The application of positive matrix factorization with diagnostics to BIG DATA

Over the past decade, Positive Matrix Factorization (PMF) has become the most commonly used tool for source apportionment of air pollutants with a large fraction of those studies using the implementation developed by the U.S. Environmental Protection Agency (EPA-PMF) and its very useful diagnostic tools. During the same period, more studies are utilizing monitoring tools that provide hourly or even higher time resolution data leading to large data sets (millions of data points). Because of the nature of the graphic user interface in EPA-PMF, it cannot handle data sets more than approximately half a million data points. Thus, alternative approaches are needed to permit these larger data sets to be analyzed and still be able to provide the diagnostic outputs that permit better interpretations of the results. We have developed a protocol to use the multilinear engine (ME-2) that is the solver used in EPA-PMF to make such analyses. Here we report how to use this approach and present results for a representative data set of particle size distributions obtained in Rochester, New York between 2008 and 2019.

Automated summary

Over the past decade, Positive Matrix Factorization (PMF) has become the most commonly used tool for source apportionment of air pollutants. However, the implementation developed by the U.S. Environmental Protection Agency (EPA-PMF) has limitations in handling large data sets. In this study, we propose a protocol using the multilinear engine (ME-2) to analyze larger data sets and present results for particle size distributions in Rochester, New York.