the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 15 Oct 2025
Global hotspots and mechanisms of extreme humid heat and air pollution co-occurrence
Abstract. Exposure to extreme humid heat and air pollution each represent significant, well-characterized environmental hazards to human health. But the questions of where, when, and why they may co-occur, and whether humid heat may exacerbate pollution relative to high temperatures alone, remain largely unexplored. Here, we identify regions worldwide where ozone (O3) or particulate matter (PM2.5) pollution tend to be higher during humid versus non-humid extreme heat – i.e., where increased moist heat stress tends to co-occur with increased pollution, revealing a compound hazard tendency – and characterize the meteorological and chemical drivers of this co-occurrence. We analyze 19 years of near-surface concentrations of ozone, PM2.5, and related species (NO2 and HCHO) in the Copernicus Atmosphere Monitoring Service global chemical reanalysis (CAMSRA), along with meteorological conditions from the European Centre for Medium-Range Weather Forecasts Reanalysis version 5 (ERA5). We find that the global hotspots of worsened pollution during humid heat overlap with several global hotspots of extreme humid heat itself, and include multiple densely-populated areas. Altogether, more of the global population experiences worsened air quality during humid heat (versus dry heat) than experiences cleaner air quality. Overall, we find that humid heat and pollution co-occurrence hotspots typically occur where (1) the near-surface background chemical makeup is more urban (higher NO2, lower HCHO), and (2) humid heat is associated with stagnation and suppressed boundary layer heights (as is common in areas that experience severe humid heat), such that the local meteorological drivers of extreme humid heat are also conducive to pollutant accumulation.
- Preprint
(30435 KB) - Metadata XML
-
Supplement
(46618 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2025-4874', Anonymous Referee #1, 05 Nov 2025
-
RC2: 'Comment on egusphere-2025-4874', Anonymous Referee #2, 08 Dec 2025
Review: Global hotspots and mechanisms of extreme humid heat and air pollution co-occurrence
General comments:
This study deals with intensive statistical investigation of how O3 and PM2.5 are associated with extreme dry/humid heat conditions at a global scale and for some selected regions/cities as foci. Research on joint air pollution and extreme climate is contemporary and active. The study also further looks to basic underlying meteorological and chemistry conditions and then point out certain overall pictures related to the issue reasonably. Although the approaches and tools applied seem basic or standard, the work is shown to take full advantage of them with centering on anomaly quantities. The findings and discussion are fine, some of which are potentially useful policy-wise regionally. Given the importance of the topic and the overall technical quality in and scientific merits gained from this study, its publication would be in my favor. Nonetheless, some concerns still exist and should be addressed adequately (in terms of justification, clarification, and improvement) properly, my comments and suggestions are listed below (with the major ones marked with asterisks ***).+++++
Specific comments:
Word choices: It is sometimes confusing when reading the following terms:
a. "extreme non-humid/dry heat day" vs "non-humid/dry day"
b. "extreme humid heat day" and "humid heat day".
c. "extreme humid/dry heat" vs "humid/dry extreme heat"
d. In the text/figures, would "with or without "extreme" for humid/dry heat be meant the same?
e. "hot-dry/humid days" appear in a number of subplots in Supplementary Material
Given many terms defined in the study, try to state them clearly and later avoid confusion by keeping the consistency throughout the manuscript.Writing style: Suggest
"daily-maximum temperature" >> "daily maximum temperature"
"daily-mean 2-meter dewpoint temperature" >> "daily mean 2-m dewpoint temperature" or "daily mean dewpoint temperature"
"daily-maximum O3" >> "daily maximum O3"
and so on.Check the consistency of abbreviations (of variables) throughout. For example, DEWPOINT vs T_d, BL Height vs BLH, etc.
Clarify a bit more the term "global land area" whether it means the full global extent or most of it? As seen in many figures, the global maps are limited or bounded to certain latitudes.
ABSTRACT and CONCLUSIONS: The term "makeup" is mentioned but a bit vague. Suggest to have it clarified or reworded.
Give the full names of chemical species at the first encounters in text (e.g., HCHO)The terms "background" meteorology and "background" chemistry" said throughout the manuscript are unclear or even misleading in the current context. I am not sure if the authors' intention is\ "underlying" or "driving". I would suggest define or elaborate on them more specifically or using an alternative term for clarity.
LINE 45: Suggest "particulate matter less than 2.5 μm in diameter (PM2.5)" >> "particulate matter less than or equal to 2.5 μm in diameter (PM2.5), to be consistent with WHO's.
LINE 70: Recheck "Particulate matter may be increased by humidity through hygroscopic growth of aerosol particles". I am not sure whether it should be said as hygroscopic growth or heterogeneous (i.e., aqueous phase) reactions or both. Note that hygroscopic growth may affect aerosol size but not as directly on dry-basis PM mass.
LINE 88: Suggest "at 2 meter elevation" >> "at 2 m (above ground level)"
LINE 89: Suggest to separate this phrase as a new sentence from its previous text: "with the 1981–2010 period also analyzed for Tmax and TWmax to calculate extreme heat thresholds (see below)"
LINE 90: The terms "(see below)" is written but no calculation info is shown below. Please check.
For the global maps presented throughout, perhaps the US state boundaries may be changed to country's for consistency, if possible.
As compliment to FIG 1, a map of cell-wise day count for dry heat days and that for humid heat day could be useful and given in Supplementary Information. Optionally, percentages (with respect to total days of the entire study period) may be used for cell-wise day counts.
*** SEC 2
- It is unclear about using meteorological variables from ERA5 (not directly from CAMSRA) given that the authors states that CAMSRSA also generates meteorological fields. State the reason(s). Are those from ERA5 and CAMSRA are comparable in values at least for the variables considered at a daily scale? If they are greatly different, would CAMSRA-simulated O3 and PM2.5 (driven by the model-internal meteorology) be well representative in view of their association with meteorology?
- The authors review the literature and state solar radiation as a relevant meteorological variable for air pollution. Why is it not included in the analysis.? ERA5 and/or CAMSRA could have global radiation available.
- Precipitation is known for a key factor of wet scavenging and surface-wetness deposition. Should precipitation be treated in the analysis (or perhaps applied as flag somehow during identifying heat days)?
SEC 2.1:
- Would approximate thicknesses of the lowest model layer in CAMSRA be given and mentioned?
- I admit that I have never been familiar with CAMSRA. As said in the manuscript, it incorporates the MACCity anthropogenic inventory. Is the MACCity inventory used in CAMSRA time-varying and corresponding to the entire study period, i.e., providing annual/monthly emissions reasonably? Given this study's focus on the hotspot regions where cities are within, I think that this point is important and should be discussed; otherwise pointed out or recommended in the last section of the manuscript.SEC 2.3:
- LINE 124: Suggest "their 95th percentile value" >> "their corresponding 95th percentiles"-
- LINE 136: Suggest as a new sentence" "with the linear trend in the anomalies’ daily time series over 2003–2021 removed" >> "The linear trend in the daily anomalies' time series over 2003–2021 is then removed to finally produce the daily anomalies for use."SEC 2.3 and FIG S7: The presented approach of identifying dry/humid heat days is quite good and logical, as explained in more detail in FIG S7 (and its caption). But I feel difficulty understanding the contents well given three cases shown and compared many variable I would suggest to organize the contents and, if possible, describe each case more clearly.
*** SEC 2.3: Initially, I followed the idea of the linear-trend removal on daily anomalies. After some later thoughts, I think that doing so needs more objectivity. To me, daily anomalies without trends removed can be representative of actual driving conditions (emissions/chemistry/weather). Trend removal is still reasonable for correlation because trends could affect or obscure underlying correlations. I think the authors may recheck the objectivity of each key data processing step, state it well, and re-compute when needed if any change.
SEC 3.1 (from LINE 175) and FIG 6: Would it help if listing the hotspot regions considered, as well as their associated countries and city locations (central and bounding lon/lat values), as a table in the main text or Supplementary Information for clarity. Also see "All cities (means)" in FIG 6, I assume the term "mean" is of only across all the cities in question (i.e., 8 for O3 and 7 for PM2.5). Perhaps, state this briefly in the figure caption.
*** SEC 3.2:
- Give citation(s) to the sentence "HCHO and NO2 are often used as proxies for total VOC reactivity and all NOx species (NO + NO2), respectively"
- I think CAMSRA also reportt NO (nitric oxide) as output, which can be summed with NO2 for NOX. Using NOx (as opposed to NO2 alone) may provide its chemical pool better than NO2 alone. Similarly, CAMSRA may produce other VOCs, some of which can also be used and combined with HCHO. It is not clear to me about the current proxy choices. I hope this could be more clarified.SEC 2.2: The 2nd paragraph is quite important and may be more clarified, as the authors focus on daily O3 and PM2.5 anomalies:
- Climatological seasonal cycles. Also, this study set it to a 1-month or 30-day scale, correct?
- A total of 6 US-Embassy sites considered (as hotspot)? How many such sites with monitored data are available and the reason of having just the 6 sites compared?
- State briefly about the adequacy or necessary treatment of the US-embassy data used (percent missing, outliers, ...)
- Lack of the definitions of "humid heat" and "non-humid heat" therein. Thus, referenced to the section where these terms are described/LINE 234: Suggest "higher during humid heat than dry heat days" >> "during humid heat days than dry heat days" for readability (also check other locations in the manuscript)
It would be helpful for FIGs 4 and 5 fully showing all subplots (as seen in Supplementary Information).
I am not sure, when mentioning figures (stored in Supplementary Information) from the main text, if one should say or attach the phrase "see Supplementary Information" next to where they are mentioned?
FIG 2:
- The dashed boxes (hotspot regions) are still not easy to see. Use a more distinct color.
- Would each dashed box be possibly named or numbered (as legends) in the maps? Some readers like me may not be much familiar with global regions/cities.
- Is it correct that the maps in a) and c) are of average over 2003-2021 (19 years) and that the histograms in b) and d) are based on daily values? Anyhow, I would suggest having them stated clearly in the figure caption.
- Suggest: The caption should be rearranged as humid first and add the word "level", i.e., "Difference in pollution level between --humid-- heat and --dry-- heat ...".
- For the histogram plots in b), given the x-axis using a linear scale (as opposed to log scale). add more minor linear ticks on the axis for clarity.SEC 3.2, I feel that it comes a bit out of nowhere for the discussion about urban and remote areas. Describe how these urban and remote areas are assigned cell-wise, directly using LAND COVER data or visually? If visually, ensure reliability.
FIG 6:
- Would the titles of b) and d) be "Across global O3/PM2.5 difference deciles" or "Across global O3/PM2.5 anomaly difference deciles"?
- Given many quantities shown in b) and d) (i.e., solid, dashed, and dotted lines), the subplots become too small to read. Suggest to rearrange them to be much larger or have them separate them as a new figure.For all figures, check if the font sizes used within the (sub)plots are too small, if graphics are blurred, and if the yellow lines when drawn are too hard to see. Suggest to improve the problems when found.
*** SEC 3.5:
Multiple linear regression include a large number of input variables (also including interaction terms). The regression framework presented is good but I am not sure whether collinearity and overfitting would arise and become a concern on the resulting models and interpretation (on regression coefficients and their statistical significance). In that case, stepwise regression and regression diagnostics may possibly be helpful to mitigate or remedy such concern.FIG 7c: Recheck the sentence "Tildes represent an interaction term between the left side and right side predictors; for example, the regression coefficient
for the correlation of wind speed with wet-bulb temperature overall is near-zero (WS ≈ 0), but becomes negative with increasing background NO2 (WS∼NO2 < 0) or decreasing HCHO (WS∼HCHO > 0).", it is not clear why saying "correlation of wind speed with wet-bulb temperature" for the term alone (ie., not interaction). Of note. the meanings of "joint variables (as interaction)" and "correlation" are typically different.+++++
Citation: https://doi.org/10.5194/egusphere-2025-4874-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 237 | 86 | 20 | 343 | 36 | 16 | 12 |
- HTML: 237
- PDF: 86
- XML: 20
- Total: 343
- Supplement: 36
- BibTeX: 16
- EndNote: 12
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript investigates an important and increasingly relevant issue, the co-occurrence of extreme humid heat and air pollution. The analysis relies heavily on spatial and temporal processing steps that lack sufficient validation, and key interpretations are not supported by robust statistical evidence or a nuanced understanding of the underlying mechanisms.
(1) The handling of different spatial resolutions between CAMSRA (0.75°) and ERA5 (0.25°) is problematic. Merely stating that "ERA5 grid cell at the center" of each CAMSRA cell is used is insufficient. This spatial mismatch introduces potential biases that need to be quantified and addressed more thoroughly. The method should be explained with consideration of how this choice affects results and whether it may bias the analysis toward certain regions.
(2) The use of 95th percentile for defining extreme heat days is reasonable but requires more justification. Alternative thresholds should be tested or discussed to demonstrate robustness of findings.
(3) The manuscript reports significant results based on data processing steps (e.g., compositing, anomaly calculations) that have not been adequately validated. The interpretation of statements like "increased humidity corresponds with worsened pollution" requires robust support. The authors should demonstrate the reliability of their methodology through independent validation or sensitivity testing.
(4) The use of a bootstrapping procedure for regional composites is positive, but its lack of application to the global-scale analysis is a significant oversight. Statistical significance should be consistently reported for all major findings to ensure the reliability and reproducibility of the results.
(5) The claim that “humid heat is associated with stagnation” is presented as a factual statement rather than a correlation. The authors must acknowledge the complex and often bidirectional relationship between humid heat and stagnation, drawing on existing literature on humid heat mechanisms. Stagnation creates humid heat, but humid heat also creates stagnation through various feedback processes.
(6) The suggestion of a causal link between background chemistry regimes and co-occurrence patterns lacks sufficient experimental design to support it. While the correlation between NO2/HCHO ratios and co-occurrence is interesting, it requires cautious interpretation. Correlation does not equal causation, and alternative explanations should be considered.
(7) Data and Method section requires substantial revision to address the issues outlined above, particularly regarding spatial resolution, threshold selection, and validation of processing steps. The results section should be revised to reflect the limitations of the methodology and avoid overstating the strength of the findings.