Predicting Lead Concentration of Soil using Readily Available Properties Based on Artificial Neural Network Model

Increased generation of pollutants such as heavy metals is one of the serious and developing environmental issues threatening human society. Heavy metal pollution not only affects the physical and chemical properties of the soil but also it is dangerous to human health through entering into the food chain and finding its way into the groundwater. The present study was conducted to predict soil lead concentration, as one of the most important heavy metals, using readily available soil properties based on artificial neural network model. For this purpose, 63 soil samples were collected from 60-cm depth of the land surrounding Kashafrud River located in Mashhad City. Measured parameters included pH, electrical conductivity, particle size distribution, organic carbon, and Pb content in soil. The multilayer perceptron (MLP) as an artificial neural network model was used to predict the Pb concentration in soil. The performance of this model was assessed by the coefficient of determination (R2), mean absolute error (MAE), and also root mean square error (RMSE). The results showed that artificial neural network model is a suitable method to determine Pb concentration in soil rather than the direct laboratory measurement, which is an expensive and time-consuming method.