| 07 Oct 2026
Spatially uneven impacts of stratospheric aerosol injection on compound dry-heat wave events in the GeoMIP G6sulfur experiment
Abstract. Compound dry-heat wave (CDHW) events, characterized by the joint occurrence of extreme heat and dry conditions for at least three consecutive days, pose substantial risks to ecosystems, agriculture, and human societies. However, the implications of stratospheric aerosol injection (SAI) for such compound extremes, particularly at regional scales, remain poorly understood. Here, we investigate changes in CDHW frequency, duration, and severity under the GeoMIP G6sulfur scenario relative to the high-emission SSP5-8.5 scenario using simulations from three Earth system models. We find that SAI substantially offsets the projected intensification of CDHWs over global land, with marked reductions in their frequency, duration, and severity. However, this mitigation is highly heterogeneous: Central Africa emerges as a prominent regional hotspot where CDHWs become more persistent in all three models and, while their frequency also increases under the percentile-based dry-day criterion, primarily in MPI-ESM1-2-LR. The enhancement is driven mainly by the dry side of the compound: dry spells lengthen in all three models (by 0.2–0.9 days per event) and the number of hot-and-dry days per summer rises, even though the total number of heat wave days falls in two of the three models, whereas the intensity of the individual compound events is reduced, consistent with the overall cooling. Comparison with SSP2-4.5 scenario, which has a global-mean radiative forcing comparable to that G6sulfur, indicates that the enhanced CDHW response over Central Africa is not driven by the greenhouse gas forcing component alone. The reduction in greenhouse gas forcing alone produces little change in CDHW exposure over Central Africa, whereas SAI reduces surface shortwave radiation and evaporation over tropical land, promoting mid-tropospheric subsidence and reductions in cloud cover, precipitation, and soil moisture over the region. The contrasting response over the Amazon Basin, where CDHWs decrease, is largely explained by the greenhouse gas component. Reduced greenhouse gas forcing substantially alleviates the projected drying over the Amazon but provides no comparable benefit over Central Africa, where aerosol-specific drying dominates. These hydroclimatic changes are further coupled with declining soil moisture and reduced latent heat flux, partly offsetting the SAI- induced cooling and favoring persistent heat extremes. Our results demonstrate that reducing global mean warming through SAI does not necessarily translate into uniform reductions in compound climate extremes, highlighting the importance of distinguishing the removal of greenhouse gas forcing from aerosol-specific effects, and of land-atmosphere interactions when assessing the climate risks and regional trade-offs of solar climate intervention.
The authors used G6sulfur simulations to analyze the effect of stratospheric aerosol injection (SAI) on heatwave, dry event, and combined dry-heat wave event (CDHW). Temporal evolution and spatial distribution of these events are analyzed. It is found that SAI would decrease frequency, duration, and severity of CDHW over large part of the land area. However, Central Africa is identified as one hotspot region where SAI enhance the persistency of CDHW. This study makes a useful contribution to our understanding of how compound events might respond to SAI.
One major issue is that the discussion of G6sulfur vs. SSP2-4.5 is not clearly presented. One example is in the abstract (see below). Also, the hotspot region of Central Africa vs. Amazon is not clearly presented. In some places, I found the discussion of Amazon comes from nowhere.
Specific comments:
It seems the first half of the abstract focuses on the comparison between G6sulfur and SSP5-8.5, and the second half focuses on the comparison between G6sulfur and SSP2-4.5. However, the distinction is not clear. In particular, I found the discussion of comparison between G6sulfur and SSP2-4.5 is confusing and hard to understand (lines 32-43). I suggest the focuses should be on the net effect of SAI (G6sulfur minus SSP5-8.5). Also, Central Africa is focused in the comparison between G6sulfur and SSP5-8.5, but the Amazon region is included in the comparison between G6sulfur and SSP2-4.5. This is very confusing.
Lines 51-52: There are two ‘which’ in one sentence. Please rephrase.
Lines 76: The study of Seneviratne et al. 2021 is not on solar radiation modification. And, more references other than Weill et al., 2024 should be provided here.
More updated literatures should be cited for the first three paragraphs in the Introduction section, Currently, most studies cited are from more than 10 years ago. Extreme climate and solar radiation modification studies evolve fast, and more recent relevant studies should be noted and cited.
Lines 117-121: This sentence is a bit too long and hard to understand. Please rephrase.
Lines 122-123: Again, more updated relevant studies should be cited here.
Lines 149-152: ‘XX, which was not included in the present analysis’. Is UKESM1-0-LL included or not? It’s not clear from this sentence. Please rephrase.
Line 180: Table 1. I suggest put more info here. For example, model resolution and the way how aerosol forcing is prescribed. Info of specific member names such as r1i1p1f2 can be listed in the supporting material.
Lines 343-345: ‘The spatial maps therefore characterize the magnitude and geographical distribution of the end-of-century response, while the time series describe its temporal evolution.’
This kind of statement is self-evident and not necessary.
Lines 348-349: ‘Severity decreases with widespread negative anomalies over tropical and subtropical land region’. This sentence is not complete.
Line 357: Figure 5 are mentioned before Fig.3 and 4.
Line 442 and 444: Why Amazon here? Should it be Central Africa?
Lines 457-461； Please comment on why in some regions (e.g., central Africa), the response of heatwave frequency is in contrast to response of heatwave duration and severity.
Lines 501-504: Figure S2 shows CNRM model (Fig. S2a) simulate a decrease in heatwave frequency in Central Africa. This should be explicated mentioned and discussed.
Lines 619-620: ‘single-point injection experiments indicate that equatorial injection produces a weaker tropical circulation and ITCZ response than injection at off-equatorial latitude’. Please provide evidence for this statement.
Lines 586-622: How is the Hadley circulation change related to the CDHW enhancement in Central Africa? Please elaborate.
Line 687: It does not make sense to talk about ensemble mean heat flux if different model-simulated flux have different signs. Temperature change is determined by each model-simulated heat flux, and ensemble temperature change is not determined by the ensemble mean heat flux.
Lines 699-701: ‘favor the persistence or occurrence of heat extremes despite the reduction in absolute temperature’. I have trouble in understanding the reasoning here. I think it is important to provide a more in-depth discussion of SAI-induced changes in mean temperature, extreme temperature, and heatwave here.
Line 706: I am not sure what ‘a general increase in mean temperature’ refers to. Lines 693-694 states that Tmax decrease in two out of three models.
Lines 713-715: The anomalies of G6sulfur relative to SSP5-8.5 reflects only the effect of aerosol, and the anomalies of G6sulfur relative to SSP2-4.5 reflects the combined effect of enhanced GHG and aerosol.
Lines 738-739: I don’t understand what this statement means.
Line 746: should be ‘is consistent’
Line 756: To make it clear, please add ‘relative to SSP2-4.5’ before ‘the stratospheric aerosol layer diminishes ..’
Lines 788-789: This statement seems like regional warming is an established feature of SRM in general, but it will depend on SAI strategy. Please rephrase.
Lines 821-822: This statement and the following arguments are confusing. I think what is concerned here is the net effect of SAI and the underlying mechanims. That is, the effect of G6sulfur relative to SSP5-8.5, but not relative to SSP2-4.5.
Lines 829-830: ‘the removal of the greenhouse gas induced response under G6sulfur relative to SSP5-8.5 XX’.
This is not correct. G6sulfur is applied under SSP5-8.5, and there is no removal of GHG under G6sulfur relative to SSP5-8.5
Figure 1：Each panel has too many lines, making it a bit hard to tell which is which. For example, ‘Thick curves are 11-year running means and the faint curves behind them are the annual values. Dotted lines are least-squares linear trends 299 over 2020–2100.’. It’s not easy to distinguish these lines. Also, please be clear what Tmax here exactly means. Annual maximum value of daily maximum temperature or just daily time series of daily temperature?