Modeling of discharge-sediment yield relationship in Middle Euphrates basin using genetic programming

This study investigates the use of Gene-expression programming (GEP) in modeling of suspended sediment yield (SSY) based on monthly averaged river discharge (Q) measurements in Middle Euphrates Basin. Conventional sediment rating curve (SRC) and multiple linear regression (MLR) techniques are often applied to determine the average relationship between SSY and Q. However, these methods are observed to generally underestimate or overestimate the amount of sediment. In the last two decades, new methods, under the name of ?soft computing?, have been proposed as alternative to the conventional methods. Artificial neural networks (ANNs), genetic programming (GP), support vector machines (SVMs) are some of the most widely validated branches of soft computing techniques. Except GP, the other methods generally work as ?black-box? models, which are implicit that can?t be simply used by other investigators. Therefore it is still necessary to develop an explicit model for the discharge?sediment relationship. The aim of this study is to develop explicit models by using GEP. GEP based explicit models, which is predicting the monthly suspended sediment yield, were compared to the rating curves and MLP techniques. The daily discharge and suspended sediment yield data from five stations on Euphrates River in Middle Euphrates Basin were used in validation of the proposed GEP, SRC and MLP models. The results indicate that the proposed GEP formulations perform superior to SRC and MLP models and these formulations are quite practical for use of water engineering society.