Source apportionment of atmospheric aerosol in a marine dusty environment by ionic/composition mass balance (IMB)

Cardoso, João; Almeida, Susana M.; Nunes, Teresa; Almeida-Silva, Marina; Cerqueira, Mário; Alves, Célia; Rocha, Fernando; Chaves, Paula; Reis, Miguel; Salvador, Pedro; Artiñano, Begoña; Pio, Casimiro

doi:https://doi.org/10.5194/acp-18-13215-2018

Articles | Volume 18, issue 17

https://doi.org/10.5194/acp-18-13215-2018

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

https://doi.org/10.5194/acp-18-13215-2018

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

Articles | Volume 18, issue 17

Research article

|

13 Sep 2018

Research article |

| 13 Sep 2018

Source apportionment of atmospheric aerosol in a marine dusty environment by ionic/composition mass balance (IMB)

João Cardoso, Susana M. Almeida, Teresa Nunes, Marina Almeida-Silva, Mário Cerqueira, Célia Alves, Fernando Rocha, Paula Chaves, Miguel Reis, Pedro Salvador, Begoña Artiñano, and Casimiro Pio

Download

Final revised paper (published on 13 Sep 2018)
Supplement to the final revised paper
Preprint (discussion started on 02 Mar 2018)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'reviewer comment on acp-2018-10', Anonymous Referee #2, 19 Mar 2018
RC2: 'Comments on Cardoso et al.', Anonymous Referee #3, 28 Mar 2018
RC3: 'Comments on Cardoso et al.', Anonymous Referee #1, 02 Apr 2018
AC1: 'Authors Reply to Reviewers Comments', Casimiro A. Pio, 30 Apr 2018

Peer-review completion

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

AR by Anna Wenzel on behalf of the Authors (04 May 2018) Author's response

ED: Referee Nomination & Report Request started (12 May 2018) by Chak K. Chan

RR by Anonymous Referee #2 (20 May 2018)

RR by Anonymous Referee #1 (21 May 2018)

Suggestions for revision or reasons for rejection

The quality of this manuscript has improved to some extent in the revised version, especially the addition of “Accuracy and Errors” section which is important to alert the readers that both methods have limitations. However, there are still quite a number of questions which hindered me from accepting it for publication. Below are detailed comments.

1. I totally agree that the accuracy of source apportionment can be increased by using different methods, as also suggested by the European Guide. However, I feel that IMB and PMF results, in this case, are not mutually complementary. With respect to PMF, IMB is not totally an “independent methodology”. They both use the same input dataset, and essentially attempt to find factors or species that co-vary with time. The edge line methodology is essentially to constrain some factor loadings to be zero in the factor analysis. Therefore, it is quite normal that IMB and PMF results reach a good agreement. Such an agreement is resulted from the similarity in algorithm, which could not provide us additional confidence to the source apportionment results. It would be a completely different story if the results from PMF and CTMs converge or can be improved by their reconciliation, which is the essential meaning of the argument quoted from the European Guide.
2. In response, the authors claimed that “IMB may help in the obtaining of a more real and accurate solution than that the one obtained only with PMF”, which I cannot agree. In this study, the authors admit that IMB cannot discriminate Dust+Ind which PMF can. In addition, PMF can discriminate different sources of carbonaceous species (e.g. combustion, soil…) which IMB cannot. By looking at the results, probably the only drawback of PMF results is the mixing of elements in the sea-salt and combustion sources which are resulted from the overwhelming presence of soil at this specific location, and such a drawback can be overcame by the latest version of PMF which has the function of constraining source profiles by priori information. It is not a good practice to selectively emphasize some points while neglecting others to support some pre-designed viewpoints.
3. In my opinion, IMB is essentially not a ‘source’ apportionment tool. The core value of doing source apportionment, apart from understanding the atmospheric processes, is more to investigate contributions from different source categories to ambient pollutants so as to make prioritized control strategies to effectively reduce ambient pollution levels. Those identified by IMB, e.g. WetSoilDust ins, WetSIC am, WetCarbon, cannot directly linked with pollution sources, therefore this IMB method has limited utility from a pollution control point of view, especially in some areas with ambient pollution contributed from diverse pollution sources.
4. In IMB, all OC and EC are simply grouped together and named “Non-carbonated carbonaceous matter”, and the reason for doing so is their low and near-detection-limit concentrations. In comparison, with such low concentrations, PMF can at least identify two factors for OC and EC. The authors further claimed, without any supporting information provided, that in another situation with considerably higher carbonaceous matter, it can be divided into biomass burning, vehicle exhaust and secondary formation process by IMB, and they compare very well with PMF. This indicates that in both situations, PMF performs better than, at least equal to, IMB, so what’s the additional value of IMB in this case?
5. The 13-step calculation procedure is the core part of IMB, and the first step of identifying Fe/Na+ and Fe/Mg2+ edge lines is pivotal and would impact all further calculations. As suggested by the authors, the sampling location is “in the middle of the ocean where it is not expectable to have absence of sea-salt”. However, identifying edge lines by the scatterplot essentially assumes that for points along the edge lines, there are little contribution from sea-salt, which is contradictory with the previous argument that sea-salt contribution is ubiquitous. Therefore, we consider the Fe/Na+ and Fe/Mg2+ values from the edge lines are underestimated to unknown degrees (could be well above 10%), leading to significant uncertainties to the following calculations. The ‘real’ edge might be way left of the current edge line.
6. There are two ‘Figure 6’ and two ‘section 4.5’ in this revised version.

Hide

RR by Anonymous Referee #3 (03 Jun 2018)

Suggestions for revision or reasons for rejection

The authors discussed more on the model uncertainty and comparison with PMF results in the revised manuscript. However, many important findings are not mentioned in the abstract and ACP Discussion Forum. The inconsistency between IMB and PMF are significant for many sources, based on Figure 6 and the authors’ explanations, and this proves that the good agreement in the average proportions may just be a coincidence and misleading. It is not fair to say (in the abstract) they: “giving comparable results throughout the sampling campaign. This gives confidence in the capability of both methods, which are complementary, for the source apportionment of aerosol particles.” Disagreement and uncertainties/limitations should be mentioned adequately in the abstract. In addition, many of the authors’ responses lack details, which are only presented in the revised manuscript, beating the purpose of having the ACP discussion forum. Key revisions should be mentioned (even by copying and pasting) in the discussion forum. The reviewer does not recommend publication without this being done.

There is a slight difference in Figure 2 in the response and in the revised manuscript as the upper line passes different points (the one in the Response pass more points than the one in the revised manuscript). Why is there such an inconsistency? How do you decide the number of points to be included as “edge”? In addition, the authors should think about how the various uncertainties can be propagated and presented in Table 3. In the reviewer’s opinion, a value without appropriate uncertainty can be misleading.

Finally, in the conclusion, the authors stated “source composition and contribution knowledge obtained with IMB can be used to constrain a model run with the new versions of the PMF model”. Actually this (constraint) can be achieved by the current EPA PMF 5.0 already. You do not need to wait for a “new” version of PMF.

Hide

ED: Reconsider after major revisions (11 Jun 2018) by Chak K. Chan

AR by Anna Wenzel on behalf of the Authors (10 Jul 2018) Author's response Manuscript

ED: Referee Nomination & Report Request started (23 Jul 2018) by Chak K. Chan

RR by Anonymous Referee #1 (27 Jul 2018)

RR by Anonymous Referee #2 (04 Aug 2018)

RR by Anonymous Referee #3 (10 Aug 2018)

ED: Publish subject to minor revisions (review by editor) (12 Aug 2018) by Chak K. Chan

AR by Casimiro A. Pio on behalf of the Authors (17 Aug 2018) Author's response Manuscript

ED: Publish as is (23 Aug 2018) by Chak K. Chan

Post-review adjustments

AA: Author's adjustment | EA: Editor approval

AA by Casimiro A. Pio on behalf of the Authors (03 Sep 2018) Author's adjustment Manuscript

EA: Adjustments approved (06 Sep 2018) by Chak K. Chan

Download

Article (1672 KB)
Full-text XML

Short summary

Atmospheric aerosol was sampled in Cabo Verde and analysed for chemical composition. Very high levels of dust were measured during the winter months, as a result of transport from Africa. Mass balances were applied to the particulate material, permitting the determination of seven different source contributions to the aerosol loading. The balance methodology compared favourably with standard source apportionment methods, showing its capability for the source apportionment of aerosol particles.