Amirkabir University of TechnologyAmirkabir Journal of Civil Engineering2588-297X52220200420Estimation of Roughness Coefficient in Erodible Channels by ANNs and the ANFIS MethodsEstimation of Roughness Coefficient in Erodible Channels by ANNs and the ANFIS Methods495512311110.22060/ceej.2018.14532.5678FAMortezaZanganehfaculty of engineering,Department of Civil Engineering, Golestan University,AbdolmotalebRastegarCivil Engineering Department, Faculty of Engineering, Golestan UniversityJournal Article20180530 Estimating the roughness coefficient of erodible open channels plays an important role in their hydraulic design. This parameter also is important for the development of numerical models. For this reason, several empirical methods have been presented so far to estimate the roughness coefficient, while these methods are not sufficiently accurate. In this paper, the so-called Artificial Neural Networks (ANNs) and Adaptive Network-Based Fuzzy Inference System (ANFIS) methods as soft computing methods are used to estimate the roughness coefficient in erodible open channels. To achieve this, none-dimensional water depth with sediment particle averaged size <strong>( </strong><em>h</em><em>d</em><sub>50 </sub><strong>),</strong>, shear Reynolds numbers (<em>R<sub>*</sub></em>) , Sheilds parameter (θ), and none-dimensional sediment falling velocity with shear velocity (<em>w</em><em>U </em><em><sub>f </sub></em>) in channel obtained by Buckingham dimensional analysis are considered as input variables. Final results show ANFIS (<em>R</em>2 =0<em>.</em>8433) and ANNs (<em>R</em>2 =0<em>.</em>8515) model performance in comparison to empirical methods and regression-based methods like Multilinear regression and multi nonlinear regression methods to estimate the roughness coefficient. Evaluation of the input variables’ effectiveness on the coefficient via a sensitivity analysis versus the variation of error estimation by elimination of variables shows effectiveness of variables like shear Reynolds number and none-dimensional water depth usually ignored in empirical methods. The final results showed that due to complicity of sediment transport mechanism in erodible channels, models developed here can be a suitable alternative to estimate roughness coefficient. Estimating the roughness coefficient of erodible open channels plays an important role in their hydraulic design. This parameter also is important for the development of numerical models. For this reason, several empirical methods have been presented so far to estimate the roughness coefficient, while these methods are not sufficiently accurate. In this paper, the so-called Artificial Neural Networks (ANNs) and Adaptive Network-Based Fuzzy Inference System (ANFIS) methods as soft computing methods are used to estimate the roughness coefficient in erodible open channels. To achieve this, none-dimensional water depth with sediment particle averaged size <strong>( </strong><em>h</em><em>d</em><sub>50 </sub><strong>),</strong>, shear Reynolds numbers (<em>R<sub>*</sub></em>) , Sheilds parameter (θ), and none-dimensional sediment falling velocity with shear velocity (<em>w</em><em>U </em><em><sub>f </sub></em>) in channel obtained by Buckingham dimensional analysis are considered as input variables. Final results show ANFIS (<em>R</em>2 =0<em>.</em>8433) and ANNs (<em>R</em>2 =0<em>.</em>8515) model performance in comparison to empirical methods and regression-based methods like Multilinear regression and multi nonlinear regression methods to estimate the roughness coefficient. Evaluation of the input variables’ effectiveness on the coefficient via a sensitivity analysis versus the variation of error estimation by elimination of variables shows effectiveness of variables like shear Reynolds number and none-dimensional water depth usually ignored in empirical methods. The final results showed that due to complicity of sediment transport mechanism in erodible channels, models developed here can be a suitable alternative to estimate roughness coefficient.https://ceej.aut.ac.ir/article_3111_f79bead228da630b402bd6d99e0ec3b1.pdf