Estimation of Yield Sediment Using Artificial Neural Network at Basin Scale

Forecasting of sediment discharge in river on regional scale is very important process for water resources assignment development and managements. The sediment yield is usually calculated from the direct measurement of sediment concentration of river or from sediment transport equations with hydrological stations in basin outlet point. Apparatus direct measurement is very costly and cannot be perform for all river measurement stations also in some basins are station miss. However, total sediment transport equations do not correspond with each other and require many detailed data on the flow and sediment characteristics. ANN model has the capability of identifying complex nonlinear relationships between inputs and output data sets. The rainfall-runoff relationship is one of the most complex hydrologic phenomena due to the tremendous spatial and temporal variability of the watershed characteristics and unpredictable rainfall pattern. The capability of ANN model for mapping the nonlinearity makes it a suitable tool for assessing the hydrological impacts of land modification and determine rate produce yield sediment in basin down stream. In general, the forecasting performance of ANN techniques is found to be advanced to the other predictable statistical and stochastic methods in terms of the selected performance standard. The ANN is well known as a flexible mathematical structure and has the ability to generalize patterns in imprecise or noisy and ambiguous input and output data sets. The study area is sorkhab River in upstream DEZ basin, IRAN country..This paper presents the proposed ANN model and Multiple Regression (MR) for prediction of total sediment at basin scale. Results show that estimated rate of sediment yield by Artificial neural networks is much better fits with the observed data in comparison to MR model. So that the differences between the estimated and the measured amount of sediment yield were respectively for Artificial neural networks model