Application of neural network model for the prediction of chromium concentration in phytoremediated contaminated soils

The assessment of chromium concentrations in plants requires the quantification of a large number of soil factors that affect their potential availability and subsequent toxicity and a mathematical model that predicts their relative concentrations. Many soil characteristics can change the availability of chromium (Cr) to plants in soils. However, accurate, rapid and simple predictive models of metal concentrations are still lacking in soil and plant analysis. In the present work a novel artificial neural network (ANN) model was developed as an alternative rapid and accurate tool for the prediction of Cr concentration in dwarf bean leaves grown in the laboratory on phytoremediated contaminated soils treated with different amendments. First, sixteen (4 × 4) soil samples were harvested from a phytoremediated contaminated site located in south-western France. Second, a series of measurements were performed on the soil samples. The inputs are the soil amendment, the soil pH, the soil electrical conductivity and the dissolved organic carbon of the soil, and the output is the concentration of Cr in the dwarf bean leaves. Third, an ANN model was developed and its performance was evaluated using a test data set and then applied to predict the exposition of the bean leaves to the Cr concentration versus the soil inputs. The performance of the ANN method was compared with the traditional multi linear regressions method using the training and test data sets. The results of this study show that the ANN model trained on experimental measurements can be successfully applied to the rapid prediction of plant exposition to Cr.

► Development of a neural network for the prediction of chromium toxicity in soil
► Inputs: soil treatment, pH, electrical conductivity, and dissolved organic carbon
► Output: chromium concentration in the bean leaves
► Investigation of chromium toxicity versus inputs for four different treatments
► The neural network model leads to an accurate prediction of the soil outputs