Can we separate industrial CH4 emission sources from atmospheric observations? - A test case for carbon isotopes, PMF and enhanced APCA. (August 2018)

• Record Type: Journal Article
• Authors: Assan, Sabina; Vogel, F.R.; Gros, V.; Baudic, A.; Staufer, J.; Ciais, P.
• Is Part Of: Atmospheric environment. Volume 187(2018)
• Page Start: 317
• Page End: 327
• Publication Date: 2018-08
• Subjects: Methane -- Isotopes -- VOCs -- PCA -- PMF -- Natural gas -- Ethane; Air -- Pollution -- Periodicals; Air -- Pollution -- Meteorological aspects -- Periodicals; 551.51
• DOI: 10.1016/j.atmosenv.2018.05.004
• Languages: English
• ISSNs: 1352-2310

Abstract

Abstract: A large factor contributing to the uncertainties associated with sector specific anthropogenic methane emissions is the lack of available methods and data to reliably discriminate the different production processes. In this study, a variety of source apportionment techniques were investigated and developed to improve CH4 apportionment for co-located CH4 sources. The goal was to distinguish emissions from different systems at a mid-stream natural gas (NG) site (compressor station). Continuous measurements of atmospheric CH4 and co-emitted volatile organic compounds (VOCs) were analysed using Principle Component Analysis (PCA) and Positive Matrix Factorisation (PMF) receptor models. After sensitivity studies, significant extensions were made to the classical PCA, (Monte Carlo Absolute PCA 'MC-APCA' and Monte Carlo moving Absolute PCA 'MC-mAPCA') to better suit this application. Results from the receptor models are compared and combined with isotopic analysis, examining both long and short-term temporal variations. This work determined MC-APCA and PMF to be the most appropriate techniques for the long-term analysis of dominant sources; according to MC-mAPCA, this was identified to be 77.5 ± 0.6% natural gas, while 66% of the remaining variability was associated with traffic-related proxies. Techniques such as the moving Miller-Tans method for isotopic source identification and MC-mAPCA gave an insight into the short-term variability of source composition. During most CH4 enhancement periods, δ 13 CH4 sources range from −40‰ to −45‰, while MC-mAPCA typically show methane to ethane ratios of 4%–8%, confirming the prevalence of NG emissions. Both techniques identified CH4 enhancements from intermittently contributing sources (in this case a ruminant source characterised by δ 13 CH4 -62 ± 3‰ and 0% C2 H6 :CH4 ), and differentiate small fluctuations in NG source composition. Overall, the best method to identify CH4 sources from atmospheric local measurements remains strongly dependant on the characteristics of said source. The campaign investigated here required sensitive analysis as it was predominantly single sourced, and focussed on the identification of two gas streams. In such a case, it was found that a combination of old and novel techniques provide the greatest information on the characteristics of CH4 sources and gives confidence in the results.