جستجوی مقالات مرتبط با کلیدواژه « : مدل ترکیبی » در نشریات گروه « آب و خاک »

Rainfall-runoff modeling and predicting play an essential role in water resource managing, urban planning, reservoir operating, etc. Support vector machine (SVM), as one of the new models of artificial intelligence, has high capability and flexibility in predicting hydrological data. In this research, the idea of rainfall-runoff process modeling using the hybrid model of Particle Swarm Optimization-Wavelet Transform-Support Vector Machine (PSO-WT-SVM) is proposed. There are constant parameters in the SVM algorithm that should be appropriately determined by the user, whereas a wrong choice of these parameters results in a significant reduction in the model performance. In order to solve this problem, the Particle Swarm Optimization (PSO) algorithm is employed to find the best values of SVM constant parameters introducing the PSO-SVM hybrid model. In the next step, applying the Wavelet Transform (WT) pre-processing method on the raw data, this research aims at proposing PSO-WT-SVM hybrid model. Finally, the daily rainfall-runoff time series of the Silakhor plain located in Lorastan province are modeled and forecasted using the SVM single model, PSO-SVM, and PSO-WT-SVM hybrid models. The models' accuracy is assessed using DC and RMSE criteria. The results indicate that PSO-SVM and PSO-WT-SVM hybrid models with DC of 0.72 and 0.89, respectively, supersede the SVM single model with DC of 0.57 in the verification step for Silakhor plain rainfall-runoff time series modeling.

Due to drought and climate change, estimation and prediction of rainfall is quite important in various areas all over the world. In this study, a novel artificial intelligence (AI) technique (WGEP) was developed to model long-term rainfall (67 years period) in Anzali city for the first time. This model was combined using Wavelet Transform (WT) and Gene Expression Programming (GEP) model. Firstly, the most optimized member of wavelet families was chosen. Then, by analyzing the numerical models, the most accurate linking function and fitness function were selected for the GEP model. Next, using the autocorrelation function (ACF), the partial autocorrelation function (PACF) and different lags, 15 WGEP models were introduced. The GEP models were trained, tested and validated in 37, 20- and 10-years periods, respectively. Also, using sensitivity analysis, the superior model and the most effective lags for estimating long-term rainfall were identified. The superior model estimated the target function with high accuracy. For instance, correlation coefficient and scatter index for this model were 0.946 and 0.310, respectively. Additionally, lags 1, 2, 4 and 12 were proposed as the most effective lags for simulating rainfall using hybrid model. Furthermore, results of the superior hybrid model were compared with GEP model that the hybrid model had more accuracy.

Potential evaporation is the result of the combined effects of several meteorological elements, including air temperature, relative humidity (or vapor pressure for saturation), wind speed, sunshine hours and air pressure. The amount of potential evaporation depends on how these variables interact in each climate region. Potential evaporation response of each of these variables depends on the importance that variable plays in the environment. For example, in windy places, the importance of wind speeds in the potential evaporation rate increases relative to places with calm air. By changing each of these meteorological elements, while the rest of the elements react to the given change, the overall effect of these changes and reactions is reflected in the amount of potential evaporation. It is therefore obvious that the potential evaporation response to meteorological variables due to spatial and time variations of these variables is of a complex nature.

For this study, monthly data of air temperature, air pressure at sea level, wind speed, relative humidity and sunshine hours were used as independent variables and monthly data of evaporation pan at Tabriz Synoptic Station as response or dependent variable. In this study, firstly, the nonlinear and linear relationship between meteorological elements and potential evaporation were identified through Generalized Additive Model (GAM), MARSplines Model, and Generalized Linear Model (GLM), respectively. In the next step, by applying the simplex algorithm on the MARSplines model, the evaporation response gradient levels were determined individually for the meteorological variables. Also, to understand the process of pure evaporation response to each of these variables under different climatic conditions, first three weather conditions based on Tabriz Synoptic Station data were defined in three scenarios as S-1, S-2 and S-3. Then, by controlling and maintaining the meteorological variables under these three scenarios and combining the simplex algorithm with the MARSplines Model, the net evaporation reaction curves for the meteorological variables changes were evaluated.

The computational results show that in all combinations, the computational error of the GAM model is less than the GLM model. Also considering the significant variables in each model, the combination of temperature, pressure, wind speed and sunshine are considered as the best subset of the effective variables in the distribution of potential evaporation in both models. On the one hand, relative humidity in these two linear and nonlinear models, in combination with other variables, does not show a significant relationship with potential evaporation. The results of the graphs of Splin smoothing components of the GAM model show that the overall effect of temperature on the evaporation is incremental. But the unit amount of this effect increases with increasing temperature. The individual evaporation reaction against air temperature is similar to its combined reaction. It is thus clear that other meteorological variables do not play a significant role in the influence of air temperature on the evaporation gradient. The overall and hybrid effect of air pressure variations on the amount of evaporation is singular and decreasing. Instead, the individual effect of this variable on evaporation is very intense, decreasing, and partly linear. Therefore, the major influence of air pressure on evaporation in the environment is due to the performance of other variables that interfere with the relationship between these two variables. The evaporation hybrid response to wind velocity was also incremental, although the single and nonlinear evaporation response to wind velocity was not significant, but its tendency was to increase its slope with respect to wind velocity changes. Sunny hours also have a net effect on the amount of evaporation. However, the slope of the solitary effect of this variable, like wind speed, is more than its combined effect. Based on the GLM model results, except for relative humidity, the other variables have a significant linear effect on the potential evaporation. Evaporation response to changes in meteorological variables under S-1, S-2 and S-3 scenarios, while accurately determining the interaction of these variables in plotting absolute evaporation, implicitly implying the synergistic role of these variables in determining absolute evaporation. The lowest distance between the absolute values of evaporation under these three scenarios is related to air temperature, which implies less influence of air temperature than the other variables. That is, the effect of each of the meteorological variables on the amount of evaporation depends to a large extent on the relationship of this variable to other meteorological variables, if such a matter is less weighted for temperature.

The results of this study show that, except for air pressure, which has an increment-reducing effect on evaporation, other variables have only an incremental influence on evaporation and the intensity of this relationship has changed. This process has resulted in a nonlinear component in the relation of independent variables to evaporation. Since hybrid spline smoothing graphs determine evapotranspiration response to each of the predictor variables by eliminating the effect of other variables, therefore, consideration of the composition of these meteorological variables provides more accurate information on evaporation behavior against environmental changes. Through individually fitting evaporation against these meteorological elements, one cannot find how evaporation works against environmental changes. Comparing individual and combined evaporation responses to meteorological variables, while identifying the net effect of each of these variables, explains why evaporation responses within a given unit differ from changing meteorological variables over different times and locations.

Accurate prediction of river daily discharge is a suitable tool for water resources planning and management. In this paper, cross station discharge of the Arkansas River in U.S.A, were examined using Gaussian Process Regression (GPR), Extreme Learning Machine (ELM) and complete ensemble empirical mode decomposition combined models. For this Purpose, in the first step, the daily and monthly discharge was predicted via GPR and ELM models. Then, the discharge time series were broken up by CEEMD method into cages, and these subclasses were introduced into the Gaussian process regression end ELM modeling to simulate discharge. Furthermore, direct correlation (DC), Root Mean Square Error (RMSE), correlation coefficient (R) and Mean Absolute Percentage Error (MAPE) were used to evaluate the efficiency of the models. The results showed that the CEEMD approach improved the performance of the above mentioned models dramatically. For instance, the values of MAPE correspond to GPR hybrid model in forecasting discharge in the first, second and third station with CEEMD pre-processing were reduced by 34, 27and 32 percent, respectively, as compared to those in the GPR model without pre-processing. Also, the effect of each of the sub-series of ensemble empirical mode decomposition model (Residual and IMFs) was studied to improve predictive outcomes. It was observed that the most inefficient subseries in the complete ensemble empirical mode decomposition model is the residual subseries. The CEEMD- ELM model can be used in watershed management and flood control in Iran.

In general, rapid transformation of supercritical flow regime into subcritical flow is accompanied with hydraulic jump. The phenomenon usually occurs at downstream of the hydraulic structures such as ogee spillway. The length of hydraulic jump is one of the most important parameters used to determine the dimension of stilling basins. In current study, a hybrid method for predicting the hydraulic jump length on sloping rough bed was developed. In the other words, the hybrid method (ANFIS-GA) was presented using combination of Adaptive Neuro-Fuzzy Inference Systems (ANFIS) and Genetic Algorithm (GA). In this study, to examine the performance of ANFIS-GA models, the Monte Carlo simulation (MCs) was used. At first, the effective parameters on length of hydraulic jump such as; Froude number in upstream of the hydraulic jump, the ratio of bed roughness, sequent depth ratio, and bed slope, were identified. Next, regarding these parameters, five ANFIS-GA models were defined. Then, the results of the ANFIS-GA models were examined and the superior model was introduced. The superior model predicted the experimental measurements with acceptable accuracy. For example, the Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) were respectively computed 4.520 and 0.781. In addition, the results of modeling revealed that the Froude number at upstream of hydraulic jump was the most effective parameter in modeling the length of hydraulic jump on sloping rough bed using ANFIS-GA method.

In this study, the discharge coefficient of the circular side orifices was predicted using a new hybrid method. Combinations made in this study were divided into two sections: 1) the combination of two algorithms including Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) and providing the PSOGA algorithm 2) using the PSOGA algorithm in order to optimize the Adaptive Neuro Fuzzy Inference Systems (ANFIS) network and providing the ANFIS-PSOGA method. Next, by identifying the parameters affecting on the discharge coefficient of the circular side orifices, 11 different combinations were provided. Then, the sensitivity analysis conducted by ANFIS showed that the Froude number and the ratio of the flow depth to the orifice diameter (Ym/D) were identified as the most effective parameters in modeling the discharge coefficient. Also, the best combination including the Froude number (Fr), the ratio of the main channel width to the side orifice diameter (B/D), the ratio of the orifice crest height to its diameter (W/D) and the ratio of the flow depth to the orifice diameter (Ym/D) for estimating the discharge coefficient was introduced. For this model, the values of Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE) and correlation coefficient (R) were obtained 0.021, 0.020 and 0.871, respectively. Additionally, the performance of the ANFIS-PSOGA method was compared with the ANFIS-PSO and ANFIS methods. The results showed that the ANFIS-PSOGA method for predicting the discharge coefficient was the superior model

The need to simulate rainfall time series at different scales for engineering purposes on the one hand and lack of recording such parameters in small scales because of administrative and economic problems, on the other hand, disaggregation of rainfall time series to the desired scale is an essential topic. In this study, for disaggregating the Tabriz and Sahand rain gauges time series, according to nonlinear characteristics of time scales, wavelet- Least Square Support Vector Machine (WLSSVM) hybrid model is proposed and daily data of four rain gauges and monthly data of six rain gauges from Urmia Lake Basin for ten years were decomposed with wavelet transform and then by using mutual information and correlation coefficient criteria, the subseries were ranked and superior subseries were used as input data of Least Square Support Vector Machine (LSSVM) model for disaggregating the Tabriz and Sahand rain gauges monthly rainfall time series to the daily time series. Results obtained from the WLSSVM disaggregation model were compared with the results of LSSVM and traditional multiple linear regression models. The results of WLSSVM model to LSSVM and multiple linear regression models at validation stage in the optimized case for Tabriz rain gauge were increased 10% and 37.5% and in the optimized case for Sahand rain gauge were increased 24.5% and 46.7% respectively. It was concluded that hybrid WLSSVM model has a higher accuracy than two other methods and can be considered as an accurate disaggregation model to disaggregate the rainfall time series.

The use of clay blanket in reservoirs is one of the main methods of seepage reducing. In this study, with clay blanket modeling in a proposed reservoir by finite element method, 350 dataset was obtained using SEEP/W. Validation of SEEP/W was carried out by comparing seepage results obtained from a laboratory tests. For evaluation of suitable model for predicting seepage values (results of modeling), used from five artificial intelligence techniques comprising: multilayer perceptron neural network (MLP), radial base function (RBF), gene expression programming (GEP), support vector regression (SVR) and a novel hybrid model of the firefly algorithm (FFA) with the multilayer perceptron (MLP-FFA). All the techniques were trained with 70% of available dataset and tested using the remaining 30% dataset. Different combinations of input data that include the ratio of the permeability coefficient of foundation to the permeability coefficient of clay blanket (K_f/K_b ), the ratio of the length of blanket to upstream head (L_1/H), the ratio of thickness of foundation to thickness of blanket (h_f/t), the ratio of length of blanket to thickness of core (L_1/L_2 ) and the ratio of horizontal to vertical permeability coefficient of foundation (K_(f_x )/K_(f_y ) ) were used for evaluation of mentioned methods. The results were evaluated using four performance criteria metrics: root mean square error (RMSE), mean absolute error (MAE), Nash-Sutcliffe efficiency (NS), Willmott’s Index of agreement (WI) and Taylor diagram. The results of study showed that the MLP-FFA method provides better estimation results than the other models and therefore, could be applied an optimized for predictive model of earth fill dam seepage.

During the last decades, important research efforts were conducted to identify and quantify the contribution of different sources delivering suspended sediment to the rivers. This knowledge also proved to be essential to provide estimations of catchment sediment budgets. The type of sources (i.e. soil types, rock types, and land uses) to discriminate depends on the local catchment context. Generally, the targeting of sediment management strategies is a key requirement in developing countries because of the limited resources available. Proper implementation of the soil conservation plans and sediment control programs should be done to inform of the relative importance of contribution the sediment resources as well as identification of crisis centers in the watersheds. During the last decades, this approach has been increasingly applied to identify and ‘trace’ several distinctive characteristics of the source material that can be compared to the same characteristics measured on river suspended sediment samples. Todays, fingerprinting techniques, provide an appropriate method for rapid and low cost information on main sources of sediment. in this study, the mentioned technique in the contribution of sediment resources, identify the critical units using the seven geochemical tracers' properties in the Dare Anar basin of Baghmalek in the Khuzestan province. The focus of this paper is upon quantifying the sources of suspended sediment transported on the Bakhmalek River in order to help guide future surface water sediment reduction efforts for turbidity-impaired streams. The statistical methods were used by the comparison of means and discriminant analysis, to select the optimal combination of tracers and contribution sediment sources. The geochemical tracers tested for their ability to distinguish between sediment sources with the Kruskal–Wallis one-way analysis of variance H test, which is able to test for the independence of more than two variables without presuming either normal or non-normal distributions. Tracers proving significance (p<0.05) between sources were retained. Tracers passing the Kruskal–Wallis H test that were non-conservative (suspended sediment tracer values that were not bracketed by sediment source tracer values) and removed before the performance of the mixing analysis. Tracers passing the first stage of statistical analysis were entered into a stepwise Discriminant Function Analysis (DFA) intended to optimize the number used in the mixing model. This analysis results in the smallest combination of tracers that are capable of correctly distinguishing 100% of the sources through the minimization of Wilks’ Lambda (Collins et al.1998). The analysis was run separately for each drainage basin using IBM SPSS Statistics v. 20.0. From the seven measurement fingerprinting properties, three of them were selected for geology formations and land use by statistics method such as discriminate analysis and compare means tests. Then, a portion of each source determinate by mixed models. Outputs from the discriminant function analysis show the discriminatory power of the final composite of tracers to be 100% successful in the sources classification for Catchment. Finally, among the seven selected tracer included the Lead, Zinc, Copper, Iron, Manganese, Nickel, and Chromium, have identified sediment sources by three elements included the Copper, Manganese, and Iron the amount 54.7, 31 and 14.3 percent respectively. Quaternary and Gachsaran formations, having the highest share in the sedimentary; the aspect ratio was 1.4 and 1.38 respectively. The poor pasture and forest land uses were responsible the highest and the lowest values of the basin sediment with 71.5 and 0.3 percent, respectively. The mitigation of nonpoint-source pollutants, such as sediment, in larger basins is rarely a straightforward procedure due to the number of sources and erosional processes contributing to their concentration in waterways. Therefore, the fingerprinting techniques with the relative efficiency 98.2 percent, having the high accuracy and precision in determinate appropriate method to sediment sources basin and separated of the sediment active units. Low relative error and high model efficiency coefficient confirm the results. Also the field observation is the same as model results. The results were indicating the environmental management strategies must be comprehensive for the study area, that need to reduce surface erosion and hill-slope/channel connectivity and the control gullies development by the commercial cultivation and the range reclamation. Sediment fingerprinting revealed that stream bank erosion in general, and of legacy sediments in particular, from Quaternary and Gachsaran formations to Baghmalek River is at the root of the regional sediment loading problem.

Groundwater considered as the main source of future water supply, irrigation, and food production under impacts of global climate changes phenomena. Main aquifers around the world are under pressure to meet the growing demands of water due to population growth. Management of groundwater reserves in a sustainable manner is a major challenge. A goal of groundwater resource assessment is to provide information on the current status of the resource and provide insights about the future availability of ground water. In recent years, several authors have attempted to assessment groundwater potential using different data-driven and knowledge-driven techniques combined with remote sensing (RS) and geographic information system (GIS). The main objective of this research is identification of effective parameters in groundwater recharge and assessment of groundwater potential using data-driven combined method in najaf-abad aquifer.
The study area lies between (32° 18′ 06″- 32° 50′ 12) latitude and (50° 52′ 46″- 51° 41′ 48″) longitude. It extends over an area of about 966.11 km2. In general, four steps must be implemented to groundwater potential mapping using combined approache. These steps are: (1) prepare groundwater well inventory map and divided into two sets: training and testing. The training data is used to investigate the statistical relationship between well locations and Geo-environmental factors influence on groundwater occurrences. The testing set is used to validate the results. (2) Build the database. In this step layers of groundwater occurrence factors are prepared using different resources such as field survey, and RS. All thematic layers must be converted to raster format to use in further analysis. (3) computation the relationship between training well locations and groundwater occurrence factors using shanoon model and their classes using EBF model are investigated. The groundwater potential map is then computed and classified into four classes using Natural Break scheme (4) the validation of the results and compare the effectiveness of model in prediction groundwater potential zones with indivdal models.
The results of the multicollinearity analysis among 20 Geo-environmental factors influence on groundwater occurrences used in this study showed that the Tolerance and VIF of 15 variables were ≥0.1 and ≤10, respectively. As a result, this parameters are selected for modeling. The computed weights for each factor using Index of entropy model, indicated that the most influencing factors on groundwater occurrence in the study area were distance from fault, LULC and geology. The results of validation of models indicate that The AUC for EBF, index of entropy and EBF-index of entropy models were 0.660, 0.431 and 0.899, respectively implying that the EBF-index of entropy was better than EBF and index of entropy.
Conclution: The main conclusions of this study is that The ensembled approach of EBF-Index of entropy combining with RS and GIS technologies provide a powerful tool for groundwater potential mapping in the study area. The results of this study could be used for efficient managing groundwater resources in the study area. Based on results of ensemble model, The areas covered by very high groundwater potential zones occupy 45.26 % of the total area, indicating that the groundwater potential is high in study area.

Identifying and monitoring of the water resources quality in basin have very special importance for quality management of a dam reservoir. Today, most natural waters arepolluted, so monitoring the distribution of pollutants in surface can control and reduce water pollution and its effects. Having such information is possible only through different analysis and pollution monitoring stations distributed across the study area. Arsenic is considered as one of the most important pollutants due to its high toxicity. Natural water pollution caused by geological resource cannot be eliminated or prevented to be spread simply; therefore, it should be evaluated carefully. Various reports in recent years indicated the presence of arsenic, anomaly with concentration more than the international standard (0.01 mg/L), in the water resources of the Sahand Dam basin which provides agriculture, industry and drinking water demands of the area. Hence, Geology Department of Tabriz University and East Azerbaijan Regional Water Authority have attempted to sampling and chemical analysis of surface water and ground water resources. Groundwater models may use for optimization by one parameter or combination of optimizations, simulation of pollutions and their management. Previous research showed a lack of adequate geostatistical linear models for predicting the total arsenic (III,V) concentration in the study area, so artificial intelligence models such as gene expression programming (GEP) and fuzzy logic (FL) models were used, inspired by nature with ability to estimate the parameters of the natural phenomena with significant accuracy compared to other methods. The number of 60 and 20 data of the hydrochemical parameters that have the highest correlation with arsenic, was used, respectively, in the training and testing level. These parameters Including pH, 〖"SO" 〗_"4" ^"2-" , 〖"NO" 〗_"3" ^"_" , F, Fe and As used as input parameters for Mamdani fuzzy logic (MFL), Larsen fuzzy logic (LSL) and Sugeno fuzzy logic (SFL) to estimate the total arsenic concentration. Fuzzy system has three main level, including: 1) fuzzification of data by defining the membership function; 2) communication of input and output by such as if-then rules; and 3) aggregation of system results and defuzzification by the fuzzy operator such as or/ and/ not. Each of the fuzzy models has its own advantages and uncertainty that can be used of the individual benefits. As the results of three fuzzy models are similar, the genetic expression programming model has been used to produce committee fuzzy logic model (CFLM). This theory is based on that the combination of models results achieves a better overall result. Up to now, several studies have taken using different methods of artificial intelligence that have demonstrated excellence in GEP methods. GEP with genetic algorithms is Like the GA and GP that uses individuals of the population and select them based on fitting and using with one or more genetic operator applied genetic changes on them. The search process is done with random, generated a series of trees that is leading to the production of expression tree. This process continues to the maximum number of replications or specific error function.
Fuzzy model by determining the optimal radius of 0.4, based on the lowest RMSE, were accomplished. The data were divided into 8 categories, and 7 if-then rules were determined. The fuzzy membership functions used for modeling of the arsenic values were Gaussian that was fitted to classified data. The output membership function of Sugeno model was linear, made based on the inputs. FCM clustering method was used in Mamdani and Larsen model. In this model, optimal number of 12 categories were determined based on minimum RMSE equal to 0.11 and 0.12 mg/L, respectively with input and output membership function of Gaussian type. The values of R2 for training level of Mamdani and Larsen model were in order 0.94 and 0.91, respectively. All three fuzzy models had acceptable results, but Mamdani model results were relatively better than the two others. Because of each these models own its performance, for simultaneous use of advantage of all, the committee model was used. All output data of three fuzzy models was used as input data in GEP model and also was selected in such a way that the minimum and maximum of the data be entered on the testing level. Production of the initial population of the program was done by selecting the number 20 chromosome with the head size of 7, 3 number of genes, and 2 constant per gene. The mathematical operator of was selected for the linking function between subtrees. To compare the results in the program, three sets of the function were used as the main operators. F3 function, includes default operators, was selected as the major functions in the program and the best fitted compared to other functions. GEP model by providing the relationship between input and output, and more accurate results in the training and testing levels with R2, 0.97 and 0.92, respectively, was evaluated as the most appropriate model to estimate the arsenic values in the region.
In this study, GEP with practical features and gene expression tree production provided the possibility of evaluating complicated and non-linear models. Also, the genetic programming model provided explicit solutions with high accuracy basis on which can be determined the relationship between input and output variables. With regard to the suitability and similarity of three fuzzy model results, the genetic expression programming was used for the production committee model of results of three single models. Considering the benefits of genetic expression programming, the mentioned model is able to present a committee model with more accuracy than three single fuzzy models. Due to lack of proper accountability of spatial statistical models to estimate the arsenic in the study area, the proposed model can be appropriate in the exact determination.

River's flow prediction and its flood are the important subjects related to the flood projects, hydroelectric power production and water allocation for agricultural, industry and drinking uses. In the present study, wavelet analysis combined with Artificial Neural Network (ANN) was performed to simulate Gamasiyab River flow, located in Nahavand.The analysis was done in daily and monthly scale. For this purpose, the original time series, using wavelet theory, decomposed to multi time sub-signals. Then, these sub-signals were used as input data to the ANN to simulate flow and then the outputs compared with the ANN’s results. In this study, we used the precipitation, temperature, evaporation and flow data of Varayeneh station during the period of 1969 to 2011.Then, 75% of the data was used for training and 25% was considered for simulating data.
In this research, the amount of N and, L were determined 516 and 2 recpectivly. Also, 1 to 4 decomposition degrees were examined to be more precise. The number of neurons in the first layer depends on the degree of wavelet decomposition. The number of input neurons to the network is “m*(j )” in which “j” is the wavelet decomposition degree and “m” is the number of input parameters. For example, for j=1, according to the input parameters which in this study is 4 (precipitation, flow, temperature and evaporation), the number of input neurons is equal to 8. The output layer also has one neuron. The number of the middle layer neurons is variable and is obtained by trial and error. However, in this study, the number of neurons analyzed in the middle layer was varied from 3 to 20. In this case, modeling was done using different training functions and transfer functions for different neurons of the hidden layers. For daily period, input parameters were also decomposed with wavelet function with the levels of 5 to 9 and used as the input data of ANN. Also, the results concluded that 5 architectures had the best performance and it was also observed that the number of neurons of the middle layer was less than 10 in the superior structures. This means that the optimal solution can be reached with a lower number of neurons. Different training functions were examined, but useing all training functions is not suggested because it is time-consuming. Therefore, as proposed in this article, the three models of “Levenberg-Marquardt”, “BFGS Quasi-Newton” and “Bayesian Regularization” due to their better performance are recommended for future studies. After examining different stimulus functions, it is concluded that four types of “Tansig”, “Logsig”, “Satlin” and “Poslin” stimulus functions are suggested to be used for future studies.
Signal decomposition with wavelet increased correlation between observation and simulation data compared to Artificial Neural Network model, so the R2 was increased by 3 percent during the daily period and 24 percent during the monthly period. Two Artificial Neural Network and Wavelet – Artificial Neural Network (WANN) models in the daily period had almost similar performance but Wavelet – Artificial Neural Network model had a better performance in predicting minimum points. Moreover, after examining different structures, is the results determined that 5 architectures, functioned better in the various models and in the daily and monthly periods of the architecture. It means that using the data of evaporation and temperature makes the performance of the models better than the architecture in which these data are not used. Accordingly, the use of temperature and evaporation parameters is suggested in further studies in addition to flow and precipitation parameters.
The results showed that hybrid model of Wavelet-Artificial Neural Network (WANN) outperformed the Artificial Neural Network model. The reason of this preference of the WANN hybrid model is that the separation affects the time series inputs before interring to the network and primary signal decomposed to the various sub-signals. By doing this, we can use the analysis that contain short term and long term effects.
Besides, wavelet function of “Db4” has a better performance than other wavelet functions in each time period. It has the best performance in the daily period in the 5 level and in the monthly period in the second level. Also, the results showed hybrid model outperformed to estimate extent points than artificial neural network model.

Chemical pollution of surface water is one of the serious issues that threaten the quality of water. This would be more important when the surface waters used for human drinking supply. One of the key parameters used to measure water pollution is BOD. Because many variables affect the water quality parameters and a complex nonlinear relationship between them is established conventional methods can not solve the problem of quality management of water resources. For years, the Artificial Intelligence methods were used for prediction of nonlinear time series and a good performance of them has been reported. Recently, the wavelet transform that is a signal processing method, has shown good performance in hydrological modeling and is widely used. Extensive research has been globally provided in use of Artificial Neural Network and Adaptive Neural Fuzzy Inference System models to forecast the BOD. But support vector machine has not yet been extensively studied. For this purpose, in this study the ability of support vector machine to predict the monthly BOD parameter based on the available data, temperature, river flow, DO and BOD was evaluated.
SVM was introduced in 1992 by Vapnik that was a Russian mathematician. This method has been built based on the statistical learning theory. In recent years the use of SVM, is highly taken into consideration. SVM was used in applications such as handwriting recognition, face recognition and has good results. Linear SVM is simplest type of SVM, consists of a hyperplane that dataset of positive and negative is separated with maximum distance. The suitable separator has maximum distance from every one of two dataset. So about this machine that its output groups label (here -1 to ), the aim is to obtain the maximum distance between categories. This is interpreted to have a maximum margin. Wavelet transform is one of methods in the mathematical science that its main idea was given from Fourier transform that was introduced in the nineteenth-century. Overall, concept of wavelet transform for current theory was presented by Morlet and a team under the supervision of Alex Grossman at the Research Center for Theoretical Physics Marcel in France. After the parameters decomposition using wavelet analysis and using principal component analysis (PCA), the main components were determined. These components are then used as input to the support vector machine model to obtain a hybrid model of Wavelet-SVM (WSVM). For this study, a series of monthly of BOD in Karun River in Molasani station and auxiliary variables dissolved oxygen (DO), temperature and monthly river flow in a 13 years period (2002-2014) were used.
To run the SVM model, seven different combinations were evaluated. Combination 6 which was contained of 4 parameters including BOD, dissolved oxygen (DO), temperature and monthly river flow with a time lag have best performance. The best structure had RMSE equal to 0.0338 and the coefficient of determination equal to 0.84. For achieving the results of the WSVM, the wavelet transform and input parameters were decomposed to sub-signal, then this sub-signals were studied with Principal component analysis (PCA) method and important components were entered as inputs to SVM model to obtain the hybrid model WSVM. After numerous run this program in certain modes and compare them with each other, the results was obtained. One of the key points about the choice of the mother wavelet is the time series. So, the patterns of the mother wavelet functions that can better adapt to diagram curved of time series can do the mappings operation and therefore will have better results. In this study, according to different wavelet tests and according to the above note, four types of mother wavelet functions Haar, Db2, Db7 and Sym3 were selected.
Compare the results of the monthly modeling indicate that the use of wavelet transforms can increase the performance about 5%. Different structures and sensitivity analysis showed that the most important parameter which used in this study was parameter BOD, and then flow, DO and temperature were important. This means that the most effective BOD and temperature with minimum impact. Also between different kernels types, RBF kernel showed the best performance. So, combined wavelet with support vector machine is a new idea to predict BOD value in the Karun River.

Estimating correct volume of sediment in alluvial rivers play the most crucial role in water and river engineering, water resources, structures, facilities, water and environmental projects. Using the observed records of sediment load is the most reliable way in estimating it. Since userunfriendly measurement tools and its remarkable fluctuation within the river section, sediment sampling is really difficult and needs high experienced professionals. In addition it is costly and time consuming.ý Because of these limitations sediment frequency of observation around the world especially ýin developing and remote areas have decreased. So, researchers considered some modeling ýapproaches based on diverse terrain attributes and þhydrological variables. Some physically-based models such as the unit stream power (USP) theory of Yang, the SHESED model of Wicks and Bathurst are able to universally predict sediment yield of a watershed but their requirements to lots of detailed information such as geological, hydraulic, and hydrological characteristics of the river basin make them difficult and costly. The same situation there is for process-based models such as the modified universal soil loss equation (MUSLE), introduced by Williams, and its family (USLE and RUSLE). So some models based on many simplifying assumptions and empirical relationships, especially for rainfall and runoff erosive effects were employed which in areas with few data would result in high uncertain results. Using of data-driven ýmodels such as different techniques of artificial intelligence are the þalternative method to the previous models. Artificial neural network (ANN) is one of the most þfamous and strong data-driven technique and ýit has been proved to be practical in modeling non-ýlinear systems or complicated hydrologic processes such as sediment transport. Also conventional sediment rating curve (SRC) has been used widely in all over the world. Although these methods might be similar in some aspects, but each of them has its own characteristics that make them differ from the other. So, authors in this study uses combinations of above tools in a new approach and a hybrid model was employed to utilize the quality of each one in estimating the bed load sediment in Yazdakan gauging station simultaneously.
For the present study Yazdakan bridge gauging site (38.476724o, 44.798362o) located on Qatorchay river was considered. ýQatorchayý river in Aras catchment þwith an area ýof 3471.9 km2 is a river ýin West Azerbaijan province of Iran. It arises in the northwest of Zagros Mountain range in Turkey. The total length of Qatorchayý main streams in Turkey before entering to Iran reaches to 115 kilometers.ý From the border of Iran and Turkey at the height of 290m flows from west to east through lands with steep slopes and passes the length of about 70 kilometers to reach to the Khoy city of west azarbaijan. After leaving Khoy the river turns north and flows through Evaghly valley where joins to Zilber Chay river and extends until ýmeets Agh Chay river at Marakan country. The river with its first name at Farhadi border outpost connects to the Aras ýriver. ý
For current study 133 data pairs of flow discharge and related bed load sediment were considered from Yazdakan Bridge gauging site. The duration of collecting data started from January of 1999 to September of 2009 by West Azarbaijan regional water authority. Monthly data of bed load sediment and discharge related to the time of experiment were recorded. For constructing each model the data set was divided into two groups which the first 25% of date pairs as group one was utilized for testing and the second group including 75% of data pairs employed for both training and validation checks.
In this study discharge flow data was utilized as input to each of ANN sediment rating curve (SRC) techniques separately so as to estimate ýbed load sediment. An idea to optimize the results of ýSRC by ANN, navigate the author to a hybrid model made ýfrom these two called HAS in a way that the error of SRC was predicted using ANN. Finally obtained results demonstrate that the HAS model is in good agreement with the observed ý bed load sediment concentration ývalues; while that depict better results than SRC and ANN models. For example the root ýof mean square error is 213.44 ton/day for HAS model, while it is 262.028 ton/day and ýý238.305 ton/day for ANN and SRC models, respectively. In general, it is illustrated from the ýresults that HAS hybrid model presents better performance to compare whit other models.

Sediment sources fingerprinting is needed as an autonomous tool for erosion prediction، validation of soil erosion models، monitoring of sediment budget and consequently for selecting soil conservation practices and sediment control methods at the catchment scale. Apportioning of eroded-soil into multiple sources using natural tracers is an integrated approach in soil erosion and sediment studies. The objectives of this study، as a first work، are to assess spatial variations of biochemical tracers and theirs validation in discriminating sediment sources under different land uses and water erosions at catchment scale and to apply them as fingerprints to determine relative contributions of sediment sources in Zidasht catchment، Iran. In view of this، 4 enzyme activities as biochemical tracers were measured in 42 different sampling sites from four sediment sources and 14 sediment samples. The results of discriminant function analysis (DFA) provided an optimum composite of two tracers، i. e. urease and dehydrogenase that afforded more than 92% correct assignations in discriminating between the sediment sources in the study area. Sediment source fingerprinting model was used based on optimum composite of two tracers resulting from DFA to explore the contributions of sediment from the four sources. The results showed that the relative contributions from rangeland/surface erosion، crop field/surface erosion، stream bank and dry-land farming/surface erosion sources were 11. 3±5. 3، 8. 1±3. 8، 75±8. 5 and 3. 6±2. 5، respectively. Therefore، we can conclude that fingerprinting using biochemical tracers may help develop sediment fingerprinting models and as a first step facilitate a more complete tool for fingerprinting approach in the future.