Combining Positive Matrix Factorization and Radiocarbon Measurements for Source Apportionment of PM2.5 from a National Background Site in North China.

To explore the utility of combining positive matrix factorization (PMF) with radiocarbon ((14)C) measurements for source apportionment, we applied PM2.5 data collected for 14 months at a national background station in North China to PMF models. The solutions were compared to (14)C results of four seasonally averaged samples and three outlier samples. Comparing the most readily interpretable PMF solutions and (14)C results revealed that PMF modeling was well able to capture the source patterns of PM2.5 with two and three irrelevant source classifications for the seasonal and outlier samples. The contribution of sources that could not be classified as either fossil or non-fossil sources in the PMF solution, and the errors between the modeled and measured concentrations weakened the effectiveness of the comparison. Based on these two factors, we developed an index for selecting the most suitable (14)C measurement samples for combining with the PMF model. Then we examined the potential for coupling PMF modeling and (14)C data with a constrained PMF run using the (14)C data as a priori information. The restricted run could provide a more reliable solution; however, the PMF model must provide a flexible dialog to input the priori restrictions for executing the constraint simulation.