Modeling and surface observations of arsenic dispersion from a large Cu-smelter in southwestern Europe

Arsenic is a toxic element for human health. Consequently, a mean annual target level for arsenic at 6 ng m−3 in PM10 was established by the European Directive 2004/107/CE to take effect January 2013. Cu-smelters can contribute to one-third of total emissions of arsenic in the atmosphere. Surface observations taken near a large Cu-smelter in the city of Huelva (Spain) show hourly arsenic concentrations in the range of 0–20 ng m−3. The arsenic peaks of 20 ng m−3 are higher than values normally observed in urban areas around Europe by a factor of 10. The Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model has been employed to predict arsenic emissions, transport, and dispersion from the Cu-smelter. The model utilized outputs from different meteorological models and variations in the model physics options to simulate the uncertainty in the dispersion of the arsenic plume. Modeling outputs from the physics ensemble for each meteorological model driving HYSPLIT show the same number of arsenic peaks. HYSPLIT coupled with the Weather Research and Forecasting (WRF-ARW) meteorological output predicted the right number of peaks for arsenic concentration at the observation site. The best results were obtained when the WRF simulation used both four-dimensional data assimilation and surface analysis nudging. The prediction was good in local sea breeze circulations or when the flow was dominated by the synoptic scale prevailing winds. However, the predicted peak was delayed when the transport and dispersion was under the influence of an Atlantic cyclone. The calculated concentration map suggests that the plume from the Cu-smelter can cause arsenic pollution events in the city of Huelva as well as other cities and tourist areas in southwestern Spain.

► We measured hourly As concentrations near a large Cu-smelter in SW Spain.
► The HYSPLIT model was used to predict arsenic emissions, transport, and dispersion from a Cu-smelter.
► The model used a variety of meteorological drivers. ▶ Best results were obtained when the WRF simulation used both four-dimensional data assimilation and surface analysis nudging.