Application of an inverse neural network model for the identification of optimal amendment to reduce copper toxicity in phytoremediated contaminated soils

• Reduction of copper toxicity in phytoremediated contaminated soils
• An inverse neural network model to predict optimal soils treatment
• Inputs: three soil parameters and target copper level
• Output: the best soil treatment to reduce copper toxicity below the target value
• Investigation of optimal treatment versus soil inputs

Artificial neural network ANN prediction approaches applied to the modeling of soil behavior are often solved in the forward direction, by measuring the response of the soil (outputs) to a given set of soil inputs. Conversely, one may be interested in the assessment of a given set of soil inputs that leads to given (target) soil outputs. This is the inverse of the former problem. In this study, we develop and test an inverse artificial neural network model for the prediction of the optimal soil treatment to reduce copper (Cu) toxicity assessed by a given target concentration of Cu in dwarf bean leaves (BL) from selected soil inputs. In this study the inputs are the soil pH, electrical conductivity (EC), dissolved organic carbon (DOC) and a given target toxicity value of Cu, whereas the output is the best treatment to reduce the given toxicity level. It is shown that the proposed method can successfully identify the best soil treatment from the soil properties (inputs). Two important challenges for optimal treatment prediction using neural networks are the non-uniqueness of the solution of the inverse problem and the inaccuracies in the measurement of the soil properties (inputs). It is shown that the neural network prediction model proposed can overcome both these challenges. It is also shown that the proposed inverse neural network method can potentially be applied with a high level of success to the phytoremediation of contaminated soils. Before large-scale application, further validation is needed by performing several experiments and investigations including additional factors and their combinations to capture the complex soil behavior.