Improving air quality forecasting with the assimilation of GOCI aerosol optical depth (AOD) retrievals during the KORUS-AQ period

Ha, Soyoung; Liu, Zhiquan; Sun, Wei; Lee, Yonghee; Chang, Limseok

doi:https://doi.org/10.5194/acp-20-6015-2020

Articles | Volume 20, issue 10

https://doi.org/10.5194/acp-20-6015-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/acp-20-6015-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 20, issue 10

Research article

|

20 May 2020

Research article |

| 20 May 2020

Improving air quality forecasting with the assimilation of GOCI aerosol optical depth (AOD) retrievals during the KORUS-AQ period

Soyoung Ha, Zhiquan Liu, Wei Sun, Yonghee Lee, and Limseok Chang

Download

Final revised paper (published on 20 May 2020)
Preprint (discussion started on 28 Aug 2019)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'RC 1 comments', Anonymous Referee #1, 07 Oct 2019
- AC2: 'Response to anonymous referee #1', Soyoung Ha, 22 Dec 2019
RC2: 'Review of "Improving air quality forecasting with the assimilation of GOCI AOD retrievals during the KORUS-AQ"', Anonymous Referee #2, 24 Oct 2019
- AC1: 'Response to anonymous referee 2', Soyoung Ha, 22 Dec 2019

Peer-review completion

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

AR by Soyoung Ha on behalf of the Authors (22 Dec 2019) Manuscript

ED: Referee Nomination & Report Request started (10 Feb 2020) by Ilona Riipinen

RR by Anonymous Referee #1 (03 Mar 2020)

RR by Anonymous Referee #2 (19 Mar 2020)

Suggestions for revision or reasons for rejection

The authors considered some of the comments from the reviewers which improved the quality of the manuscript. But on the other hand, they ignored other comments that I believe are important, which are outlined below. The English was improved although there are still some concerns, I tried to point them out below.

My major comment remains being the evaluation of the modeled/assimilated vertical profiles using data available from the KORUS-AQ campaign. They can address this in different ways. There was an HSRL in Seoul (see sample data in Peterson et al., 2019) that can provide full time series of vertically resolved aerosol extinction that can be compared to the simulations. Also, there was an HSRL in the DC8 aircraft, so the comparison can be done for some days across the Korean peninsula. Also, data from the AMS onboard of the DC-8 can be used to compute PM1 and compare to model estimates, there were 2-3 full vertical profiles over Seoul every day the aircraft flew. I’m not asking the authors to do all of these (although this would be nice), but at least show some effort of trying to assess the skill of the model in representing vertical profiles and if the changes in vertical distribution generated by the assimilation are somewhat reflected in better agreement to the observations. If discrepancies arise this is still useful as one can blame model uncertainties (e.g., computation of optical properties) for them. Louisa Emmons (NCAR ACOM, same institution as authors) is a modeler that was heavily involved in the KORUS-AQ campaign, perhaps consulting with her on this topic might be a good idea.

Comments from initial review

R1: Initial reviewer comments, A: Author response, R2: New reviewer comment

R1: Figure 1. Why show observations for a given time? Why not show maybe an average of the period analized?

A: => Figure 1 simply shows the model domain with the observing network. No changes are made.

R2: But you can also use it to show average concentrations over the period analyzed instead of showing a random time. Also, it would be nice to get a second panel with the 2nd domain to see details on the observation distribution over the Korean peninsula

R1: Figures 8 nd 7. You could model vertical distribution and impact after assimilation using airborne data and surface lidars deployed as part of KORUS-AQ

A: => Not clear on your point here. Figures 7 and 8 show how the model responded to the assimilation of observations in use. This analysis is needed to understand how our assimilation worked in the model space. It has nothing to do with verification.

R2: I apologize for the typo, I meant “You could evaluate model vertical distribution …” This is related to one of my main comments on using KORUS-AQ observations for evaluating vertical distributions, that the authors have not addressed.

Comments by line on revised manuscript

2 1. Provide references for this statement, latest IPCC report should do

3 3-8. You can add Park et al (2014) to this paragraph

4 5. You can cite LeGrand e t al. (2019) for the AFWA scheme

8 30-34. If I understood correctly, what you are trying to say here is that there are large values after thinning that were not found in the original data. Please rephrase to make this clearer and to the point.

9 7-16. I’m assuming subscript 1 and 2 in the error (eqns 3-6) correspond to the different verifying object (AERONET or satellite-based retrievals)? Please specify which is which. AERONET is generally treated as ground truth, so that’s probably the one you should be using.

9 27-29. This sentence is not clear, please rephrase.

9 30-31. “This is partly because AOD is not directly associated with surface PM2:5 …” I think what you are trying to say is that AOD is a column integrated quantity while PM2.5 is measured at the surface? Might be better to specify it that way, the way it’s currently stated is vague and not necessarily true (many approaches exist to compute surface PM2.5 from AOD)

9 30-31. “… and partly because large uncertainties in the forecast model and the emission forcing can dominate over the analysis error during the model integration” This is only applicable for forecasts, not for the analysis. Might want to split the arguments.

9 26 – 10 2. Based on the bad correlation, you might want to mention that this is why a model is needed to “translate” AODs into PM2.5. And that this “translation” might depend on the ability of the model of properly represent the aerosols vertically, and the conversion from aerosol mass to optical properties.

Figure 7. How can be sulfate in the background so low (seems to be equal to 0) and different than 0 in the assimilation experiments?

11 5-6 . “When all the observations are assimilated together (in "ALL"), it combines the effect of surface PM2:5 and GOCI retrievals, as expected”. You might want to explain that this combined effect ends up changing the vertical distribution, pulling the surface levels towards PM2.5 and the upper levels to match the AOD columns.

11 32. I think it’s worth mentioning/discussing that although errors are reduced, none of the assimilation experiments are able to reduce the bias compared to the NODA experiment, which is pretty low to start with.

13 21-29. The data used for evaluation in Fig 14 correspond to 3 urban sites where the 9 km res model with simple aerosol chemistry and emissions that have no hourly variation will have a very hard time representing the observed fluctuations. For instance, see Nault et al., (2018), the model configuration you are running doesn’t even consider secondary organic aerosol formation. You can pick sites that you assimilated that are close-by, plot them in the same way, and I would expect you find similar fluctuations. When you average many sites that are both urban and background, you expect some of these fluctuations to be smoothed out. My point is that I wouldn’t blame observation quality in this case.

14 23-24. I would add “and more detailed aerosol chemistry mechanisms”

14 26. “The best use …” I would tone down this statement

15 2-4. Similar to my previous comment, might be better to just state that one is a column integrated quantity and the other is surface, and their connection depends on the vertical distribution of aerosols and conversion from mass to optical properties.

22-23. Not correct, see Peterson et al (2019). There are periods of stagnation where local contribution generates pollution episodes that are not negligible.

Technical Corrections

1 7-10. This sentence reads as you are using MODIS and surface PM to evaluate the assimilation, but are actually assimilating them as well. Please rephrase.

Figure 6. State in the caption which DA experiment is plotted.

10 1. “… coefficient of 0.33, the two observation types …”

10 17 “… observation types separately, …“

Figure 7. State concentration units in the caption. Also, use full specie name instead of abbreviation or reference location in the text where species are defined, some of the names are not straightforward to to figure out.

10 29 “… produces the largest PM2:5 throughout the …”

14 1. “As our best experiment "ALL" analyzed,” not clear what this means, please rephrase

14 9 .”are near sea level”

References
LeGrand, S. L., Polashenski, C., Letcher, T. W., Creighton, G. A., Peckham, S. E. and Cetola, J. D.: The AFWA dust emission scheme for the GOCART aerosol model in WRF-Chem v3.8.1, Geosci. Model Dev., doi:10.5194/gmd-12-131-2019, 2019.

Nault, B. A., Campuzano-Jost, P., Day, D. A., Schroder, J. C., Anderson, B., Beyersdorf, A. J., Blake, D. R., Brune, W. H., Choi, Y., Corr, C. A., de Gouw, J. A., Dibb, J., DiGangi, J. P., Diskin, G. S., Fried, A., Huey, L. G., Kim, M. J., Knote, C. J., Lamb, K. D., Lee, T., Park, T., Pusede, S. E., Scheuer, E., Thornhill, K. L., Woo, J.-H., and Jimenez, J. L.: Secondary organic aerosol production from local emissions dominates the organic aerosol budget over Seoul, South Korea, during KORUS-AQ, Atmos. Chem. Phys., 18, 17769–17800, https://doi.org/10.5194/acp-18-17769-2018, 2018.

Park, M. E., Song, C. H., Park, R. S., Lee, J., Kim, J., Lee, S., Woo, J.-H., Carmichael, G. R., Eck, T. F., Holben, B. N., Lee, S.-S., Song, C. K., and Hong, Y. D.: New approach to monitor transboundary particulate pollution over Northeast Asia, Atmos. Chem. Phys., 14, 659–674, https://doi.org/10.5194/acp-14-659-2014, 2014.

Peterson, D.A., Hyer, E.J., Han, S.-O., Crawford, J.H., Park, R.J., Holz, R., Kuehn, R.E., Eloranta, E., Knote, C., Jordan, C.E. and Lefer, B.L., 2019. Meteorology influencing springtime air quality, pollution transport, and visibility in Korea. Elem Sci Anth, 7(1), p.57. DOI: http://doi.org/10.1525/elementa.395

Hide

ED: Publish subject to minor revisions (review by editor) (23 Mar 2020) by Ilona Riipinen

AR by Soyoung Ha on behalf of the Authors (31 Mar 2020) Author's response Manuscript

ED: Publish subject to minor revisions (review by editor) (15 Apr 2020) by Ilona Riipinen

First, thank you for your patience with the long handling time of this manuscript. Based on a careful review of the revised manuscript and the reviewer comments, I believe this manuscript can be publishable in ACP after the following minor (mostly editorial to improve readability) points have been addressed:

1. Abstract. Please give the abstract one more thorough read and revise it to be as specific and quantitative as possible to improve it's readability and make it more standalone. Here are some examples on potential points to improvement:
- Line 1: What is "high spatial and temporal resolution"? Why not be more specific?
- Line 4: "...the analysis quality". Which analysis? Please specify.
- Line 5: What do you mean by "data characteristics"? Please specify.
- Line 9: What do you mean by "...by comparing with those of other observations"? Please specify exactly which parameters were compared.

2. p. 7 line 30: Please specify that prepbufr is a data format.

3. p. 9 line 7: I agree with Reviewer 1 on the use of term "evaluation" instead of "validation". The fact that the term "validation" has been misused in the literature does not justify continuing to use it when "evaluation" would be the more accurate term. This is of course a minor point, but sometimes the terminology does matter.

4. Figures 7 and 8. Thank you for an extensive response to the Reviewer 2' s request to include an evaluation of the vertical profiles. I do understand your point of the focus of the paper being the surface observations, but it would be good to explicitly comment on the possibility of doing such evaluation in your outlook and perhaps acknowledging the data sets mentioned by the reviewer (and possibly any other relevant ones).

Hide

AR by Soyoung Ha on behalf of the Authors (16 Apr 2020) Author's response Manuscript

ED: Publish as is (17 Apr 2020) by Ilona Riipinen

AR by Soyoung Ha on behalf of the Authors (18 Apr 2020) Manuscript

Short summary

This study examines the effect of aerosol optical depth (AOD) retrieved from the Korean Geostationary Ocean Color Imager (GOCI) sensors on surface PM_2.5 forecasts using the online coupled WRF-Chem forecasting model and the GSI 3D-Var analysis system. During the KORUS-AQ campaign period, the assimilation of GOCI AOD retrieved at the 550 nm wavelength greatly improved air quality forecasting up to 24 h when assimilated with surface PM_2.5 observations, particularly for heavy pollution events.