the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Jun 2026
Cleaner air, drier land: the unintended drought consequences of near-term climate forcers mitigation
Abstract. Drought poses severe threats to water resources, ecosystems, and socioeconomic well-being worldwide. While greenhouse gas (GHG) emissions are the primary driver of global warming and the associated intensification of the hydrological cycle, the role of non-methane near-term climate forcers (NTCFs) — encompassing aerosols and their precursors, ozone-forming reactive gases, and other short-lived species — in modulating regional drought risk remains poorly constrained, particularly in arid and semi-arid regions where aerosol-induced radiative effects and circulation feedbacks interact non-linearly with the water balance. Here we employ seven Earth System Models from the CMIP6 AerChemMIP framework to quantify the contribution of NTCF emissions to drought evolution under two future pathways that share identical GHG forcing but differ in the stringency of NTCF controls: SSP3-7.0 and SSP3-7.0-lowNTCF. Using the Standardised Precipitation Evapotranspiration Index at 3-, 6-, and 12-month timescales, we characterise projected changes in drought frequency, duration, intensity, and severity by mid-21st century and identify the large-scale dynamical mechanisms driving regional responses. NTCF mitigation exerts negligible influence on global-mean drought tendency but produces strongly heterogeneous regional responses, with drought conditions worsening substantially across the Sahel, West Asia, Central Asia, and the Mediterranean, while Southeast Asia, Australia, and Central America experience significant drought reductions. These findings are robust across the majority of models and highlight that air quality policies carry regionally differentiated and sometimes counterintuitive hydrological consequences that must be considered alongside their climate and health co-benefits.
- Preprint
(4454 KB) - Metadata XML
-
Supplement
(4249 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-3416', Anonymous Referee #1, 13 Jul 2026
-
RC2: 'Comment on egusphere-2026-3416', Anonymous Referee #2, 22 Sep 2026
The study addresses an important and policy-relevant question, and the multi-model paired-scenario framework has considerable potential. However, the present analysis does not yet provide sufficient physical or methodological support for its central conclusions. In particular, (1) the use of Hargreaves–Samani PET does not match this study on aerosol-induced radiative perturbations, (2) the relative contributions of aerosols and other NTCF species is not quantified, and (3) the mechanism diagnostics do not directly establish the claimed causal mechanism. Substantial restructuring and clarification of the manuscript are also required.
Major concerns
1. The emphasis on the full suite of NTCFs is not sufficiently justified
The manuscript consistently frames the projected hydroclimatic responses as consequences of non-methane near-term climate-forcer mitigation. However, the dominant forcing responsible for these responses is likely associated with reductions in anthropogenic aerosols, particularly through changes in radiative forcing, surface temperature, atmospheric stability and regional circulation. The contributions of other NTCF species or precursors are not separately quantified.
Because SSP3-7.0-lowNTCF simultaneously changes multiple aerosol species and reactive gases, the current experimental design identifies only the combined response to NTCF mitigation. It does not demonstrate which species dominate the drought response. The authors should therefore either provide quantitative evidence separating aerosol and non-aerosol contributions or substantially revise the framing. If aerosols are indeed the dominant driver, the title, abstract and discussion should reflect this more explicitly. Conversely, if the authors wish to retain the broader NTCF framing, they need to explain and, where possible, quantify the roles of the other included species.
2. The PET method is not well suited to assessing the hydroclimatic effects of aerosol mitigation
A central methodological concern is the use of the Hargreaves–Samani method to estimate potential evapotranspiration. This method represents PET primarily through mean temperature and the diurnal temperature range, together with extraterrestrial radiation. It does not explicitly account for simulated changes in surface net radiation, humidity or wind speed.
This limitation of neglecting the radiation is particularly important for the present study, as it focuses on addressing how future aerosol reductions influence hydroclimatic evolutions. Particularly, many of the strongest projected drought responses occur over drylands, where surface energy balance and atmospheric evaporative demand can be highly sensitive to radiative perturbations. Aerosol reductions directly alter surface shortwave radiation through aerosol–radiation and aerosol–cloud interactions, and may also affect cloud cover, humidity, boundary-layer structure and near-surface winds. These processes are central to the question being investigated but are not explicitly represented by Hargreaves–Samani PET. Consequently, the inferred importance of warming-driven PET changes—and the quantitative decomposition between precipitation and PET contributions—may depend strongly on the selected PET formulation.
A Penman–Monteith-type formulation would be more physically appropriate because it explicitly incorporates net radiation, vapour-pressure deficit and wind speed in addition to temperature. At minimum, the authors should repeat the principal SPEI and attribution analyses using Penman–Monteith PET and assess the sensitivity of the main conclusions to the PET formulation. Without such an analysis, confidence in the reported PET-driven drying remains limited.
3. The mechanism analysis does not adequately support the central attribution
The circulation diagnostics—including streamfunction, velocity potential, divergent winds, wave-activity flux and Rossby-wave-source anomalies—provide evidence of atmospheric circulation patterns associated with the simulated hydroclimatic response. However, they remain largely diagnostic and do not by themselves establish that aerosol reductions, particularly those over South and East Asia, cause the remote drying over the Mediterranean, northern Africa, the Middle East and the western United States.
The manuscript’s central mechanism claim requires a more direct analysis of the forcing and energy pathways. In particular, the authors should examine the radiative and atmospheric energy-budget responses to NTCF mitigation, including, for instance:
- surface energy fluxes (shortwave and longwave radiation, sensible heat, etc.);
- surface and top-of-atmosphere net radiation;
- aerosol radiative effects and cloud-radiative effects;
- atmospheric radiative cooling and column energy convergence;
- moisture transport and convergence;
Such diagnostics would help establish how aerosol reductions alter surface heating, atmospheric diabatic heating, convection and the resulting circulation response. They would also help distinguish local thermodynamic effects from remotely forced dynamical changes.
The use of Penman–Monteith PET would provide an additional physically interpretable decomposition of changes in atmospheric evaporative demand into contributions from net radiation, temperature, humidity or vapour-pressure deficit, and near-surface wind speed. This would directly connect the aerosol-induced energy and meteorological responses to the inferred drought changes.
4. The manuscript is difficult to follow and insufficiently focused
The manuscript requires substantial revision for clarity, precision and focus. Many passages use lengthy and abstract expressions that obscure the underlying analytical steps. It is frequently difficult to distinguish among quantities that are directly calculated, mechanisms inferred from spatial correspondence, and processes that are only hypothesized. Terms such as “robust,” “physically coherent,” “drives,” “primary driver,” and “explains” are sometimes used without sufficient quantitative support.
The manuscript also devotes considerable space to describing the GHG-driven changes under SSP3-7.0, particularly in the initial figures and accompanying discussion. Some baseline context is necessary to distinguish the NTCF signal from the overall future climate response. However, the present treatment appears disproportionate to the stated objective, which is to determine how NTCF mitigation modifies future hydroclimate and drought. The authors should streamline the discussion of the common GHG-forced response and focus the main text more directly on the SSP3-7.0-lowNTCF minus SSP3-7.0 contrast. Additional background results could be moved to the Supplement.
A thorough language edit is also suggested. The revision should shorten long sentences, remove repetitive or generic statements, define analytical logic explicitly, etc.
Citation: https://doi.org/10.5194/egusphere-2026-3416-RC2 -
RC3: 'Comment on egusphere-2026-3416', Anonymous Referee #3, 01 Oct 2026
The authors look into the effects of high vs low mitigation of near-term climate forcers (NTCF; aerosols etc.) for projected future drought, globally, according to other factors (greenhouse gases etc) evolving according to SSP 3. They do this using simulations from CMIP6 (ScenarioMIP and AerChemMIP), comparing SSP370 with “SSP3-7.0-lowNTCF” that is NTCF@SSP1, for three ensemble members each for 7 GCM/ESMs at their native resolution. The authors compute the SPEI index with a reference period of 1985-2014 (also tas, precip, PET, and P-PET), and then quantify changes in 2036-2055 compared to 1995-2014. All data re regridded to 2.5x.2.5 degree from each models’ native resolution (Tab. S1), then ensemble means per model are taken, then a non-weighted MMM computed. Observational data -CRU TS at half-degree resolution for observations- are only used to compute the global-mean change observed so far. The authors show changes in terms of drought occurrence, duration, intensity, and severity. The authors find that the global mean effect of different NTCF scenarios id negligible, but that NTCF mitigation in most regions increases, in some regions decreases, drought risk, with high heteorogeneity. The additional drought risk due to NTCF mitigation is in already water-stressed and densely populated regions. The authors call for adaptation to drought alongside (needed) NTCF mitigation.
The paper has a nice introduction and a method section listing everything needed. Then the authors look at global-mean projected changes for tas, pr, SPEI; then have a mainly descriptive section on the difference between both scenarios in terms of SPEI by season and spatially resolved; then a somewhat lengthy part on water balance aspects via P-PET as one mechanism including reference to climate dynamics via wave activity fluxes; a description of how the drought characteristics change for SPEI-6 and SPEI12 in contrast to SPEI-3; a nice analysis on how drought-affected *area* changes depending on the NTCF evolution; useful summary plots on the drought (SPEI) characteristics varying across season and region; and a separate (why is this its own section?) part mixing the changes in drought as hazard in a climate risk framework with present-day population data as exposure variable.
I find the manuscript interesting and while only analysing one specific scenario, this analysis is comprehensive and covers a range of aspects. I have a couple of minor revision points.
Main points
* add the historical simulations from the same CMIP models to Fig. 1a). That adds credibility or caveats to the model projections – and if they do not align, that needs discussing! This is my main point.
* please use a map projection that is area-preserving! https://blog.prif.org/2026/09/16/a-fair-world-begins-with-a-fair-map-the-un-adopts-equal-earth-projection/
* I think you should use ssp3-projected population data additionally, can be an SI figure. Otherwise you underestimate changes, as you say, and ssp3 pop is for sure more likely to be the case than present-day pop
* do you have a sense of the robustness of your results to the reference period, and the exact periods used for the analysis?
* 5/7 models agreeing is quite a low measure of robustness, given 4/7 is if it’s split or random.
* how robust is the climate dynamics analysis? In my experience this can be quite sensitive to analysis choices, which make the results not very meaningful
L 60 climate change is also a SDG, isn't it? Sentence implies it isn’t
L 110 why does that mean it’s an upper bound?
L 225 unprecedented since when?
L 240 onward. There are lots of “marked” “pronounced” words. remove all of them. Also, ‘even more markedly’ further below. Distinct. Marked. Notable exception (why mention any exception if it’s not notable? Can remove ‘notable’). Striking is fine – that’s interesting (in line 500).
L 275 why are the differences ‘important’ (as you say)? Fill word - remove
L 280 0.5 months = roughly 2 weeks? Much easier.
L 320 environmental – do you mean climatic?
L 330 bears a marked resemblance -> resembles. There are more phrasings like this that make the text unnecessarily complicated.
L 565 reinforcing physical consistency – it’s not the consistency that is reinforced, but your sense of it being consistent
L 655 the call in the introduction – that’s the authors call expressed in the introduction? I’d hope that one addresses in the findings what one calls for the introduction.. clarify who else calls for this or change wording
L 670 where are the implications listed for adaptation? I can’t see any, let alone ‘direct’ ones
L 675 where is the paradox?
L 630 this result underscores a critical point: global ... obscure > this emphasises that global ... obscure (because critical for what? Same as ‘marked’)
L 680 who do you mean should be “explicitly attentive to vulnerability”? Adaptation happens locally/nationally, not much globally unless you mean multilateral adaptation funding. And if you write ‘explicitly attentive’ I would expect subnational differentiation between populations as vulnerabilities vary massively among populations (by age, income/wealth, housing, occupation, etc)
L 685 do you think holding co2+ch4 constant means rather over or underestimating the NTCF signal, and how big could the error made by this be – anything in the literature that may allow an indication on this?
L 690 it’s not scientists’ but societies’ role to say what future research *should* prioritise.
Fig. 1- the text is very small
Citation: https://doi.org/10.5194/egusphere-2026-3416-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 131 | 135 | 28 | 294 | 35 | 25 | 22 |
- HTML: 131
- PDF: 135
- XML: 28
- Total: 294
- Supplement: 35
- BibTeX: 25
- EndNote: 22
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of “Cleaner air, drier land: the unintended drought consequences of near-term climate forcers mitigation” by Zhou et al.
Summary
This paper uses AerChemMIP simulations to quantify the effects of NTCF mitigation on droughts (frequency, intensity, duration) using the SPEI. Although a global signal is absent, there are large and contrasting regional effects (drought conditions exacerbated in many regions, but mitigated in others). In particular, drought conditions worsen across the Sahel, West Asia, Central Asia and the Mediterranean whereas drought conditions improve across SE Asia, Central America and Australia. The authors then go on to discuss the associated mechanisms, emphasizing the role of large-scale atmospheric dynamics (e.g., remote regions with mid-tropospheric subsidence), while also presenting a population weighted analysis to quantify exposure to droughts. This paper nicely points out the contrasting regional impacts of NTCF mitigation on drought characteristics, which has not received much attention. As such, this paper is an important contribution. The paper is comprehensive, nicely organized and well written, with clear figures (outside of Fig 8). I only have a small number of minor comments.
Specific Comments
L62 “Mitigating NTCFs is therefore critical not only for climate…” --> but this accelerates warming?
L74. “Numerous studies have explored the climate impacts of NTCF mitigation, with most focusing on aerosol reductions” --> what about methane? There’s the Global Methane Pledge for example. Also, the cited Allen et al. 2021 paper uses the same AerChemMIP models and the ssp370-lowNTCFCH4 experiments to quantify the effects of methane mitigation alone.
Introduction. Are there past studies that have focused on the main topic of this paper, i.e., the effects of NTCFs or aerosols on drought? If yes, some brief information would be helpful. If not, then this further motivates the topic of this paper.
L111. “This provides an upper-bound estimate of NTCF-driven changes …” yes, and this is also related to the fact the baseline ssp370 experiment features (weak) increases in most NTCFs (as is noted later).
L214. “and the wetting signal over monsoon regions and high northern latitudes” --> this looks to be occurring over China, might want to note this here as a prominent case in point?
L284, “are already positive under SSP3-7.0, effectively compounding the GHG-driven drying” and similar “GHG-driven” verbiage. When discussing the ssp370 experiment above, you may want to explicitly say that this experiment is subsequently used to diagnose GHG-driven changes.
Figure 8 does well to present a lot of data in a concise way. However, some of the numbers are too small to read. It’s also hard to differentiate italics versus non-italics. I assume the color of each box is also indicative of the number, with warm colors indicating an increase and blue colors indicating a decrease. However, I’m also a bit confused by some of the numbers, which do not seem to match the color. For example, under the lower panel for annual mean changes, the second box for intensity is orange (indicating an increase) but the number is -0.05. There are many similar examples. I am either misinterpreting the figure, or there is an error?
Supp Table 1. Why only 7 models? There are more. For example, CESM2-WACCM should have the relevant simulations?