the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Mar 2024
Diurnal variations of NO2 tropospheric vertical column density over the Seoul Metropolitan Area from the Geostationary Environment Monitoring Spectrometer (GEMS): seasonal differences and impacts of varying a priori NO2 profile data
Abstract. Geostationary Environment Monitoring Spectrometer (GEMS), launched in 2020, provides both temporally and spatially continuous air quality data from geostationary Earth orbit (GEO). In this study, we analyzed seasonal characteristics of GEMS tropospheric NO2 vertical column density (NO2 TropVCD) diurnal patterns and impacts of a priori data from diverse chemical transport model (CTM) simulations over the Seoul Metropolitan Area (SMA), using the GEMS products retrieved by the IUP algorithm. We found that both the amounts of NO2 TropVCD and the peak time vary according to the season – the maximum value occurs earlier in July (10 KST) compared to other months (12 KST), with relatively lower value (8.53 – 9.81 × 1015 molec. cm-2). In wintertime, the decrease in NO2 TropVCD over time was relatively slower than in summertime. Also, we examined the impact of changes in a priori data on the GEMS NO2 TropVCD. When we compare GEMS NO2 data retrieved with default NOx emissions and uniformly 20 %-reduced NOx emissions, there are no notable discrepancies as simulated NO2 profiles from CTM are nearly identical over the SMA. However, when the vertical profile at 06:45 UTC (13:45 KST) was applied for retrievals at all times, there are 11.9 – 16.1 % lower values before 13:45 KST and up to 4.9 % higher values after 13:45 KST compared to the control run case. Our study highlighted two key findings: (1) GEMS NO2 products describe distinct seasonal features, including the absolute values (highest in January and lowest in July) and diurnal patterns (persisting longer in January and declining rapidly in July), (2) changes of a priori data have the impacts of up to 19.2 % on the GEMS NO2 TropVCD.
- Preprint
(1782 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 25 May 2024)
-
RC1: 'Comment on amt-2024-33', Anonymous Referee #1, 15 Apr 2024
reply
review of "Diurnal variations of NO2 tropospheric vertical column density over the Seoul Metropolitan Area from the Geostationary Environment Monitoring Spectrometer (GEMS): seasonal differences and impacts of varying a priori NO2 profile data” by Seunghwan Seo et al.
The paper "Diurnal variations of NO2 tropospheric vertical column density over the Seoul Metropolitan Area from the Geostationary Environment Monitoring Spectrometer (GEMS): seasonal differences and impacts of varying a priori NO2 profile data” by Seunghwan Seo et al., presents NO2 VCD comparisons results from GEMS retrievals over a few months in 2021, testing different a-priori profiles, from one model but changing NO2 emissions and/or only considering the profile at one time of the day, as used/seen from the LEO (TROPOMI's TM5 apriori) at 13h45 LT. They show that although the model can have very different NO2 VCD (magnitude and diurnal evolution) and vertical profile distributions, the impact on GEMS VCD is of the order of 20% at maximum.
This study is interesting, well written, easy to follow and in the scope of AMT, but it is too short in my opinion. Several points are only briefly touched, with lack of proper reference and explanation of the reason of the findings (some are well known in the community, and proper mention and reference need to be included). Also it would be nice to have more referencing and discussion of similarities and/or differences with other papers of the special issues that touch similar topics, in order to "put the results in perspective". A few additions/tests and clarifications would be beneficial to make the paper much more interesting.
I would thus suggest publication after a major revision, including the addition of at least a few of the suggested ideas below.
Major points for improvement suggestions:
------------------------------------------
1) it is well known that it is not the apriori profile itself but its shape that is relevant for DOAS-type of retrievals (ie, see Palmer et al., 2011; Eskes and Boersma 2003 or discussions eg in Yang et al. 2023a), so it would be nice to compare in a quantitative way the different apriori profiles shapes, and not only their VCD and the curtain plots as done now in Figures 3 and 4. I suggest to use a quantitative estimate of the height that contain the larger part of the profile (see details below).2) the test with the emission changes is one of the key points of this paper, but several parts are not clear to me: why choose emissions from Los Angeles while emissions for Korea exist from KORUS-AQ campaigns or their modified version (as used in Yang et al. 2023a and 2023b and in Edwards et al 2024)? It should be explained. In addition, in particular the diurnal factor mentioned in P5, line 27: it would be nice to see what is this diurnal evolution (and not only refer to a paper), to link it with the VCD evolution, and also comment its difference with respect to the above-mentioned emissions estimates for Korea. Moreover, I am not an expert on emissions, but I am wondering if 20% change is really a "big" change: in several publications larger changes are seen over the day (see my detailed comment below for P5, line 27).
I would explain more here the reasons of these choices for the emission changes tests and help the reader in understanding why these changes do not "appear" in the corresponding models and GEMS VCD. Is this because the test is too restrictive/not realistic or is this related to the WRF-Chem model (version) vertical mixing that is not strong enough to allow surface differences to also have an impact on the tropospheric column?
I am surprised that 20% changes in surface (emissions) would not affect the Planetary Boundary Layer so much, to only change the NO2 profile shape so little that this is almost not seen between v2 and v3 test (or am I missing something?).3) the AMF shown in Fig. 7 and 8 are all smaller than 1, with some diurnal and seasonal changes and different magnitude for 2 of the models used as apriori profile. When looking at other papers in the special issue, and esp. Yang et al. 2023a Figure 6 and Table 1, I am surprised how different are the AMF values. For Korus AQ conditions, May-June 2016, also over SMA region, they calculate AMF values always larger than 1. Please comment on the differences and ideally also add on Fig 8 the geometrical AMF (the Yang et al calculation is between 2.5 and 3), so that we could compare comparable quantities (only the part related to geometry and not the choice of the model used as apriori). The profile shape should be very different, with surface (or PBL) contribution probably much more peaked in your case compared to what they measure/use from KORUS-AQ ?
4) the WRF-Chem model has a higher resolution (28x28km2) than the TM5 model (1°x1°) model used in TROPOMI retrievals. It would be interesting to average the WRF-Chem model to the 1°x1° resolution, to quantify the dilution effect mentioned in p.7, line 23.
Minor points/specific comments:
--------------------------------- P2, line 14: "uniformly 20%-reduces": uniformly in space or in time? or both?
- P2, line 16: say why 13h45 KST (it is TROPOMI overpass time!)
- P4, line 24: version 0.9. It is good to have a clear info on the version, but as the paper is not yet available, it is not so easy to compare, eg with Lange et al 2024 paper in the same special issue, that also uses IUPB GEMS NO2 data. Is this the same version? If there is no cloud correction, what do you do? you filter based on GEMS cloud fraction? you keep all the points? Please specify.
- P5, sect. 2.2: some details, as the number of layers are given, which, as is, do not bring much information. Give the number of layer wrt to Top and bottom of the atmosphere, or the number of layers within the troposphere? Are the tropopause definitions the same between the models?
- P5, line 27: "a diurnal factor" --> this is an important element in my opinion, which is not explained enough. Why is Los Angeles relevant for Korea? why shifting it by 1 hour? is the diurnal factor the same for every season? Is the diurnal evolution larger than the 20% changes used for test v3? It would be nice to have an illustration of the time evolution of this factor, in order to be able to compare it to other choices made by other papers in the special issue. Eg diurnal emission pattern in Yang 2023b, or Edwards et al. 2023, using NIER/KU-CREATE inventory from KORUS-AQ filed campaign (fig2): there are up to 75% changes during the day. Jo et al 2023 uses MUSICA model with different resolutions, and there they use KORUS v5 inventory. There is a nice discussion about different emission estimates and incorporation of diurnal variation into existing monthly emission estimates. It would be nice to refer to these publications and comment the relevance of your choice and how much it is different from the others.
The 20% emission reduction test: 20% is maybe not a change "big enough"? in Yang et al. 2023b, the ratios of 2022/2015 in Korea are of 0.70 for DJF and 0.50 in JJA, so 30% and 50% reductions... please comment.
Table 1: add in the first line (v2) if it is by default with the diurnal factor.
See major comment n°2).
- P7, line 21: add the pink box in Fig 2 also in a non-empty subplot (it would be easier to follow the discussion.)
Also relevant here: line 23 "The coarser horizontal resolution of the TM5 model would be one of the reasons why these differences occur": although I tend to agree with this statement, it is too speculative. You have a model with higher resolution, so you could actually resample it to 1°x1° resolution and quantify how much of the seen difference is due to purely dilution effects related to the different resolution. See major point 4).
It would also be good to cite studies that tried to quantify NO2 dilution effects (with satellites or models).
- P8, lines 2 to 9): this discussion lacks an important point. In DOAS-like satellite retrievals, as in GEMS, only the apriori profile SHAPE is important, not its magnitude. This should be mentioned here and discussed. I would recommend adding some plots of the profile shapes comparison between WRF-Chem (v2 and v3) and TM5, maybe through H75 diurnal variation comparisons (H75 is the height at which 75% of the profile is included, see eg Vlemmix et al. 2015 (see major comment n° 1)
- P8, line 18: "diurnal profiles of GEMS" --> the term profile is misleading. I would suggest something like "diurnal VCD evolution", not to confuse the reader with the vertical profile.
- P8, line 219: why the 25-28 October 2021 period was chosen?
- P8, line 23: model tropo VCD between v2 and v3 are nearly identical --> same model, different emissions: is the profile shape the same? (I suppose so, as no changes in GEMS VCD). This would mean that the model has too small vertical mixing? or that the 20% change emission is a too small change? please comment.
Also, I am wondering why there is a small reduction of the VCD at 14h for f3 (and the others) in Fig 6 (and thus also Fig 5), while it is constant for the rest of the day? How do you explain this?
- P9, line 2: "almost no temporal variations of AMF" --> this is the part related to viewing angles, albedo, etcc, which do not change much during that period? adding the geometric AMF would help understanding better this.
- P9, line 6 to 8: "Notably, despite... " --> this is good: this is the purely SCD contribution, hopefully not all the VCD change is coming from the AMF! I would insist on this. What is the part of the VCD that is actually coming from the measurements (the SCD) and which one is from AMF? Is there a time (diurnal and seasonal) variability?
- table 1: clarify if some of these tests have constant emission factor during the day or if all share the same diurnal factor. It is not clear to me. + add a plot of the emission evolution during the day.
- Fig1: can you comment on WRF_chem v2. Is it the same than v3 for all seasons? (in Fig 5 for a few days yes)
- Fig2: in April, for the first hours of measurements there seems to be a clear yellow band on the left of the map (figure is too small and is difficult to estimate the longitude) compared to bluish columns over Korea. Is this due to a resolution effect of the model? or the albedo? or a land/ocean effect? Please comment.
references:
-------------- Edwards et al. 2023 : Edwards, D. P., Martínez-Alonso, S., Jo, D. S., Ortega, I., Emmons, L. K., Orlando, J. J., Worden, H. M., Kim, J., Lee, H., Park, J., and Hong, H.: Quantifying the diurnal variation of atmospheric NO2 from observations of the Geostationary Environment Monitoring Spectrometer (GEMS), EGUsphere [preprint], https://doi.org/10.5194/egusphere-2024-570, 2024.
- Jo et al 2023: Jo, D. S., Emmons, L. K., Callaghan, P., Tilmes, S., Woo, J.-H., Kim, Y., et al. (2023). Comparison of urban air quality simulations during the KORUS-AQ campaign with regionally refined versus global uniform grids in the Multi-Scale Infrastructure for Chemistry and Aerosols (MUSICA) version 0. Journal of Advances in Modeling Earth Systems, 15, e2022MS003458. https://doi.org/10.1029/2022MS003458
- Yang et al., 2023a: Yang, L. H., Jacob, D. J., Colombi, N. K., Zhai, S., Bates, K. H., Shah, V., Beaudry, E., Yantosca, R. M., Lin, H., Brewer, J. F., Chong, H., Travis, K. R., Crawford, J. H., Lamsal, L. N., Koo, J.-H., and Kim, J.: Tropospheric NO2 vertical profiles over South Korea and their relation to oxidant chemistry: implications for geostationary satellite retrievals and the observation of NO2 diurnal variation from space, Atmos. Chem. Phys., 23, 2465–2481, https://doi.org/10.5194/acp-23-2465-2023, 2023.
- Yang et al. 2023b: Yang, L. H., Jacob, D. J., Dang, R., Oak, Y. J., Lin, H., Kim, J., Zhai, S., Colombi, N. K., Pendergrass, D. C., Beaudry, E., Shah, V., Feng, X., Yantosca, R. M., Chong, H., Park, J., Lee, H., Lee, W.-J., Kim, S., Kim, E., Travis, K. R., Crawford, J. H., and Liao, H.: Interpreting GEMS geostationary satellite observations of the diurnal variation of nitrogen dioxide (NO2) over East Asia, EGUsphere [preprint], https://doi.org/10.5194/egusphere-2023-2979, 2023.
- Vlemmix, T., Hendrick, F., Pinardi, G., De Smedt, I., Fayt, C., Hermans, C., Piters, A., Wang, P., Levelt, P., and Van Roozendael, M.: MAX-DOAS observations of aerosols, formaldehyde and nitrogen dioxide in the Beijing area: comparison of two profile retrieval approaches, Atmos. Meas. Tech., 8, 941–963, https://doi.org/10.5194/amt-8-941-2015, 2015.
- Lange et al., 2024: Lange, K., Richter, A., Bösch, T., Zilker, B., Latsch, M., Behrens, L. K., Okafor, C. M., Bösch, H., Burrows, J. P., Merlaud, A., Pinardi, G., Fayt, C., Friedrich, M. M., Dimitropoulou, E., Van Roozendael, M., Ziegler, S., Ripperger-Lukosiunaite, S., Kuhn, L., Lauster, B., Wagner, T., Hong, H., Kim, D., Chang, L.-S., Bae, K., Song, C.-K., and Lee, H.: Validation of GEMS tropospheric NO2 columns and their diurnal variation with ground-based DOAS measurements, EGUsphere [preprint], https://doi.org/10.5194/egusphere-2024-617, 2024.
Citation: https://doi.org/10.5194/amt-2024-33-RC1 -
RC2: 'Comment on amt-2024-33', Anonymous Referee #2, 26 Apr 2024
reply
The manuscript "Diurnal variations of NO2 tropospheric vertical column density over the Seoul Metropolitan Area from the Geostationary Environment Monitoring Spectrometer (GEMS): seasonal differences and impacts of varying a priori NO2 profile data" investigates diurnal variations of tropospheric NO2 around Seoul, and investigates the impact of the a-priori profile of NO2.
The topic of the study matches the scope of AMT and is of high interest, as GEMS is the first satellite instrument providing NO2 measurements on geostationary orbit.
However, I see two major concerns with the current manuscript:
1. Clouds have a strong impact on the visibility of tropospheric NO2 from space. This is shortly mentioned in the manuscript (Page 3, Line 20), but nevertheless ignored in the analysis (Page 4, Line 25).
Clouds potentially affect both key topics of the paper:
- Cloud parameters vary over the day as well. This affects visibility (i.e., AMFs), and can lead to virtual diurnal patterns in the NO2 column if not accounted for properly.
- Cloud effects are coupled to the vertical profile, as they shield the column below, but increase the visibility above.
Thus, depending on cloud height, I would expect that for partly clouded pixels the AMF variations for different a-priori profiles can be much larger than those shown in Fig. 8, and the conclusion that "different a-priori profiles ... have only minimal impact" is potentially misleading.
Thus, this study needs (a) to explicitely consider, or (b) at least throroughly (and quantitatively) discuss cloud effects.2. The WRF simulations do not match the observed columns, but are far higher.
Before using WRF as kind of reference for investigating the impact of vertical profiles, this issue should be fixed (which is probably related to far too high emissions, definitely more than the 20% investigated in the manuscript).Due to theese issues, the manuscript needs extensive major revisions, which will also affect the drawn conclusions. Thus I do not provide detailed comments on the text at this stage, but an additional iteration for review will be required after the manuscript has been revised accordingly.
Additional comments:
- Page 2 Line 11: "decrease over time" is potentially misleading, as it might be understood as a temporal trend, while I think it is meant to describe the diurnal patterns. This should be clarified (e.g. by adding "diurnal").
- Page 3 Line 20: terrain height is not varying.
- Page 5 Line 4: what is different in the "variant" from STREAM?
- Figures: The diurnal cycle of NO2 is prominent topic of this study, but it is actually not shown; Fig. 2 displays differences of columns for different models, but not the column itself.
I would recommend to add a figure showing the diurnal cycle of VCD maps. For sake of readability, I would recommend to show only every second hour. A similar figure of the diurnal patterns of cloud fraction and cloud height might be added as well.- I don't understand Fig. 6: f2 is a scenario with fixed WRF profile over the day. How can this fixed profile show diurnal variations in the modeled column?
Citation: https://doi.org/10.5194/amt-2024-33-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 233 | 68 | 13 | 314 | 9 | 11 |
- HTML: 233
- PDF: 68
- XML: 13
- Total: 314
- BibTeX: 9
- EndNote: 11
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1