the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Jul 2026
Global atmospheric aerosol distributions and composition from the Earth's surface to the stratosphere
Abstract. Atmospheric aerosols play a crucial role in Earth’s climate system, yet their spatio-temporal distribution, particularly in the free troposphere (FT) and upper troposphere–lower stratosphere (UTLS), remains poorly constrained, a major source of uncertainty in estimates of aerosol radiative forcing. To address this, we perform ECHAM/MESSy Atmospheric Chemistry (EMAC) model simulations with a newly developed setup, bridging the tropospheric and stratospheric regimes. Model output is evaluated against a comprehensive suite of observations of aerosol mass, number concentrations, and optical properties, showing good agreement across vertical layers and most geographical regions. The evaluated simulations provide a unified description of global distributions of key aerosol species, their composition, and number concentrations from the Earth's surface to the stratosphere. Simulated aerosol mass exhibits a global minimum between 400 and 200 hPa, marking the transition between FT and UTLS, with particle numbers peaking at similar altitudes or slightly higher in the tropics. Primary particles contribute less than 3.5 % to aerosol mass in the stratospheric overworld up to 10 hPa, substantially less than suggested by previous modelling studies and in closer agreement with recent observations. Stratospheric aerosol mass is dominated by sulfate, with a notable contribution (~15 %) from secondary organic aerosol throughout the global lower stratosphere. This work provides new constraints on aerosol distributions in the FT and UTLS, which remain underrepresented in global modelling studies, and enables future research on aerosol-climate interactions in this critical atmospheric regime.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Atmospheric Chemistry and Physics.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(7346 KB) - Metadata XML
-
Supplement
(7574 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-3649', Anonymous Referee #1, 30 Jul 2026
-
RC2: 'Comment on egusphere-2026-3649', Anonymous Referee #2, 27 Aug 2026
General comments
This manuscript presents EMAC simulations of global aerosol mass, composition and number from the surface to 10 hPa using a new GMXe modal configuration ("TropStrat", TS) that unifies previously separate tropospheric and stratospheric setups, evaluated against an unusually broad observational suite and then used to describe global aerosol distributions across four atmospheric layers.
I regard this as a valuable contribution, for three reasons. First, the experimental design is clean: running all three modal configurations under QCTM conditions with ERA5 and QBO nudging isolates the effect of the size distribution from chemistry–dynamics feedback, and Fig. 7 demonstrates this convincingly. Second, the observational sampling protocol in Sect. 2.3 is exemplary; converting vacuum aerodynamic to volume-equivalent diameter, integrating the modal PSD over each instrument's size cut, and applying the HR-AMS transmission curve explicitly removes a class of error that has confused earlier comparisons in this domain. Third, the diagnosis of the stratospheric dust problem, which related to emit ~900 nm dust into a mode with median diameter near 200 nm effectively down-scales the particle, suppressing sedimentation and enhancing convective lofting.
My concerns are not with the direction of the results but with the strength of the quantitative claims relative to the documentation provided. The most consequential methodological change, the dust and sea salt emission size mapping, is described in one sentence with no numbers; the TS setup differs from its predecessors in several respects that are not separated; the term "primary particles," on which the abstract's most novel claim rests, is never defined; and the headline percentages carry no uncertainty. None of this calls the results themselves into question, and all of it can be resolved through documentation, clearer definitions, and more carefully bounded language rather than new simulations.
Main comments
- The species-to-mode emission mapping is not documented anywhere in the manuscript
The manuscript never states which species are emitted into which modes. Section 2.2 says only that "Sea salt and dust emission size distributions were adjusted to the new modal size distributions" (l. 198–199); Sect. 2.2.1 adds that "a substantial amount emitted directly into the small coarse mode" (l. 253); for the previous setups we learn only that dust was "partly emitted into the accumulation mode" (l. 597). The mode descriptions at l. 231–237 are qualitative. Nowhere is there a figure or table giving the emission mapping.
This is the paper's central methodological change, and the Conclusions attribute its most novel finding directly to it (l. 645). As current description, a reader cannot reproduce or assess it. I would ask for a single table for each mode in two setups:
- (a) which primary species are emitted directly into it (desert dust, sea salt, BC, POA, anthropogenic and biomass-burning primaries, aircraft emissions), and the fraction of each species' emitted mass and number so assigned;
- (b) whether each species enters the soluble mode, the insoluble mode, or both, and in what proportion;
Relatedly, the claim that dust and sea salt emissions "typically peak between 800 and 1000 nm" (l. 252–253) carries no citation and is the physical justification for where the new mode was placed; please clarify whether this refers to a number- or mass-weighted peak and provide a supporting reference.
- The TS setup embodies several simultaneous changes, and the attribution to emission size alone is inferred rather than demonstrated
From Table 1, TS differs from TROPO by the addition of the small coarse mode, the narrowing of the accumulation mode (1400 → 800 nm, σ 1.59 → 1.4), the raising of the coarse threshold (1400 → 3200 nm, σ 2.0 → 1.7), and the re-mapped emissions. Importantly, "evaporation of particles into smaller than the original modes was prohibited" (l. 207–209). This is a global modification to cloud-processed aerosol release that potentially, independently reduces fine-mode primary aerosol and its long-range transport. The authors recognize the confounding in the sea salt case, but the stratospheric result is attributed to emission size alone. More analysis and discussion are helpful here.
- "Primary particles" is undefined, and the model and observational definitions are not commensurable
The abstract states that "primary particles contribute less than 3.5% to aerosol mass in the stratospheric overworld... in closer agreement with recent observations" (l. 10–13), but the term is defined nowhere. From Sect. 2.2 and Fig. 9 I infer it means the summed Dust, BC, POA and sea salt tracer mass; on that reading the arithmetic checks out, since summing the four primary panels of the OW row in Fig. 9 gives a maximum near 3.3%. Two issues follow.
(a) Mass of primary material is not mass of primary particles. Modes are internally mixed, so a droplet nucleated on a dust core and grown by H₂SO₄ condensation is almost entirely secondary by mass while a single-particle instrument would classify it as dust-containing. Please state which quantity is reported.
(b) The Sect. 4 diagnostics appear to be raw modal output, whereas Sect. 3 applies the careful sampling of Sect. 2.3. This matters concretely: TS places dust largely in the 800–3200 nm mode, outside HR-AMS transmission and at the upper edge of the PALMS range. Please either apply the Sect. 2.3 sampling where a comparison to observations is claimed, or state that Sect. 4 reports unfiltered output and adjust the abstract. Minor but related: the species in Fig. 9 sum to ~100% per cell, indicating that particle-bound water is excluded. Given that the manuscript repeatedly describes deliquesced sulfate droplets, please state this convention explicitly.
- The 3.5% is mainly carried by POA, but the claim is all about dust
In the overworld row of Fig. 9, POA contributes 1.7–3.0%, BC ≤0.3%, dust ≤0.7% and sea salt <0.1%. Primary organic aerosol accounts for roughly four fifths of the primary mass; dust for well under a quarter. Sect. 5 (l. 596–605) and the Conclusions (l. 645) explain the low primary fraction entirely through the revised dust emission size. That mechanism addresses the smaller component, and it is not obvious that it affects POA transport at all, since POA is not emitted at the sizes that motivated the new mode. This is compounded by the authors' own statement that "these factors likely contribute to an overestimation of OA in the FT" (l. 594). Please discuss what controls POA transport into the overworld, how it differs between setups, and whether the known FT overestimate propagates.
- The tropical UTLS SOA fraction
Figure 9 shows SOA mass fractions of 63.6% (30°S–0°) and 43.7% (0–30°N) in the UTLS, against sulfate fractions near 23.5%. The abstract instead quotes ~15%, corresponding to the overworld row. A tropical UTLS that is SOA-dominated rather than sulfate-dominated is considerably consequential for heterogeneous chemistry, optical properties, and the common modelling assumption of a sulfuric acid–water aerosol. Discussion about its importance is absent. I assume the results depend heavily on ORACLE, which the authors note "does not account for the Kelvin effect and the kinetic limitations in condensation" (l. 588–589). Please give an uncertainty range on the reported SOA fractions.
- Three problems cluster in the southern extratropics
The manuscript documents underestimated SAOD at southern mid-to-high latitudes with the cause left open (l. 502–503, Fig. 7). Substantial underestimation of sulfate, ammonium, OA, BC and BL PNC over the Southern Ocean (Fig. 4 bottom row) and an annual-mean nitrate fraction of 54.2% in the southern polar overworld (Fig. 9, l. 538–539). Given the overlapping regions and that the first two are both consistent with missing aerosol mass, please discuss them jointly in Section 5.
Specific comments
- 2.2, l. 176–184. Please add an explicit definition of primary vs. secondary aerosol as used here, with the species list.
- Sect. 2.2.1, l. 231–237. The mode descriptions are qualitative. Please state, for each mode, which species it can contain and by which route they arrive — direct emission, nucleation, growth across the boundary from the mode below, inter-modal coagulation, or insoluble-to-soluble ageing. This would resolve much of M1 in a few sentences.
- Table 1. Please state the total number of modes per configuration (7, 7 and 9, if I understand correctly) and indicate which modes have insoluble counterparts.
- Table 2. The improvements in Na⁺ (M/O 17.59 → 6.08 → 3.0), Cl⁻ (30.16 → 10.09 → 7.55) and NO₃⁻ (4.77 → 3.05 → 1.96) are substantial and well presented. However sulfate is worse in TS than in TROPO on every metric (M/O 1.48 vs. 1.18; PF2 73 vs. 88; R²log 0.35 vs. 0.54), and total PM₂.₅ is essentially tied. The text at l. 313–316 reports the nitrate improvement but not this. Acknowledging it would strengthen the paper.
- Fig. 2. Chloride remains overestimated with R²log = 0.17 even in TS. Some discussion of why Cl⁻ specifically remains poor would be welcome.
- Fig. 4. The detection-limit band overlaps much of the upper-altitude comparison. Please state what fraction of UTLS observations lie below detection.
- Fig. 7. Please state whether omitting pyroconvective stratospheric injection (e.g. the unreproduced August 2017 event) affects the annual-mean POA and BC fractions in Sect. 4 and 5.
- Fig. 9. Consider adding a "primary total" panel so the 3.5% can be read directly rather than reconstructed from four panels.
Citation: https://doi.org/10.5194/egusphere-2026-3649-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 368 | 171 | 35 | 574 | 50 | 39 | 41 |
- HTML: 368
- PDF: 171
- XML: 35
- Total: 574
- Supplement: 50
- BibTeX: 39
- EndNote: 41
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript presents a new setup of the EMAC model that allows to seamlessly simulate tropospheric and stratospheric aerosols. Compared to the two standard EMAC versions (one for the troposphere and one for the stratosphere), this new version includes an additional “small coarse” mode for particles between 800 and 3200 nm diameter. The authors evaluate extensively the new model setup against air-, ground-, and satellite-borne observations.
This is an very good model evaluation paper. The new setup is generally well described, the comparison with observations is detailed and rigorous, the text is clear, and the figures are meaningful and very well thought out (I really appreciate how they deliver a lot of information while remaining clearly legible). If this paper was submitted to GMD, I would only have minor comments to improve the readability of the manuscript (see below). However, I don't think this paper is within the scope of ACP (“Articles should have important and clearly argued implications for our understanding of the state and behavior of the atmosphere and climate or present substantial new insights into the atmosphere's role in other parts of the Earth system.”).
General comment: Throughout the manuscript, the references to the relevant figures are generally at the beginning of the section that discuss them. Since each figure has many panels, it would be clearer if the panels were numbered and the text contained the reference to the panels where each of them is discussed. I often had trouble understanding which figure I was supposed to look at while reading the text.
L165: is half-year spin up enough for the stratosphere? If starting from a clear atmosphere, the spin up for the stratosphere is generally several years.
L177: I am a bit confused by the description of the aerosol speciation. Are the concentrations of sulfate, ammonium, nitrate, sodium, chloride, potassium, calcium, magnesium, organic aerosols (separated in 26 species), black carbon, dust, and sea salt prognostically calculated for each of the modes (5 soluble and 5 insoluble modes)? Are soluble species allowed in the insoluble modes to simulate coating?
L230: I think it should be specified that the past studies are only the ones with EMAC.
L250: which parameterization is used for sedimentation? Is it the same for all modes (just varying the radius)?
L265: I don’t understand what “respectively” refers to. Maybe a word is missing?
L291: I am not sure I understand what is done to compare with aircraft measurements. Shouldn’t the aircraft measurement (not the model) be binned on the model vertical levels? Is the vertical distance between consecutive aircraft measurements larger than one model level?
L315: I am surprised that both TROPO and STRATO overestimate NO3, and TS does it better. Is there an explanation for why?
L335: I don’t think MERRA2 is a good benchmark to compare speciation. In MERRA2, the total (not speciated) AOD is assimilated, but the speciation is calculated based on the one of the underlying forecast model (GEOS). If GEOS gets the speciation incorrectly, this error will remain after the assimilation adjust the AOD.
Table 2: generally, M/O is not bad, but R2 is quite low. Can the authors draw any conclusion from this? Could it be that compensating errors in different regions cancel each other, giving a good globally averaged M/O but making the R2 low? It would be interesting to see a map of M/O at each GHOST location: maybe it could show if there are systematic under/overestimation based on the aerosol species dominating PM and give indications on which processes are better simulated.
Fig 3: is “log” in base 10 or e? For clarity, I’d use ln if base e and log_10 if base 10.
L355: why is the pattern opposite in the comparison with MODIS vs AERONET? Is it because MODIS and AERONET are different even when they observe the same locations, or because they measure different locations?
L377 “low M/O … datasets.”: I am not sure I understand this. If concentrations are below the detection limit, are they considered zero or equal to the detection limit? Couldn’t the same detection limit be applied to the model output?
L386: how are biomass burning and volcanic emissions distributed in the vertical?
L486: BC is missing for StratoClim in Fig. 6. Was it not measured?
Fig A1: the vertical labels say PM, but only the sea salt component of PM is shown, correct? Also, I have the same comment as for L335. I don’t think comparing speciation to MERRA2 is useful (and in any case the label on the color calendar should not say “observed” but rather “reanalysis” or “MERRA2”). If GEOS has a bias in sea salt, this bias will remain after assimilating the AOD.