Suspended Sediment Load Estimation in Lower Sakarya River By Using Artificial Neural Networks, Fuzzy Logic and Neuro-Fuzzy Models

In water resources management and projects, correct estimation of suspended sediment load being carried by a river is very important in the determination of economical lifetime of facilities built on rivers. The suspended sediment load of the river is generally determined from direct measurement of the sediment load or from sediment transport equations. Although direct measurement is the most reliable method, it is very expensive and can not be conducted for as many streams as the measurement of water discharge. Also during floods direct measurement of suspended sediment load can not be measured. On the other hand, most of the sediment transport equations require detailed information on the flow and sediment characteristics and also do not agree with each other and it makes difficult to choose which equation is the best, the methods in the literature for sediment load estimation are very complex and time consuming. In this study because of the complexity of the phenomena soft computing methods which are the powerful tool for input-output mapping are used, these are artificial neural networks (ANNs), fuzzy logic (FL) that is Mamdani Fuzzy interference system (FIS-Mamdani) and Sugeno fuzzy interference system (FIS- Sugeno), adaptive neuro-fuzzy inference system (ANFIS) approaches used to estimate suspended sediment load values. This application is modeled to predict suspended sediment load in Lower Sakarya River in Sakarya, Turkey. The results show that adaptive neuro -fuzzy inference system (ANFIS) technique is found to b e significantly superior to others and to ease the model building process.