Principal component analysis of summertime ground site measurements in the Athabasca oil sands with a focus on analytically unresolved intermediate-volatility organic compounds

Tokarek, Travis W.; Odame-Ankrah, Charles A.; Huo, Jennifer A.; McLaren, Robert; Lee, Alex K. Y.; Adam, Max G.; Willis, Megan D.; Abbatt, Jonathan P. D.; Mihele, Cristian; Darlington, Andrea; Mittermeier, Richard L.; Strawbridge, Kevin; Hayden, Katherine L.; Olfert, Jason S.; Schnitzler, Elijah G.; Brownsey, Duncan K.; Assad, Faisal V.; Wentworth, Gregory R.; Tevlin, Alex G.; Worthy, Douglas E. J.; Li, Shao-Meng; Liggio, John; Brook, Jeffrey R.; Osthoff, Hans D.

doi:https://doi.org/10.5194/acp-18-17819-2018

Articles | Volume 18, issue 24

https://doi.org/10.5194/acp-18-17819-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

Atmospheric emissions from oil sands development and their...

https://doi.org/10.5194/acp-18-17819-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 18, issue 24

Research article

|

14 Dec 2018

Research article |

| 14 Dec 2018

Principal component analysis of summertime ground site measurements in the Athabasca oil sands with a focus on analytically unresolved intermediate-volatility organic compounds

Travis W. Tokarek, Charles A. Odame-Ankrah, Jennifer A. Huo, Robert McLaren, Alex K. Y. Lee, Max G. Adam, Megan D. Willis, Jonathan P. D. Abbatt, Cristian Mihele, Andrea Darlington, Richard L. Mittermeier, Kevin Strawbridge, Katherine L. Hayden, Jason S. Olfert, Elijah G. Schnitzler, Duncan K. Brownsey, Faisal V. Assad, Gregory R. Wentworth, Alex G. Tevlin, Douglas E. J. Worthy, Shao-Meng Li, John Liggio, Jeffrey R. Brook, and Hans D. Osthoff

Download

Final revised paper (published on 14 Dec 2018)
Supplement to the final revised paper
Preprint (discussion started on 25 Jan 2018)
Supplement to the preprint

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review comments on “Principal component analysis of summertime ground site measurements in the Athabasca oil sands: Sources of IVOCs”, authored by Tokarek et al.', Anonymous Referee #1, 27 Feb 2018
RC2: 'Comments', Anonymous Referee #2, 24 Apr 2018
RC3: 'Review', Anonymous Referee #3, 05 May 2018
AC1: 'Final author comments', Hans Osthoff, 15 Jun 2018

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Hans Osthoff on behalf of the Authors (15 Jun 2018) Author's response Manuscript

ED: Referee Nomination & Report Request started (18 Jun 2018) by Lynn M. Russell

RR by Anonymous Referee #1 (25 Jun 2018)

RR by Anonymous Referee #4 (27 Jul 2018)

Suggestions for revision or reasons for rejection

I have been asked to review this manuscript at this late stage as I understand that two of the initial reviewers were not available to review the revisions. I find the authors have done a good job of responding to many specific comments, but that a number of issues were not addressed. While I also have some new suggestions/comments, I think the most important thing for the authors to do is more fully address the first round of comments, especially the more general comments of Reviewer 3 for which no changes were made in the manuscript (some of which I repeat and rephrase below).

1. Given the limitations on source sampling:
a. Is the subtitle “Sources of IVOCs” really appropriate? I suggest that if the main source is not really verifiable, it’s not appropriate.
b. Are there GCMS profiles of similar operations available in the literature that could be compared to, providing also a more substantial review of the relevant literature, and comparison of the results of this work to that? This is not that onerous and seems a minimum that the authors can do to provide appropriate context.
2. Correlations:
a. I suggest using the scale and nomenclature of a standard stat reference rather than making one up.
b. Have the authors investigated the degree of autocorrelation in the measurements? Is each measurement independent? If so, the low r values are more meaningful.
3. Is the paper new and interesting? (See general comments from Reviewer 3.)
a. I realize these are not specific comments that can be addressed easily, but I believe the authors can do a better job of more explicitly telling readers what is interesting and new about their work. Is it really the sources? If no, don’t emphasize that in title. If yes, support it with any available literature.
b. For example, additional and specific discussion in response to Reviewer 2 Comment 7 could add some depth to the information presented.

I hope this helps.

Hide

ED: Reconsider after major revisions (27 Jul 2018) by Lynn M. Russell

AR by Hans Osthoff on behalf of the Authors (17 Sep 2018) Author's response Manuscript

ED: Referee Nomination & Report Request started (19 Sep 2018) by Lynn M. Russell

RR by Anonymous Referee #1 (27 Sep 2018)

Suggestions for revision or reasons for rejection

I went through the authors’ responses and was still not very convinced by their rebuttal or still confused by their explanation about the PCA analysis. I noticed that the authors provided more evidence to confirm the unsolved peaks were from bitumen. However, this still cannot well explain the good correlation of gas-phase IVOCs with particle-phase LO-OOA (0.72, factor 5, Table 5) and poor correlation with particle-phase HOA (0.25, factor 5, Table 5). Their argument of “We do not agree, as such correlations could arise from a fraction of IVOCs partitioning to the aerosol phase, perhaps assisted by oxidation” can’t convince me because the oxidative processes of particle-phase IVOCs will most likely destroy the correlation between gas-phase IVOCs and particle-phase LO-OOA, no need to mention that not only IVOCs from bitumen but also IVOCs from other sources can partition to the aerosol phase and be oxidized, which could further ruin the correlation between IVOCs vapor and LO-OOA. In contrast, with regard to the poor correlation between IVOCs vapor and HOA, the authors provided confused response “there are many heavy haulers emitting HOA driving all over the ground covered in bitumen that is dug up and carrying fresh bitumen. It would hence be surprising to not see a correlation between bitumen vapors and HOA”. In fact, the correlation between bitumen vapors and HOA is poor (r=0.25). Should we be surprised about it or not? This actually brought in another issue which I mentioned earlier i.e. collocation of factors. Because most pollutants used by the authors are gas-phase pollutants including IVOCs while HOA and LO-OOA are actually particle-phase pollutants. There is argument in the community that it may not be suitable to run PCA using parameters with different physical and chemical characteristics because of factor collocation problems. Hence, the confused PCA results at least for Factor 5 in Table 5 may be due to the collocation of more than one source. If this is true, the PCA results could mislead the readers. How will the PCA results be if the HOA, LO-OOA and/or PM10-1 are removed from the PCA running?

Hide

RR by Anonymous Referee #5 (11 Oct 2018)

Suggestions for revision or reasons for rejection

1. GC-ITMS: what type of column was used for measuring the IVOCs? Would the IVOCs be limited to hydrocarbon species or would oxygenated species be detected too? Depending on that answer, are the IVOC measurements discussed here primary and/or secondary species?
2. Line 160-167: This discussion would likely benefit from broadly citing and describing the IVOC work that has been done with fossil fuel sources other than aircraft. For example, Gentner et al. (PNAS, 2012) and Zhao et al. (ES&T, 2014, 2015, 2016) that look at IVOC emissions from evaporated fuel and mobile sources.
3. The manuscript at one point (line 173-175) says that species are conserved between emission and measurement but at other points allude to the role of secondary formation and aging (e.g., PCA with secondary pollutants). A short paragraph that discusses the transport times between emission and measurement (based on distances in Table 1 and average wind speeds), average OH concentrations, and kOH of IVOCs would be useful to understand the potential influence of photochemical aging on the measurements.
4. Line 223: As the authors are aware, there are nuanced approaches to dealing with values below the detection limit based on the distribution of the data. See for example, https://www.tandfonline.com/doi/pdf/10.1080/08940630.1989.10466534
5. Section 2.3.2: Does log-transforming the data instead of normalizing help with the PCA?
6. Line 285-286: The source signatures for a single source could also change with environmental conditions and hence result in varying association of the IVOCs with VOCs.
7. Line 328: Stating the coefficient of variation (CoV) would help compare the variability between pollutants better since the CoV is a normalized metric.
8. Line 578-582: Wouldn't photochemical oxidation make the IVOCs less prone to aerosol formation since they would be more likely to fragment? Also, would the GC-ITMS be able to detect them if the IVOCs were oxidized by the microbes? It's possible I have not understood the reasoning for differences in the IVOCs in components 1, 2, and 5 correctly. If that is the case, please consider rephrasing the last sections of that paragraph.

Hide

ED: Reconsider after major revisions (11 Oct 2018) by Lynn M. Russell

AR by Hans Osthoff on behalf of the Authors (19 Nov 2018) Author's response Manuscript

ED: Publish as is (25 Nov 2018) by Lynn M. Russell

AR by Hans Osthoff on behalf of the Authors (03 Dec 2018)

Download

Article (3270 KB)
Full-text XML

Short summary

Measurements of air pollutants at a ground site near Fort McKay in the Athabasca oil sands region in the summer of 2013 are presented. A large number of intermediate-volatility organic compounds (IVOCs) were observed; these molecules were shown previously to generate atmospheric particles downwind of the region. A principal component analysis was performed to identify major pollution source types, including which source(s) is(are) associated with IVOC emissions (e.g., freshly mined bitumen).