Source identification of $PM_{10}$ using positive matrix factorization in the case of a real industrial French site

Abstract

The aim of this study is to identify the main sources responsible of airborne particles around an industrial site. For this purpose, two sampling points were installed around a real industrial site, characterized by a variety of steel-plant processes. Several chemical analyses were performed in order to determine the concentrations of 18 trace metals. An advanced technique called Positive Matrix Factorization (PMF) was used to find the sources and their relative contribution to $PM_{10}$. PMF has a substantial advantage over more conventional methods in receptor modeling due to non-negativity of factors and use of error estimates. The present study is the first reported comparison between sources profiles given by PMF and the real iron and steel sources such as Basic Oxygen Furnace, Electric Arc Furnace, steelworks and coke plants. The results show that the choice of trace metal is very important to judge their representativeness compared to measure total mass $PM_{10}$. Globally, the identification, according to real source profile, is in a good agreement with PMF source profile, and their inter-comparison revealed the most robust association of elements. The contribution of identified emission sources has large temporal and spatial variations and can provide a good information to validate identification using others exogenous information.