Uncertainty Analysis in Sediment Load Modeling Using ANN and SWAT Model

Sediment load estimation is essential in many water resources projects. In this study, the capability of two different types of model including SWAT as a process-based model and ANNs as a data-driven model in simulating sediment load were evaluated. The issue of uncertainty in the simulated outputs of the two models which stems from different sources was also investigated. Calibration and uncertainty analysis of SWAT were performed using monthly observed discharge and sediment load values and through the application of SUFI-2 procedure. The issue of uncertainty in the ANN model was also accounted for by training a network several times with different initial weights and bias values as well as randomly-selected training and validation sets, each time a network trained. Trying different input variables to find the best and most efficient network structure, it was found that in the forested watershed of Kasilian, adding average daily rainfall or previous values of discharge dose not change the performance of the ANN model significantly. Comparing the results of SWAT and ANN, it was found that SWAT model has a superior performance in estimating high values of sediment load, whereas ANN model estimated low and medium values more accurately. Moreover, prediction interval for the results of ANN was narrower than that of SWAT which suggests that ANN outputs are with less uncertainty.     

自适应变步长BP神经网络在水质评价中的应用

为克服传统的BP网络的不足，采用自适应变步长算法（ABPM）来训练前馈人工神经网络。根据黄河流域的大汶河水系的水质监测的数据，建立了一个对地面水质进行判别的多层前馈网络数学模型。以地面水质七项污染指标为训练样本，对网络进行训练，并将训练好的网络用于水质进行评价，将计算结果与BP网络评价结果、单因子评价结果进行了比较分析。结果表明，ABPM神经网络方法收敛速度较快，预测精度很高，为水质评价提供了一种新方法。     

ANN for Hybrid Energy System Evaluation: Methodology and WSS Case Study

Water Supply Systems (WSS) are large consumers of energy mainly used in pumping stations and treatment plants. Therefore, the improvement of energy efficiency is a major priority for water utilities. The current research work presents a new methodology and a computational algorithm based on renewable energy concepts, hydraulic system behaviour, pressure control and neural networks for the determination of the best hybrid energy configuration to be applied in a typical water supply system. The Artificial Neural Network (ANN) created to determine the best hybrid system uses scenarios with only grid supply, grid combined with hydro turbine, with wind turbine and mutual solutions with hydro and wind turbine. The ANN is trained based on values obtained from a configuration and economical simulator model (CES), as well as from a hydraulic and power simulator model (HPS). The results obtained show this ANN advanced computational model is useful for decision support solutions in the plan of sustainable hybrid energy systems that can be applied in water supply systems or other existent hydro systems allowing the improvement of the global energy efficiency. A real case study is analysed to determine the best sustainable hybrid energy solution in a small WSS of Portugal.     

Optimal Management of Coastal Aquifers Using Linked Simulation Optimization Approach

Saltwater intrusion management models can be used to derive optimal and efficient management strategies for controlling saltwater intrusion in coastal aquifers. To obtain physically meaningful optimal management strategies, the physical processes involved need to be simulated while deriving the management strategies. The flow and transport processes involved in coastal aquifers are difficult to simulate especially when the density-dependent flow and transport processes need to be modeled. Incorporation of this simulation model within an optimization-based management model is very complex and difficult. However, as an alternative, it is possible to link a simulation model externally with an optimization-based management model. The GA-based optimization approach is especially suitable for externally linking the numerical simulation model within the optimization model. Further efficiency in computational procedure can be achieved for such a linked model, if the simulation process can be simplified by approximation, as very large number of iterations between the optimization and simulation model is generally necessary to evolve an optimal management strategy. A possible approach for approximating the simulation model is to use a trained Artificial Neural Network (ANN) as the approximate simulator. Therefore, an ANN model is trained as an approximator of the three dimensional density-dependent flow and transport processes in a coastal aquifer. A linked simulation – optimization model is then developed to link the trained ANN with the GA-based optimization model for solving saltwater management problems. The performance of the developed optimization model is evaluated using an illustrative study area. The evaluation results show the potential applicability of the developed methodology using a GA- and ANN-based linked optimization – simulation model for optimal management of coastal aquifer.     

混合智能算法在引水冲污方案优选中的应用

考虑水质、经济和生态环境影响等因素,建立佛山水道的引水规划优化模型。利用河网水环境数学模型模拟多组引水冲污方案的水质,将输入输出数据作为样本用于人工训练神经网络;将训练好的网络嵌入遗传算法,形成混合智能算法,求解引水规划优化模型。结果表明,混合智能算法能够自动求出不同引水流量下的最优方案,精度较高,无需人工试算,运算速度快,不必对遗传算法与河网模型进行接口处理,具有普遍适用性,为求解耦合复杂模拟模型的优化问题提供了一种理想的工具。     

Artificial Neural Network Models of Watershed Nutrient Loading

This paper illustrates the use of artificial neural networks (ANNs) as predictors of the nutrient load from a watershed. Accurate prediction of pollutant loadings has been recognized as important for determining effective water management strategies. This study compares Haith??s Generalized Watershed Loading Function (GWLF) and Arnold??s Soil and Water Assessment Tool (SWAT) to multilayer artificial neural networks for monthly watershed load modeling. The modeling results indicate that calibrated feed-forward ANN models provide prediction which are always essentially as accurate as those obtained with GWLF and the SWAT, and some times much more accurate. With its flexibility and computation efficiency, the ANN should be a useful tool to obtain a quick simulation assessment of nutrient loading for various management strategies.     

River Flow Forecasting using Recurrent Neural Networks

Forecasting a hydrologic time series has been one of the most complicated tasks owing to the wide range of data, the uncertainties in the parameters influencing the time series and also due to the non availability of adequate data. Recently, Artificial Neural Networks (ANNs) have become quite popular in time series forecasting in various fields. This paper demonstrates the use of ANNs to forecast monthly river flows. Two different networks, namely the feed forward network and the recurrent neural network, have been chosen. The feed forward network is trained using the conventional back propagation algorithm with many improvements and the recurrent neural network is trained using the method of ordered partial derivatives. The selection of architecture and the training procedure for both the networks are presented. The selected ANN models were used to train and forecast the monthly flows of a river in India, with a catchment area of 5189 km² up to the gauging site. The trained networks are used for both single step ahead and multiple step ahead forecasting. A comparative study of both networks indicates that the recurrent neural networks performed better than the feed forward networks. In addition, the size of the architecture and the training time required were less for the recurrent neural networks. The recurrent neural network gave better results for both single step ahead and multiple step ahead forecasting. Hence recurrent neural networks are recommended as a tool for river flow forecasting.     

Intermittent Streamflow Forecasting and Extreme Event Modelling using Wavelet based Artificial Neural Networks

Forecasting of intermittent stream flow is necessary for water resource planning and management at catchment scale. Forecasting of extreme events and events outside the range of training data used for artificial neural network (ANN) model development has been a major bottleneck in their generalization capabilities till date. Despite of several studies using wavelet analysis in water resource modelling, no study has yet been conducted to explore capabilities of hybrid ANN modelling techniques for extreme events outside the training range. In this study a wavelet based ANN model (WANN) is proposed for intermittent streamflow forecasting and extreme event modelling. This study is carried out in a watershed in semi arid middle region of Gujarat, India. 6 years of hydro-climatic data are used in this study. 4 years of data are used for model training, 1 year for cross-validation and remaining 1 year data are used to evaluate the effectiveness of the WANN model. Two different approaches of data arrangement are considered in this study, in one approach testing data are within the range of training dataset, whereas in another approach testing data are outside the training dataset range. Performance of four different training algorithms and different types of wavelet functions are also evaluated for WANN model development. In this study it is found that WANN model performed significantly better than standard ANN models. It is observed in this study that different wavelet functions have different role in modelling complexities of normal and extreme events. WANN model simulated peak values very well and it shows that WANN model has the potential to be applied successfully for intermittent streamflow forecasting even for the data outside the training range and for extreme events.     

Artificial Neural Network Modeling for Groundwater Level Forecasting in a River Island of Eastern India

Forecasting of groundwater levels is very useful for planning integrated management of groundwater and surface water resources in a basin. In the present study, artificial neural network models have been developed for groundwater level forecasting in a river island of tropical humid region, eastern India. ANN modeling was carried out to predict groundwater levels 1 week ahead at 18 sites over the study area. The inputs to the ANN models consisted of weekly rainfall, pan evaporation, river stage, water level in the drain, pumping rate and groundwater level in the previous week, which led to 40 input nodes and 18 output nodes. Three different ANN training algorithms, viz., gradient descent with momentum and adaptive learning rate backpropagation (GDX) algorithm, Levenberg–Marquardt (LM) algorithm and Bayesian regularization (BR) algorithm were employed and their performance was evaluated. As the neural network became very large with 40 input nodes and 18 output nodes, the LM and BR algorithms took too much time to complete a single iteration. Consequently, the study area was divided into three clusters and the performance evaluation of the three ANN training algorithms was done separately for all the clusters. The performance of all the three ANN training algorithms in predicting groundwater levels over the study area was found to be almost equally good. However, the performance of the BR algorithm was found slightly superior to that of the GDX and LM algorithms. The ANN model trained with BR algorithm was further used for predicting groundwater levels 2, 3 and 4 weeks ahead in the tubewells of one cluster using the same inputs. It was found that though the accuracy of predicted groundwater levels generally decreases with an increase in the lead time, the predicted groundwater levels are reasonable for the larger lead times as well.     

Deriving Hedging Rules of Multi-Reservoir System by Online Evolving Neural Networks

This paper presents an online optimization scheme for combined use of Artificial Neural Networks (ANN), hedging policies and harmony search algorithm (HS) in developing optimum operating policies for Tehran water resources system. Past efforts in this area are concentrated on using an offline approach. In that approach, an optimization method is first used to derive a long-term set of optimum reservoir releases. These releases are then used as the target vector for training the ANN model. The online method simultaneously uses the optimization and ANN methods and can adopt objective functions other than minimizing the error indices. Therefore, it requires methods other than the backpropagation for training the ANN model. Hence, under the proposed online approach the application of a heuristic method, such as HS, is inevitable for training the network. This is accomplished by using an optimization-simulation procedure where different objective functions and system constraints could be easily handled. The proposed approach is a novel and efficient method for finding the parameters of hedging policies where earlier methods suffered from high computational costs and the curse of dimensionality. The results show the superiority of the proposed online scheme. Moreover, a surrogate model for the hedging policy is presented, which by adhering to the principle of parsimony is more efficient in large scale systems involving many decision variables.     

Applying Genetic Algorithm and Neural Network to the Conjunctive Use of Surface and Subsurface Water

The conjunctive use of surface and subsurface water is one of the most effective ways to increase water supply reliability with minimal cost and environmental impact. This study presents a novel stepwise optimization model for optimizing the conjunctive use of surface and subsurface water resource management. At each time step, the proposed model decomposes the nonlinear conjunctive use problem into a linear surface water allocation sub-problem and a nonlinear groundwater simulation sub-problem. Instead of using a nonlinear algorithm to solve the entire problem, this decomposition approach integrates a linear algorithm with greater computational efficiency. Specifically, this study proposes a hybrid approach consisting of Genetic Algorithm (GA), Artificial Neural Network (ANN), and Linear Programming (LP) to solve the decomposed two-level problem. The top level uses GA to determine the optimal pumping rates and link the lower level sub-problem, while LP determines the optimal surface water allocation, and ANN performs the groundwater simulation. Because the optimization computation requires many groundwater simulations, the ANN instead of traditional numerical simulation greatly reduces the computational burden. The high computing performance of both LP and ANN significantly increase the computational efficiency of entire model. This study examines four case studies to determine the supply efficiencies under different operation models. Unlike the high interaction between climate conditions and surface water resource, groundwater resources are more stable than the surface water resources for water supply. First, results indicate that adding an groundwater system whose supply productivity is just 8.67 % of the entire water requirement with a surface water supply first (SWSF) policy can significantly decrease the shortage index (SI) from 2.93 to 1.54. Second, the proposed model provides a more efficient conjunctive use policy than the SWSF policy, achieving further decrease from 1.54 to 1.13 or 0.79, depending on the groundwater rule curves. Finally, because of the usage of the hybrid framework, GA, LP, and ANN, the computational efficiency of proposed model is higher than other models with a purebred architecture or traditional groundwater numerical simulations. Therefore, the proposed model can be used to solve complicated large field problems. The proposed model is a valuable tool for conjunctive use operation planning.     

Deriving Reservoir Refill Operating Rules by Using the Proposed DPNS Model

The dynamic programming neural-network simplex (DPNS) model, which is aimed at making some improvements to the dynamic programming neural-network (DPN) model, is proposed and used to derive refill operating rules in reservoir planning and management. The DPNS model consists of three stages. First, the training data set (reservoir optimal sequences of releases) is searched by using the dynamic programming (DP) model to solve the deterministic refill operation problem. Second, with the training data set obtained, the artificial neural network (ANN) model representing the operating rules is trained through back-propagation (BP) algorithm. These two stages construct the standard DPN model. The third stage of DPNS is proposed to refine the operating rules through simulation-based optimization. By choosing maximum the hydropower generation as objective function, a nonlinear programming technique, Simplex method, is used to refine the final output of the DPN model. Both the DPNS and DPN models are used to derive operating rules for the real time refill operation of the Three Gorges Reservoir (TGR) for the year of 2007. It is shown that the DPNS model can improve not only the probability of refill but also the mean hydropower generation when compare with that of the DPN model. It's recommended that the objective function of ANN approach for deriving refill operating rules should maximize the yield or minimize the loss, which can be computed from reservoir simulation during the refill period, rather than to fit the optimal data set as well as possible. And the derivation of optimal or near-optimal operating rules can be carried out effectively and efficiently using the proposed DPNS model.     

Forecasting Surface Water Level Fluctuations of Lake Van by Artificial Neural Networks

Lake Van in eastern Turkey has been subject to water level rise during the last decade and, consequently, the low-lying areas along the shore are inundated, giving problems to local administrators, governmental officials, irrigation activities and to people's property. Therefore, forecasting water levels of the Lake has started to attract the attention of the researchers in the country. An attempt has been made to use artificial neural networks (ANN) for modeling the temporal change water levels of Lake Van. A back-propagation algorithm is used for training. The study indicated that neural networks can successfully model the complex relationship between the rainfall and consecutive water levels. Three different cases were considered with the network trained for different arrangements of input nodes, such as current and antecedent lake levels, rainfall amounts. All of the three models yields relatively close results to each other. The neural network model is simpler and more reliable than the conventional methods such as autoregressive (AR), moving average (MA), and autoregressive moving average with exogenous input (ARMAX) models. It is shown that the relative errors for these two different models, are below 10% which is acceptable for engineering studies. In this study, dynamic changes of the lake level are evaluated. In contrast to classical methods, ANNs do not require strict assumptions such as linearity, normality, homoscadacity etc.     

Drought Forecasting using Markov Chain Model and Artificial Neural Networks

Water resources management is a complex task. It requires accurate prediction of inflow to reservoirs for the optimal management of surface resources, especially in arid and semi-arid regions. It is in particular complicated by droughts. Markov chain models have provided valuable information on drought or moisture conditions. A complementary method, however, is required that can both evaluate the accuracy of the Markov chain models for predicted drought conditions, and forecast the values for ensuing months. To that end, this study draws on Artificial Neural Networks (ANNs) as a data-driven model. The employed ANNs were trained and tested by means of a statistically-based input selection procedure to accurately predict reservoir inflow and consequently drought conditions. Thirty three years’ data of inflow volume on a monthly time resolution were selected to enable calculation of the standardized streamflow index (SSI) for the Markov chain model. Availability of hydro-climatic data from the Doroodzan reservoir in the Fars province, Iran, allowed us to develop a reservoir specific ANN model. Results demonstrated that both models accurately predicted drought conditions, by employing a randomization procedure that facilitated the selection of the required data for the ANN to forecast reservoir inflow close to the observed values over a validation period. The results confirmed that combining the two models improved short-term prediction reliability. This was in contrast to single model applications that resulted into substantial uncertainty. This research emphasized the importance of the correct selection of data or data mining, prior to entering a specific modeling routine.     

Estimate of Suspended Sediment Concentration from Monitored Data of Turbidity and Water Level Using Artificial Neural Networks

Artificial neural networks (ANNs) are promising alternatives for the estimation of suspended sediment concentration (SSC), but they are dependent on the availability data. This study investigates the use of an ANN model for forecasting SSC using turbidity and water level. It is used an original method, idealized to investigate the minimum complexity of the ANN that does not present, in relation to more complex networks, loss of efficiency when applied to other samples, and to perform its training avoiding the overfitting even when data availability is insufficient to use the cross-validation technique. The use of a validation procedure by resampling, the control of overfitting through a previously researched condition of training completion, as well as training repetitions to provide robustness are important aspects of the method. Turbidity and water level data, related to 59 SSC values, collected between June 2013 and October 2015, were used. The development of the proposed ANN was preceded by the training of an ANN, without the use of the new resources, which clearly showed the overfitting occurrence when resources were not used to avoid it, with Nash-Sutcliffe efficiency (NS) equals to 0.995 in the training and NS = 0.788 in the verification. The proposed method generated efficient models (NS = 0.953 for verification), with well distributed errors and with great capacity of generalization for future applications. The final obtained model enabled the SSC calculation, from water level and turbidity data, even when few samples were available for the training and verification procedures.     

Performance Evaluation of Reference Evapotranspiration Estimation Using Climate Based Methods and Artificial Neural Networks

This study is an attempt to find best alternative method to estimate reference evapotranspiration (ET_o) for the Mahanadi reservoir project (MRP) command area located at Raipur (Chhattisgarh) in India, when input climatic parameters are insufficient to apply standard Food and Agriculture Organization (FAO) of the United Nations Penman–Monteith (P–M) method. To identify the best alternative climatic based method that yield results closest to the P–M method, performances of four climate based methods namely Blaney–Criddle, Radiation, Modified Penman and Pan evaporation were compared with the FAO-56 Penman–Monteith method. Performances were evaluated using the statistical indices. The statistical indices used in the analysis were the standard error of estimate (SEE), raw standard error of estimate (RSEE) and the model efficiency. Study was extended to identify the ability of Artificial Neural Networks (ANNs) for estimation of ET_o in comparison to climatic based methods. The networks, using varied input combinations of climatic variables have been trained using the backpropagation with variable learning rate training algorithm. ANN models were performed better than the climatic based methods in all performance indices. The analyses of results of ANN model suggest that the ET_o can be estimated from maximum and minimum temperature using ANN approach in MPR area.     

Integration of Optimal Dynamic Control and Neural Network for Groundwater Quality Management

This study integrates an artificial neural network (ANN) and constrained differential dynamic programming (CDDP) to search for optimal solutions to a nonlinear time-varying groundwater remediation-planning problem. The proposed model (ANN-CDDP) determines optimal dynamic pumping schemes to minimize operating costs and meet water quality requirements. The model uses two embedded ANNs, including groundwater flow and contaminant transport models, as transition functions to predict groundwater levels and contaminant concentrations under time-varying pumping. Results demonstrate that ANN-CDDP is a simplified management model that requires considerably less computation time to solve a fine mesh problem. For example, the ANN-CDDP computing time for a case involving 364 nodes is 1/26.5 that of the conventional optimization model.     

Artificial Neural Network (ANN) Based Modeling for Karstic Groundwater Level Simulation

A relatively new method of addressing different hydrological problems is the use of artificial neural networks (ANN). In groundwater management ANNs are usually used to predict the hydraulic head at a well location. ANNs can prove to be very useful because, unlike numerical groundwater models, they are very easy to implement in karstic regions without the need of explicit knowledge of the exact flow conduit geometry and they avoid the creation of extremely complex models in the rare cases when all the necessary information is available. With hydrological parameters like rainfall and temperature, as well as with hydrogeological parameters like pumping rates from nearby wells as input, the ANN applies a black box approach and yields the simulated hydraulic head. During the calibration process the network is trained using a set of available field data and its performance is evaluated with a different set. Available measured data from Edward??s aquifer in Texas, USA are used in this work to train and evaluate the proposed ANN. The Edwards Aquifer is a unique groundwater system and one of the most prolific artesian aquifers in the world. The present work focuses on simulation of hydraulic head change at an observation well in the area. The adopted ANN is a classic fully connected multilayer perceptron, with two hidden layers. All input parameters are directly or indirectly connected to the aquatic equilibrium and the ANN is treated as a sophisticated analogue to empirical models of the past. A correlation analysis of the measured data is used to determine the time lag between the current day and the day used for input of the measured rainfall levels. After the calibration process the testing data were used in order to check the ability of the ANN to interpolate or extrapolate in other regions, not used in the training procedure. The results show that there is a need for exact knowledge of pumping from each well in karstic aquifers as it is difficult to simulate the sudden drops and rises, which in this case can be more than 6 ft (approx. 2 m). That aside, the ANN is still a useful way to simulate karstic aquifers that are difficult to be simulated by numerical groundwater models.     

Long-term deformation analysis of Shuibuya concrete face rockfill dam based on response surface method and improved genetic algorithm

Due to the size effects of rockfill materials,the settlement difference between numerical simulation and in situ monitoring of rockfill dams is a topic of general concern.The constitutive model parameters obtained from laboratory triaxial tests often underestimate the deformation of high rockfill dams.Therefore,constitutive model parameters obtained by back analysis were used to calculate and predict the long-term deformation of rockfill dams.Instead of using artificial neural networks(ANNs),the response surface method(RSM) was employed to replace the finite element simulation used in the optimization iteration.Only 27 training samples were required for RSM,improving computational efficiency compared with ANN,which required 300 training samples.RSM can be used to describe the relationship between the constitutive model parameters and dam settlements.The inversion results of the Shuibuya concrete face rockfill dam(CFRD) show that the calculated settlements agree with the measured data,indicating the accuracy and efficiency of RSM.     

Deterministic Insight into ANN Model Performance for Storm Runoff Simulation

The artificial neural network (ANN) theory has been widely applied to practical applications in hydrology. Since watershed rainfall–runoff processes are nonlinear and exhibit spatial and temporal variability, the ANN model, which considers watershed nonlinear characteristics, can usually but not always obtain satisfactory simulation results. The training of an ANN network is based completely on the reliability of the available hydrologic records. The objective of this study was to provide deterministic insight into the limitations of storm runoff simulation when using ANN. Hydrologic records of 42 storm events from two watersheds in Taiwan were adopted for analysis. A deterministic runoff model was used to classify the hydrologic records into “usual” and “unusual” storm events. The analytical results show that the ANN model could provide good simulation results for “usual” storm events; however, its performance was poor when it was applied to “unusual” storm events because no consistent hydrologic characteristics could be extracted from the storm event records using ANN. The success of the ANN model in usual storm discharge simulations may be mainly due to the input vectors including the previous observed discharge. Moreover, the number of past periods of rainfall that were set as the input vectors of the ANN model was found to be highly correlated with the watershed time of concentration. It can be used to efficiently determine the ANN network structure instead of using iterative network training.