Source apportionment for online dataset at a megacity in China using a new PTT-PMF model

Air pollution negatively impact human health and the environment. Source apportionment is an important research method for designing effective air pollution control policies. In this study, a new PTT-PMF (Partial Target Transformation-Positive matrix factor) source apportionment method (based on Multilinear Engine 2 program) was developed, to better recognize the extracted source categories, for online dataset. By incorporating measured source profiles, this model can automatically extract factors that have physical significance and close to actual source profiles. The accuracy of the PTT-PMF method was first evaluated by using several artificial datasets, and the results were favorable. Results of the new model can better explain similar sources such as dust and coal combustion (when lack of markers such as Ai and Si), comparing with the base run obtained using PMF. Then, this model was applied to online measurement data collected in Tianjin, China. Six factors were obtained and they were close to actual source profiles: dust (6.06%), coal combustion (4.70%), secondary sulfate sources (26.36%), secondary nitrate sources (42.91%), vehicle exhaust (9.97%) and biomass burning & SOC (10.01%). Time series of source contributions can also help to identify the source categories. The work presented here may advance receptor modelling techniques and could also enhance the validity of source apportionment results.