Chemometric approach for prediction of uranium pathways in the soil

Abstract

Understanding the effect of soil parameters (pH, Eh and organic and inorganic ligands availability) on uranium mobility under different geochemical conditions is fundamental for reliable prediction of its behaviour and fate in the environment. In this study, the impact of total and available phosphorus content, humus and acidity of Serbian agricultural soils on the content of total and available uranium were evaluated by Response Surface Methodology (RSM), second order polynomial regression models (SOPs) and artificial neural networks (ANNs). The performance of ANNs was compared with the performance of SOPs and experimental results. SOPs showed high coefficients of determination (0.785-0.956), while ANN model performed high prediction accuracy: 0.8893-0.904. According to the results, total and available uranium content in the soil were mostly affected by pH, statistically significant at p LT 0.05 level. For the same responses the total phosphorus was found to be also very influential, statistically significant at p LT 0.05 and p LT 0.10 levels. The impact of available phosphourus and humus was much more influential on total and available uranium content, compared to total phosphorus content. Proposed chemometric approach will be very helpful in preserving the natural resources and practical application for risk assessment modeling of uranium environmental pathways.