Effects of Water Level and Flow Discharge Fluctuations on Suspended Sediment Load Using Artificial Neural Networks (Case study: Ahar Chay River)

Evaluation of suspended sediments and parameters affecting them is of great importance in river engineering. In this research the effects of water level and flow discharge fluctuations on suspended sediment load were studied using Artificial neural networks (ANNs). Ahar chay river basin, located in north of Iran, with area totaling about 2400 km2 as a sub-basin of Aras river was chosen. Data from Tazeh-Kand, Orang, Casein, Oshdologh and Bermice (upstream of Sattarkhan dam) stations were employed in ANNs method and the suspended sediment were predicted. The results showed that, suspended load forecasted by water level data had low accuracy than that forecasted by the flow discharge. The maximum and minimum coefficients of correlation for water level data were 0.69 and 0.08 in Orang and Oshdologh stations, respectively. The corresponding values for flow discharge were 0.84 in Oshdologh and 0.7 in Bermice. The main reason for the low coefficient of fittness in some stations probably were due to shortage of data, lack of temporal sequence and inaccuracy of water level fluctuations compared to the flow measurements. It appeared that, in this basin, in moving from high to low land with increasing discharge and sediment rates, the results of ANN became more reliable. Water level fluctuations did not show this trend.