Seyed Ahmad Hosseini; Ahmad Tabatabaei
Abstract
Introduction
Simulating suspended sediment in hydrological systems has always been challenging due to inherent complexities and uncertainties. This issue has led to the use of intelligent models such as Artificial Neural Networks (ANNs) as a suitable approach for predicting suspended sediment load. ...
Read More
Introduction
Simulating suspended sediment in hydrological systems has always been challenging due to inherent complexities and uncertainties. This issue has led to the use of intelligent models such as Artificial Neural Networks (ANNs) as a suitable approach for predicting suspended sediment load. Therefore, the use of intelligent models like ANNs has expanded in this field. However, determining the optimal network structure (including the number of neurons, layers, weights, and biases) is usually done through trial and error, which is both time-consuming and inefficient. In this study, a multilayer perceptron neural network was used to simulate the daily suspended sediment load in the Qarasu Sarab watershed (Quri Chay and Hir Chai rivers) located in Ardabil province, Iran.
Materials and methods
In this research, an Artificial Neural Network (ANN) of the Multilayer Perceptron (MLP) type was utilized to simulate the daily suspended sediment load in the Sarab Qareh Su watershed (including the Quri Chay and Hir Chay rivers) in Ardabil province. The neural network models were trained not only whit the conventional backpropagation algorithm but also using the Particle Swarm Optimization (PSO) algorithm to optimize the weights and biases of the neurons. Furthermore, to increase the models' generalization capability, a Self-Organizing Map (SOM) clustering was employed. In addition to the backpropagation algorithm, the Particle Swarm Optimization (PSO) algorithm was also employed to optimize the network weights and biases. Furthermore, to enhance the model's generalization power, SOM clustering was used. The use of evolutionary algorithms such as PSO in training neural networks is an effective approach to improve the accuracy of intelligent models, especially in simulating river suspended sediment and applications related to water resources and watershed management structures.
Results and discussion
Using SOM clustering and the Davies-Bouldin index, the optimal number of clusters was determined as 12 for Koozeh Toupraqi station and 15 for Hir Chai station. Statistical analysis and the Kolmogorov-Smirnov (KS) test showed that data distributions across training, validation, and testing sets were similar, which enhances the generalization capability of the models. Training the neural network models with PSO yielded better performance and lower prediction errors compared to backpropagation. The ANN-PSO-Sig and ANN-PSO-Tan models achieved the best results at Koozeh Toupraqi and Hir Chai stations, respectively. Bias analysis further confirmed that PSO-trained models had lower errors in total sediment load estimation. Overall, results showed that PSO-based training outperforms pure backpropagation training. At Koozeh Toupraqi station, the hybrid ANN-PSO model with sigmoid activation function (ANN-PSO-Sig), and at Hir-Hirchai Topraghi station, the hybrid model with hyperbolic tangent activation function (ANN-PSO-Tan) were selected as optimal models, showing biases of +5.25% and -19.2% and RMSE values of 86.28 and 89.2 tons per day, respectively. These findings demonstrate the improvement in suspended sediment load prediction accuracy by using PSO in neural network training.
Conclusion
The use of the PSO metaheuristic algorithm in training neural network models improved their performance in simulating suspended sediment load. This method outperformed gradient-based error algorithms and provided more accurate weight optimization. The improved bias accuracy in PSO-trained models is crucial for designing hydraulic structures and water resource management. Furthermore, SOM clustering helped select homogeneous and representative datasets for model training, enhancing model generalizability. Overall, considering the complexities and uncertainties in hydrological systems, employing intelligent models combined with evolutionary optimization algorithms like PSO is an effective approach for simulating and monitoring suspended sediment loads. The obtained results can be applied in planning and implementing watershed engineering measures and water resource management.
Mahmoudreza Tabatabaei; Amin Salehpour Jam; Jamal Mosaffaie
Abstract
IntroductionThe cycle of soil erosion (including removal, transport and deposition) that controls the sedimentation of watersheds, includes a set of complex and highly nonlinear processes. On the other hand, the factors influencing sedimentation in watersheds are very diverse, and according to the specific ...
Read More
IntroductionThe cycle of soil erosion (including removal, transport and deposition) that controls the sedimentation of watersheds, includes a set of complex and highly nonlinear processes. On the other hand, the factors influencing sedimentation in watersheds are very diverse, and according to the specific conditions of climate, soil, vegetation, geology, topography, etc., in each basin, the weight and role of each of the mentioned factors in sediment production is very different. Accurately determining and measuring these factors and making mathematical relationships between them are often difficult, expensive, time-consuming and error-prone, and this is the case with the use of models based on computational intelligence and the use of a limited number of basin dynamic variables, it is possible to simulate the behavior of the watershed in sediment production. Regardless of the type of intelligent models, in most of the conducted research (especially in internal research), the simulation of suspended sediment is mainly based on the discharge variable and the role of variables such as precipitation (especially precipitation obtained from satellite images), which are effective in the sedimentation of basins, have received less attention. In addition to precipitation, the skewness of sediment measurement data is also one of the issues that lack of recognition and attention will reduce the efficiency of estimator models. In the present study, the role of variable daily rainfall (taken from CHIRPS satellite) in the simulation of suspended sediment of Qarachai River has been investigated. Materials and methodsMulti-layer perceptron artificial neural network was used in order to simulate the daily suspended sediment concentration of Qarachai River (at the Ramian hydrometer station in Golestan province). In this regard, the variables of discharge and previous discarge (in instantaneous and daily scales) as well as the average daily and previous rainfall of the basin (taken from CHIRPS satellite) for a statistical period of 37 years (1980-2017) as variables model input was used. In order to increase the generalization power of the models, self-organized mapping neural network (for data clustering) and gamma test was used to find the best combination of input variables. In order to improve the efficiency of network training, a variety of activation and loss functions as well as the overfitting prevention algorithm were used. In order to investigate the effect of using activation and loss functions in suspended sediment estimation, different scenarios were considered, which led to the construction of 9 models. After that, using validation indicators, the effectiveness of the models in simulating suspended sediment was investigated and compared, and then the best model was selected. Results and discussionThe results obtained from the present research showed that among the different models, the neural network model with Huber's activation function and ReLU loss function, having the average absolute value of the error equal to 368 mg/l, the root mean square error equal to 597 mg per liter, the Nash-Sutcliffe coefficient of 0.87 and the percent bias -2.2% were selected as the best model. The results also showed that the use of the rainfall variable (as one of the important factors in causing erosion and sediment transfer in the basin) has improved the efficiency of the models, therefore, considering the ease of using CHIRPS satellite rainfall data, it is suggested in order to simulate the suspended sediment of rivers, this data is also used along with other predictive variables. ConclusionIn the simulation of suspended sediment, discharge variable is often used as the only predicting variable of suspended sediment, while in basins with rainy, or rainy-snow regimes, the role of precipitation in the production of surface runoff and soil erosion is very important and plays an important role in the production and transport of sediment in the basin. In this regard, although the use of rainfall data obtained from ground rain gauge stations has played an effective role in increasing the efficiency of data-based models in estimating suspended sediment, however, the preparation of hundreds of spatial distribution layers of daily rainfall from the data point data of ground stations, the use of this variable in the simulation of the suspended sediment of the basin has been faced with many problems (such as the lack or inappropriateness of the spatial distribution of rain gauge stations, statistical deficiencies, the use of inappropriate interpolation methods and time-consuming calculations). Therefore, in practice, the variable of river flow is often used as a predictor of sediment, and precipitation is used less often. One of the solutions to the problem mentioned in the present study is the use of CHIRPS satellite data, which was investigated for the first time in this study. These data, available since 1981, can easily be used to simulate suspended sediment or other applications related to watersheds. Another important point that needs to be taken into account in the simulation of suspended sediment is the presence of high skewness in sediment measurement data (both suspended sediment and flow rate), which lack of attention in the process of training (or recalibration) and testing the models leads to It will lead to the construction of weak models in terms of efficiency and the existence of uncertainty in the accuracy of their results. In this regard, it is necessary to use logarithmic transformations or suitable functions of activation and loss in the training process, which in this research, two functions, ReLU and Huber, were proposed respectively. Another important point is to pay attention to the generalization power of data-based models, which is largely dependent on the data used in their calibration or training process. These data should be selected in such a way that while they are representative of the data in the entire statistical period, they are similar and have the same distribution with other data sets (such as cross-validation or test sets). According to the results obtained from the present research and in order to increase the efficiency of artificial neural network models in estimating the suspended sediment of watershed hydrometric stations, it is suggested to use the experiences obtained in this research in other sediment measuring stations of the country.
Mahmoudreza Tabatabaei; Amin Salehpour Jam
Abstract
Relationships between river water quality parameters and physical, geochemical and biological processes carried between basin resources (soil, vegetation, geology, land use, etc.), meteorological variables (temperature, precipitation, snowmelt, etc.), River hydrological variables (flow discharge), as ...
Read More
Relationships between river water quality parameters and physical, geochemical and biological processes carried between basin resources (soil, vegetation, geology, land use, etc.), meteorological variables (temperature, precipitation, snowmelt, etc.), River hydrological variables (flow discharge), as well as human interventions are often very complex, nonlinear and non–deterministic in a way that makes their complete understanding impossible. In this situation, the use of computational intelligence (such as artificial neural networks) is a useful tool in simulating and estimating river water quality variables such as suspended sediment load. In the present study, by combining open source GIS libraries and neural network models (with and without supervisor), an intelligent GIS system has been designed and coded that can estimate daily suspended sediment load under univariate or multivariate conditions. The results of applying this system to Mazaljan River Watershed at Razin hydrometric station showed that this system is able to simulate suspended sediment load with proper performance and validation (with root mean square error of 1033 tonday-1, mean absolute error of 455 tonday-1 and Nash-Sutcliffe efficiency of 0.89 for the test data set). In general, this system can be used as a national infrastructure in the simulation and management of suspended sediment in all hydrometric stations in the country by relevant organizations.
Adele Alijanpour Shalmani; Alireza Vaezi; Mahmoudreza Tabatabaei
Abstract
Analysis of suspended sediment load data in rivers is the basis for understanding the trend of erosion and sediment in the management and planning of soil and water resources. Due to lack of access to daily suspended sediment loading data with direct measurement, it is important to use methods for modeling ...
Read More
Analysis of suspended sediment load data in rivers is the basis for understanding the trend of erosion and sediment in the management and planning of soil and water resources. Due to lack of access to daily suspended sediment loading data with direct measurement, it is important to use methods for modeling and estimating it in watersheds. One of the best methods used in this field is the use of artificial neural networks. To evaluate daily suspended sediment load, Sira hydrometric station was studied in Karaj River watershed. The number of data used in this study included 624 information records of 31 years (1981–2011) statistical period .Input data to the artificial neural network models included instantaneous flow discharge, average daily flow discharge, average daily flow discharge with a delay of three days, average daily precipitation and average daily precipitation with a delay of three days. Output data to models was daily suspended sediment load. In this research, gamma test and genetic algorithm were used to obtain optimal variables and best combination of variables for entering the model. Then, these combinations with some combination of test and error variables were entered to artificial neural network models. The self-organizing map neural network was used for data clustering and all data were divided into three homogeneous groups: 70 percentage training data, 15 percentage validation data and 15 percentage test data. Then, the combination of variables entered to neural network models with activation functions log sigmoid and tangent sigmoid. The results showed that the neural networks using the optimal variable combinations in comparison with manual combinations have a more accurate estimate for suspended sediment load. In all combinations of inputs to neural network models, a model with tangent sigmoid activation function, with input variables combination including, instantaneous flow discharge (Q), average daily flow discharge (Qi), average daily flow discharge for two day ago (Qi-2), average daily flow discharge for three day ago (Qi-3), average daily precipitation (Pi), average daily precipitation for two day ago (Pi-2) and average daily precipitation for three day ago (Pi-3), was the best model for estimating daily suspended sediment load. This model has the lowest of error (MAE=500.05 (ton/day), RMSE=1995.33(ton/day) and Erel=7%), the highest accuracy (R2=0.96), the highest performance model (NSE=0.96) and has the lowest general standard deviation (GSD=0.97) compared to other models. Also, this model is the best combination with the most influential input variables derived from gamma test and genetic algorithm for estimating SSL.
