Source Identification of Chittagong Aerosol by Receptor Modeling

Abstract:

Air pollution has become an important environmental concern globally, especially in urban areas, in view of the adverse health effects. The national air quality standard for fine particles in many countries around the world including Bangladesh has also created the need to identify the particle sources, so that effective control strategies could be designed and implemented. Receptor modeling technique, a powerful tool for air pollution source apportionment, can provide the relative contributions of different sources at the receptor site where PM samples are collected.
Samples of fine and coarse airborne particulate matter (PM) were collected between February and July 2007 at the Continuous Air Monitoring Station (CAMS) in Chittagong, the second largest city in Bangladesh. The samples were collected using a dichotomous sampler in two fractions of <2.5 µm (fine) and 2.5 to 10 µm (coarse). The samples were analyzed for elemental concentrations by proton induced X-ray emission (PIXE), hydrogen by proton elastic scattering analysis (PESA), and black carbon by reflectometry. The elemental data sets together with black carbon were analyzed by positive matrix factorization to identify the possible sources of mass for the coarse and fine PM fractions. The best solutions were found to be six and seven factors for elemental compositions for coarse and fine fractions at the CAMS at Chittagong, respectively. The sources were identified as biomass burning/brick kiln, soil dust, road dust, Zn source, motor vehicle, CNG vehicle and sea salt. The PMF results show that about 35.5% of PM2.5 mass at this site comes from biomass burning. The second largest contribution of fine PM comes from motor vehicle including CNG vehicles. The third one is Zn source that include emissions from two-stroke vehicles and galvanizing factories. In case of coarse PM, about 40% of PM2.5-10 mass comes from soil dust including road dust.

Keywords: Source Identification, Receptor Model, PIXE, PESA, PMF