Prediction of Longitudinal Dispersion Coefficient in Natural Channels Using Soft Computing Techniques

Accurate estimate of longitudinal dispersion coefficient is essential in many hydraulic and environmental problems such as intake designs, modeling flow in esturies and risk assessment of injection of hazardous pollutants into river flows. Recent research works show that in the absence of knowledge about explicit relationships concerning longitudinal dispersion coefficient and its influencing parameters, data driven techniques can be used to predict it with reasonable degree of accuracy. In this paper, the usefulness of Support Vector Machines (SVM) and Genetic Programming (GP) are examined for predicting longitudinal dispersion coefficient in natural channels. The hydraulic variables such as flow depth (H), flow velocity (U) and shear velocity (u(*)) along with the width of channel (B) are used as input variables to predict longitudinal dispersion coefficient (K-x). The performance evaluation based on multiple error criteric confirm that GP shows remarkably good performance in capturing non-linear relationship between the predictors and predictant in the estimation of longitudinal dispersion coefficient when compared with empirical approaches, the traditional Artificial Neural Networks (ANN) and S V M. Bence GP can be used as an efficient computational paradigm in the prediction of longitudinal dispersion coefficient in natural channels.