Prediction of flow and sediment transport in Euphrates - Tigris basin by artificial neural network

In our country, which is in a semi-arid climate zone, it is estimated that drought will be experienced more in the future. In this context, it is very important to determine the future amounts of flow and sediment, which are among the hydrological characteristics of the catchment basins. In this study, the possibilities of estimating the amount of flow and sediment, among the hydrological variables measured in the Fırat-Dicle basin, with 3 different artificial neural networks (ANN) that can be used in many areas today, were investigated and the most appropriate network structure was tried to be determined. The ANN results obtained were compared with the multiple linear regression (MLR) method. For this purpose, different combinations of monthly average flow and sediment data from 20 stations and precipitation, humidity, wind speed, minimum and maximum temperature values from 24 stations were used. Model performances were evaluated by using Correlation coefficient (R), Mean Square Error (MSE), Nash-Sutcliffe Efficiency Coefficient (NSE) which are among different statistical parameters. As a result, it was determined that ANN methods can be used safely in flow and sediment estimation. It was determined that the best network is the combination of RBANN and LM, and the worst network is the combination of MLP and CG.