaminant transport modelling has always been a difficult task due to a large degree of uncertainties and complexities. These complexities reside between hydrological, hydrochemical, hydrobiological processes and the dynamics of the point and diffuse pollutants in the catchment. The analysis and development of a model that is capable of predicting water pollution in a body of water has challenged many researchers and practitioners worldwide. In this paper, a somewhat novel modelling approach for contaminant transport modelling based on a Support Vector Machine (SVM) has been examined. This approach belongs to the class of models that is based on inference from data sets and can be very effective in situations where little or no knowledge about the phenomena is available. The data from the Wairau Stream (New Zealand) is used to identify a dynamic model of contaminants (bacteria, suspended solids and heavy metals) as a function of precipitation and flow rates. For the variables of interest the fifteen minutes measurements were available and utilised in this analysis. From the analysis of results, it was found that the SVM model performed well on all three test datasets.

This paper describes the complexity in using SVM (Support Vector Machine) models in flood forecasting, with focus on identification of a suitable model structure and its relevant parameters for rainfall runoff modelling. Initially developed in the Machine Learning community, SVM has been applied in many fields and has a high success rate in classification tasks such as pattern recognition, OCR, etc. The applications of SVM in regression of time series are relatively new and they are more problematic in comparison with classifications. The study found that exhaustive search of an optimum model structure and its parameter space is prohibitive due to their sheer sizes and unknown characteristics. Some parameters are very sensitive and can increase the CPU load tremendously (hence result in very long computation time). All these make it very difficult to efficiently identify SVM models, which has been carried out by manual operations in all study cases so far. The paper proposed a few approaches for the future activities in the field.

The paper presents a novel method based on Piecewise-Polynomial Truncated Singular Value Decomposition to train Multi-Layer Perceptrons. It is a two-stage technique. As well as in Support Vector Machines framework, second layer weights are estimated after first layer ones in order to obtain a sparse solution (some weights equal to zero) according to function mapping level of accuracy. The technique can work in large-scale problems. It requires the specification of the tolerance ε, as well as for Support Vector Machines, which is strictly related to the Multi-Layer Perceptron's hidden neuron number.

A simple clustering method is proposed to extract most representative data from chaotic hydrological time series. Most clustering techniques need several parameters to be specified. However, this method needs only a single parameter to be specified. Yet it is shown to be very effective. Another advantage of this method is domain specific knowledge is not necessary to tune the parameter. The effectiveness of the proposed method to extract compact set of most representative data is demonstrated on chaotic Lorenz series and two river flow time series. The advantages of the new method over the SCM (Chiu [5]) for data extraction purposes are highlighted. This new clustering method is expected to be more suitable for data extraction for function approximation purposes.

The training time of a learning machine is an important aspect for real time forecasting studies. Identification of an optimal training set can significantly contribute in reducing the training time. This paper suggests a SVM based novel way of extracting a small subset of the training data set which contains all the required information without sacrificing the prediction accuracy. The proposed approach has been applied to the case studies on flood stage data from Dhaka (Bangladesh). The results indicate a reduction of 65-75% of the total training data set. Further, it is noted that the prediction accuracy may sometimes improve with the reduced training set (particularly for higher lead days) indicating that sometimes the presence of redundant training data may even be counter productive.

The settling velocity of sand grains is an important parameter in the study of sedimentary processes in the coastal and estuarine environment. Different settling velocity equations for sand grains have previously been presented in the literature. In this paper a data driven discovery technique, genetic programming, is used to generate new sand grain settling velocity equations from experimental data. The evolutionary processes used resulted in the generation of a number of expressions, the most interesting of which are presented in this paper. These new expressions and the existing equations from the literature are analysed and compared with the experimental data. The results of this analysis are very encouraging. The analysis shows that the relationships generated using genetic programming techniques are good and provide an improvement on the existing equations fitted to the experimental sand grain data. This is the first application of genetic programming techniques to sand grain settling.