the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 13 Jan 2025
Global Ground-based Tropospheric Ozone Measurements: Reference Data and Individual Site Trends (2000–2022) from the TOAR-II/HEGIFTOM Project
Abstract. Tropospheric ozone trends from models and satellites are found to diverge. Ground-based (GB) observations are used to reference models and satellites but GB data themselves might display station biases and discontinuities. Re-processing with uniform procedures, the TOAR-II Working Group Harmonization and Evaluation of Ground-based Instruments for Free-Tropospheric Ozone Measurements (HEGIFTOM) homogenized public data from 5 networks: ozonesondes, In-service Aircraft for a Global Observing System (IAGOS) profiles, solar absorption Fourier-Transform Infrared (FTIR) spectrometer measurements, Lidar observations, and Dobson Umkehr data. Amounts and uncertainties for total tropospheric ozone (“TrOC”, surface to 300 hPa), free and lower tropospheric ozone, are calculated for each network. We report trends (2000 to 2022) for these segments using Quantile Regression (QR) and Multiple Linear Regression (MLR) for 55 datasets, including 6 multi-instrument stations. The findings: (1) Median TrOC trends computed with QR and MLR trends are essentially the same; (2) Pole-to-pole, across all longitudes, TrOC trends fall within +3 ppbv/decade to -3 ppbv/decade, equivalent to (-4 % to + 8 %)/decade depending on site. (3) The greatest fractional increases occur over most tropical/subtropical sites with decreases at northern high latitudes but these patterns are not uniform. (4) Post-COVID trends are smaller than pre-COVID trends for Northern Hemisphere mid-latitude sites. In summary, this analysis conducted in the frame of TOAR-II/HEGIFTOM shows that high-quality, multi-instrument, harmonized data over a wide range of ground sites provide clear standard references for TOAR-II models and evolving tropospheric ozone satellite products for 2000–2022.
- Preprint
(6353 KB) - Metadata XML
-
Supplement
(1798 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-3736', Anonymous Referee #1, 03 Feb 2025
This manuscript uses homogenized ozone data from ozonesondes, IAGOS, FTIR, Lidar, and Umkehr instruments to analyze global trends in total, free, and lower tropospheric ozone. The authors compared different regressions and explored reasons for trend differences among stations with co-located instruments. Overall, this manuscript is excellent, and these ozone datasets and analyses are critically needed in the ozone community. I appreciate the authors’ careful consideration and handling of sparse and complex ozone data. I have only minor comments.
Page 2, Line 72-73: There is a reference to “Ref to Elementa collection”. Please fix this so it is a real, traceable citation.
Page 5, Line 154: Are there any results from the ASOPOS WMO GAW report that influenced how the ozonesonde data was handled?
Page 7, Line 175: I have some concerns about calculating monthly averages from locations where only one or two ozonesonde measurements are available within a given month. Previous literature suggests that somewhere between 3-18 observations per month are needed for accurate and representative time series (e.g., Christiansen et al., 2022; Lu et al., 2019; Chang et al., 2020; Wang et al., 2022). Could the authors perform some kind of short analysis that compares trends from once- or twice-monthly samples to trends derived from more frequently sampled sites? One thought could be to use a site that has many observations each month, then randomly select two observations to use from each month. Would the trends be similar to those derived from the full dataset? At the least, a discussion of the uncertainty involved in using these very sparse ozonesonde datasets is warranted.
Page 8: The IAGOS profiles are integrated so that the concentration is also reported in DU. Why is this not also done for ozonesondes?
Section 2.5: While the inclusion of Lidar data is desirable, as it summarizes nighttime trends, I am not sure it is appropriate to compare those nighttime trends to other instruments that measure during daylight hours. Nighttime ozone trends are known to be different from daytime (e.g., Yan et al., 2018). Perhaps the authors could discuss those nighttime trends as a separate section without comparison to daytime observations. Does the fact that these are nighttime measurements or some other aspect of Lidar sensing (e.g., sensitivity to the lower atmosphere, the filling in of missing data using models) explain the biases identified in Section 4.1.1 when Lidars are compared to the other instruments?
Figures:
Figure 6. The legend in the upper left corner is difficult to use. A few suggestions: 1) box off the legend so it does not appear to be another data point, and 2) provide more than one length/concentration for an easier visual reference.
References:
Chang, K.-L., Cooper, O. R., Gaudel, A., Petropavlovskikh, I., and Thouret, V.: Statistical regularization for trend detection: an integrated approach for detecting long-term trends from sparse tropospheric ozone profiles, Atmos. Chem. Phys., 20, 9915–9938, https://doi.org/10.5194/acp-20-9915-2020, 2020.
Christiansen, A., Mickley, L. J., Liu, J., Oman, L. D., and Hu, L.: Multidecadal increases in global tropospheric ozone derived from ozonesonde and surface site observations: can models reproduce ozone trends?, Atmos. Chem. Phys., 22, 14751–14782, https://doi.org/10.5194/acp-22-14751-2022, 2022.
Lu, X., Zhang, L., Zhao, Y., Jacob, D. J., Hu, Y., Hu, L., Gao, M., Liu, X., Petropavlovskikh, I., McClure-Begley, A., and Querel, R.: Surface and tropospheric ozone trends in the Southern Hemisphere since 1990: possible linkages to poleward expansion of the Hadley circulation, Sci. Bull., 64, 400–409, https://doi.org/10.1016/j.scib.2018.12.021, 2019.
Wang, H., Lu, X., Jacob, D. J., Cooper, O. R., Chang, K.-L., Li, K., Gao, M., Liu, Y., Sheng, B., Wu, K., Wu, T., Zhang, J., Sauvage, B., Nédélec, P., Blot, R., and Fan, S.: Global tropospheric ozone trends, attributions, and radiative impacts in 1995–2017: an integrated analysis using aircraft (IAGOS) observations, ozonesonde, and multi-decadal chemical model simulations, Atmos. Chem. Phys., 22, 13753–13782, https://doi.org/10.5194/acp-22-13753-2022, 2022.
Yan, Y., Lin, J., and He, C.: Ozone trends over the United States at different times of day, Atmos. Chem. Phys., 18, 1185–1202, https://doi.org/10.5194/acp-18-1185-2018, 2018.
Citation: https://doi.org/10.5194/egusphere-2024-3736-RC1 -
RC2: 'Comment on egusphere-2024-3736', Anonymous Referee #2, 17 Feb 2025
Review of manuscript entitled “Global Ground-based Tropospheric Ozone Measurements: Reference Data and Individual Site Trends (2000-2022) from the TOAR-II/HEGIFTOM Project by R. Van Malderen et al.
This manuscript describes the homogenized 2000-2022 tropospheric ozone records (column, free tropospheric and lower tropospheric partial columns) archived as part of the TOAR-II/HEGIFTOM project. Data are from five ground-based networks; measurements from each network included in HEGIFTOM have be reprocessed using standardized procedures and records include requisite uncertainty estimates. The focus of this work is intercomparisons of mean values, seasonal cycles and trends among the many individual instrument records to demonstrate the usefulness and limitations of the data for direct analysis as well as a reference for model and satellite-based tropospheric ozone records.
The manuscript is very well organized, especially given the difficulties intercomparing so many individual records with different long-term and daily sampling, station distribution, instrument type and related vertical resolution, uncertainty, etc.. The number of tables and figures are substantial but also very well organized, and appropriate for this type of comprehensive review of a large collaborative project and resulting data archive. The analysis is very thorough, the authors use/compare several statistical approaches, suggest explanations for resulting biases and trends, and include relevant references throughout the manuscript.
I have only a few comments, and otherwise editorial corrections and suggestions. I recommend publication after these minor comments are addressed.Comments:
Section 4.1.4: The authors present an analysis of the seasonal cycle changes over the record and discuss both the impact of Covid-19 and Bowman et al. (2022) results. Although not relevant to the trends, I think it would be very interesting to also compute the seasonal cycle over several years pre-Covid (say 2014-2019, which is 6 years matching 2000-2005) to see how this differs from the seasonal cycle change including the Covid years. One would assume this pre-Covid time period is a better comparison in reference to the Bowman et al. results.
L551-555: In reference to the Law et al. “dipole effect” did the authors analyze the monthly trends in the Arctic ozonesonde records as done for other stations in Figure 17? It seems they should be able to further comment on the seasonal variation in the trend. I am not familiar with the study but note they say the “dipole effect” in the vertical tropospheric ozone. If by “vertical” it means this feature varies with altitude it may be more difficult to resolve from the partial columns but I was curious if this could be directly checked.
Section 4.2.2: I did get a bit confused reading this section and trying to understand all the permutations. Most notably, the last two sentences seem to say the same thing. For Natal TrOC and LTOC “have increases” over the period but the FTOC increase in greater. The next example for several European stations says the same but “now with positive LTOC rates” but Natal LTOC was also increasing, so I do not see what distinguishes these cases.
Also in Line 588, I am not sure what the “least negative relative LTOC trends” are referring to. Is this saying that the FTOC is more sensitive than the LTOC due to mid-tropospheric/lower stratospheric dynamics?
Line 759 (and L520): the authors mention here in the conclusion that multiple observations per day can exist for several instruments (looking back I see this was well covered, I just missed the details as I was reading through). I see the Daily Mean (L2) used in the bias computations. For the QR analysis using all measurements, does this include multiple daily measurements, or are the L2 averages used? Also, when discussing the sample numbers within the month (around L520) does the sample count include multiple profiles in a day? This would make the monthly sampling problem even more of an issue, if say the 12 Umkehr samples per month actually occur on fewer than 12 days. If the sample numbers (SN) are notably different than the number of days typically sampled in the month, that would be useful to note.Figures:
The Figures are understandable complex, but I have a couple of suggestions to consider.
Figures 2 and 3, showing the time series and mean value, it is difficult to see the mean value dashed line in the figures, but since they are constant they could be extended into the white space (i.e. to 2025) then the reader will be able to see the individual dashed lines for easier comparison.
Figure 7 and similar: It is difficult to see the station names, although this may be the best that can be done. Did the authors try listing the station names along the plot axes, for example horizontally on the right hand side in panel a (would have to shift the legend), and vertically along the top axis in panel b? Or possibly where there is overlap, some stations are listed on the right/left or top/bottom so they all can be read.
Time Series Plots: For stations with extended time gaps I suggest removing the line over the gaps, for example in Figures 3 a+b and 15 a+b. Many of the longer lines connecting points with large gaps distract from the pertinent results of the plot, and the fact of the limited coverage is still clear because the color is missing in the gaps.Minor Editorial Comments/Suggestions
L55: depending on site; (3)
L71-73: “In the first phase …” This sentence is incomplete, or maybe it is supposed to be a clause, in which case the period before Gaudel et al. (2018) should be a comma.
L77: remove comma after 2019
L85: comma after De Maziere et al., 2018)
L121: and lidar
L136-137: suggest “… in electrochemical cells (ECC). Known as the ECC sonde, this type is used in the HEGIFTOM analyses…”
L152: suggest “… in a WMO/GAW Report (Report 201 by Smith et al., 2014).”
L159: suggest “(i) Removing all known inhomogeneities … ; (ii) ensuring consistency …; and (iii) providing …”
L230-231: (Bjorklund et al., 2023; Gordon et al., 2022) (correct punctuation after et al)
L234: having -> have
L235: suggest continuously -> commonly or consistently
L255: relevant -> relative
L264-265: the minimum -> a minimum
L273: 30-m is that 30 meters? If so, 30 m and 2 km.
L283: introduce XO3 (I didn’t see it before)
L349: within a region (delete -)
L415: 700>p>300 hPa
L448: Suggest removing “?” from section title
L463: (Fig. 6b)
L501: suggest “with the sign of the Umkehr trend at some collocated sites differing from the other instrument(s).
L521: suggest … only 3 airports with sufficient coverage to compute trends, the sample …
L521-522: … most divergent: ATL and DAL have only … ( note remove extra period after : )
L 524: … or Chang et al. (2004) …
L526: I’m not sure what “On the other hand” is referring to here. It seems the text before is describing the potential complications of the differing monthly sampling and that the different uncertainties by network type similarly make intercomparisons different. If this is correct, I would suggest “In addition, the different techniques … “
L526: … with mean values of: …
L 527: comma after IAGOS rather than period
L621: The wording here was just a little confusing to me, maybe “There is a trend reduction for all but one Arctic site (Churchill ozonesondes) and for all but one North American site (IAGOS Dallas).”
L634: suggest “Differences due to instrument technique”
L641: that -> which
L654: details -> detail
L691-692: I think the sentence “This adds extra information …” could be removed here.
L694: suggest “the trend estimates are robust across statistical methods and the DLM results complement the previously reported results.”
L696: Figure 15. (remove : )
Line 700: in function of -> as a function of
L706: A reference to the ozonesonde “total ozone drop off” might be useful here. If the Hilo tropospheric column is impacted as identified though intercomparison with other instruments at the same station location, do we expect other ozonesonde stations which experienced the drop off to be similarly impacted even though it is difficult to quantify due to lack of co-located data, or is Hilo thought to be a special case?
Line 734: In Appendix A …
Line 750-751: suggest slight rewording so that the main point is that the data are available rather than the data include uncertainties, maybe “The HEGIFTOM data and associated uncertainties, covering more than 350 individual datasets, are available via http: … “
L782: the word sparse confused me at first because it made me think of the sparse data issue (probably just me) but maybe consider using sporadic or intermittent, such as “to highlight intermittent periods over which the trend is significant, where trends estimated with the traditional QR and MLR methods to not show any significance.”
L805: “in the sense North America … “ (remove “the”)
L815: remove “will”
L846: change period to comma after “stations”Citation: https://doi.org/10.5194/egusphere-2024-3736-RC2
Data sets
HEGIFTOM homogenized ozone profile and TrOC datasets R. Van Malderen et al. http://hegiftom.meteo.be
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 242 | 32 | 10 | 284 | 21 | 8 | 5 |
- HTML: 242
- PDF: 32
- XML: 10
- Total: 284
- Supplement: 21
- BibTeX: 8
- EndNote: 5
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 93 | 33 |
| Belgium | 2 | 27 | 9 |
| China | 3 | 23 | 8 |
| Germany | 4 | 22 | 7 |
| France | 5 | 17 | 6 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 93