Document Type : Original Article
Authors
1
PhD student, Department of Water Engineering Lahijan Branch, Islamic Azad University, Lahijan, Iran
2
Department of Water Engineering, Lahijan branch, Islamic Azad University, Lahijan, Iran.
3
Assistant Professor, Department of Geology, Lahijan Branch, Islamic Azad University, Lahijan, Iran
4
Department of Agronomy and Plant Breeding, Lahijan Branch, Islamic Azad University, Lahijan, Iran.
Abstract
The erosion and transfer of sediment caused by it affects the quality of surface water, and the estimation of sediment and flow quality plays an important role in the management of Water Resources. In the present study, modeling of quality and load indicators of flow sedimentation and providing relationships to predict the amount of flow sedimentation of Khorasan Razavi rivers in 6 stations and statistical period 1986 to 2016 with non-parametric and nonlinear methods of the gated recursive unit Network (GRU) and random forest algorithm (RF) were carried out. By using the bagging method, modeling of flow, sedimentation and total anion and Cation flow in both networks was modeled. MAE, MAPE and RMSE statistical indicators were used to evaluate the performance of the model and the R2 coefficient were used to evaluate the accuracy of data segmentation. Results of the average value of MAE, MAPE and RMSE indicators the water quality data of the stations examined in the GRU network is equal to 286.16, 0.06 and 1685.34 respectively and in the RF network is equal to 317.4, 0.08 and 1954.11 respectively and the sediment load data in the GRU network is equal to 200.95, 0.55 and 1434.44 respectively and in the RF network respectively equal to 52921.84, 0.90 and 1544.29 showed the superiority of the GRU model over the RF model in the modeling of sediment and water quality indicators of the Khorasan Razavi rivers.
Keywords
)