Abstract
Positive matrix factorization (PMF) has been applied extensively in source apportionment of atmospheric pollutants. Data uncertainty is one of the most important input information for PMF analysis. Currently, data uncertainty is calculated by some simple formulae referenced from previous studies, lacking evaluation of their reasonableness and applicability. In this study, we develop an uncertainty assessment framework (UAF) by utilizing three common uncertainty calculation algorithms. The effectiveness of this UAF is assessed by comparing with the source apportionment results on parallel measurement of PM10 compositions at Tsuen Wan station in Hong Kong during 1998-2008. It is found that applying UAF could further decompose some factors which couldn't by traditional methods. The derived scaled residuals are even smaller than those from source apportionment on parallel measurement. Source contributions derived by UAF are well in between those from traditional methods. All above indicates the reliability and completeness of source apportionment by UAF. It is thus concluded that this UAF has great applicability and usefulness in ensuring the accuracy of source apportionment results.
| Translated title of the contribution | Establishment of an uncertainty assessment framework for atmospheric pollutant monitoring data and its impact on PMF source apportionment |
|---|---|
| Original language | Chinese (Traditional) |
| Pages (from-to) | 95-104 |
| Number of pages | 10 |
| Journal | Huanjing Kexue Xuebao/Acta Scientiae Circumstantiae |
| Volume | 39 |
| Issue number | 1 |
| DOIs | |
| Publication status | Published - 6 Jan 2019 |
Bibliographical note
Publisher Copyright:© 2019, Science Press. All right reserved.
Keywords
- Positive matrix factorization
- Source apportionment
- Uncertainty assessment framework