Source apportionment using particle number size distribution

Air pollution is a complex phenomenon with many interdependent factors such as meteorology, source emission, humidity, temperature, etc. affecting it. This makes modelling and predicting the air pollution an immensely hard task riddled with uncertainties. To overcome this challenge of modelling air pollution, the approach was flipped from “modelling air pollution from known source to a receptor” to “modelling the relative contributions of different sources at a known receptor”. This process is called source apportionment. Source apportionment is a big field of research and studies, but a large portion of these belong from the past. In other words, in recent times, no significant innovations have been made to the task of source apportionment. The motivation for an innovative approach lies not only in the importance of the task but also in the inadequacy of the conventional methods, namely they were slow, expensive and laborious.

This study dwells into an innovative approach to perform the task of source apportionment, namely switching from chemical composition data, to Particle Number Size distribution (PNSD) data. This single change helps overcome all the limitation of the older methods, so the paper covers the approaches and models from its construction, experimentation and finally validation. The study demonstrates that MLRGMM, is a sufficiently capable method that outperforms the conventional method in specific conditions.