Contributions of Non-Residual (Fe Oxides, Mn Oxides and Organic Materials) and Residuals in Surficial Sediments to Atrazine Adsorption Using Artificial Neural Network Model

Abstract

A three-layer artificial neural network (ANN) model was developed to predict the contributions of non-residual and residual components in surficial sediments (SSs) on atrazine (AT) adsorption based on 32 experimental sets obtained in a laboratory batch study, in which the inputs were selected as contents of Fe oxides, Mn oxides, organic materials (OMs), residual component and the initial concentrations of AT, the output was set as the amount of AT adsorption onto SSs. The performance of the BP ANN model was assessed through the mean square error (MSE), relative deviation (RD), coefficient of determination (r <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ) (square of the correlation coefficient), and Nash-Sutcliffe Simulation efficiency coefficient (NSC) estimated from the experimental and predicted values of the dependent variables. The results indicated that the model could describe AT adsorption onto different contents of SSs components well. The influence of Fe oxides, Mn oxides and OMs on the adsorption of AT could be also predicted via the established BP ANN model. The results show that Mn oxides restrain the AT adsorption and play the most important role in the adsorption process, Fe oxides and OMs in SSs facilitate the sorption of AT.