Aquatic Ecosystem Risk Assessment Generated by Accidental Silver Nanoparticle Spills in Groundwater

This paper aims to create a new model for assessing the ecosystem risk in rivers and wetlands that are linked to accidental spills of silver nanoparticles (AgNPs) in soil/groundwater. Due to the uncertainty of the modeling inputs, a combination of two well-known risk assessment methodologies (Monte Carlo and fuzzy logic) were used. To test the new model, two hypothetical, accidental AgNP soil spill case studies were evaluated; both of which were located at the end of the Llobregat River basin within the metropolitan area of Barcelona (NE Spain). In both cases, the soil spill reached groundwater. In the first case, it was discharged into a river, and in the second case, it recharged a wetland. Concerning the results, in the first case study, a medium-risk assessment was achieved for most cases (83%), with just 10% of them falling below the future legal threshold concentration value. In the second case study, a high-risk assessment was obtained for most cases (84%), and none of the cases complied with the threshold value. A sensitivity analysis was conducted for the concentration and risk. The developed tool was proven capable of assessing risk in aquatic ecosystems when dealing with uncertain and variable data, which is an improvement compared to other risk assessment methodologies.

Automated summary

This paper proposes a new model for assessing ecosystem risk in rivers and wetlands caused by accidental spills of silver nanoparticles in soil/groundwater. The model combines Monte Carlo and fuzzy logic methodologies to deal with uncertainty in modeling inputs. Two hypothetical case studies of accidental AgNP soil spills in the Llobregat River basin, Barcelona, were evaluated using the new model, showing medium and high risk assessments respectively. A sensitivity analysis was conducted, and the new model was proven to be an improvement compared to other risk assessment methodologies in dealing with uncertain and variable data.