the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 01 Oct 2025
Global escalation of more frequent and intense compound heatwave-extreme precipitation events
Abstract. Compound heatwave-extreme precipitation (CHWEP) events, the rapid succession of heatwaves and extreme precipitation, pose growing compound and cascading risks. However, global-scale comparisons of their spatiotemporal evolution against single extremes remain limited. This study systematically examines the changes in CHWEP and corresponding single extremes from 1980 to 2100 using climate observations and projections under SSP (Shared Socioeconomic Pathway) 2–4.5 and SSP5-8.5 scenarios. We find that CHWEP exhibit higher frequency, stronger precipitation, and longer heatwave duration in mid-to-high latitudes, while tropical CHWEP feature more intense heatwaves than single heatwave events. These spatial contrasts persist in future projections. Under both scenarios, CHWEP and single extreme metrics intensify globally by 2056–2100, with post-heatwave precipitation exceeding that of single precipitation extremes, particularly under SSP5-8.5, highlighting sensitivity to greenhouse forcing. Critically, the co-occurrence is non-random, indicating an emerging physical linkage. In the tropics, the likelihood of extreme rainfall following heatwaves increases markedly. Our findings demonstrate that CHWEPs are evolving into a distinct, intensifying hazard class, necessitating their integration into climate resilience, early warning, and adaptation frameworks.
- Preprint
(10275 KB) - Metadata XML
-
Supplement
(7530 KB) - BibTeX
- EndNote
Status: open (until 11 Dec 2025)
- RC1: 'Comment on egusphere-2025-4289', Anonymous Referee #1, 23 Oct 2025 reply
-
CC1: 'Comment on egusphere-2025-4289', Marta Moreels, 15 Nov 2025
reply
The method of this study raises some questions for me. They use three reanalysis datasets
(ERA5, MERRA-2, JRA-55) as historical data and use four CMIP6 models under two SSP
scenarios for the future. Firstly, they address the systematic biases between climate model
simulations and reanalysis data. They highlight that they performed a simple bias correction
on the CMIP6 model output although, they note that, since they are comparing CHWEP
events and single extreme events, the precise accuracy of the absolute value is not
important for the conclusion. However, this is too short of an explanation in my opinion.
Limitations of the bias correction method and explaining it shortly would be more complete.Citation: https://doi.org/10.5194/egusphere-2025-4289-CC1 -
RC2: 'Comment on egusphere-2025-4289', Anonymous Referee #2, 28 Nov 2025
reply
This study analyzes historical and future changes in compound heatwave–extreme precipitation events (CHWEPs) and compares them with single extreme events at the global scale from 1980–2100. They used three reanalysis datasets as observations and four CMIP6 models under two SSP scenarios (SSP2-4.5 and SSP5-8.5). They found that CHWEPs exhibit higher frequency, stronger precipitation, and longer heatwave duration in mid-to-high latitudes, and that CHWEPs occurring in the tropics feature more intense heatwaves than single heatwave events. They also suggest that these patterns will persist in future projections. In addition, they found that although both CHWEPs and single extreme events increase across most of the globe in the future, extreme precipitation following heatwaves will become notably more intense and frequent.
I agree that this is a very important topic in the field, but I find the manuscript still quite far from the standard typically expected for HESS. The choices of observational and model data are not reasonably justified, the statistical techniques are not analyzed in a way that convinces me of the robustness of the results, and the results/discussion remain largely descriptive and shallow (not quantitatively or statistically convincing). Many of the figures repeat similar information without adding new insights, and some key quantities are not clearly defined. In my view, the authors will need extensive methodological revision, additional testing, and substantial rewriting before the paper can be considered for publication in HESS. Below I list some of my concerns for the authors to address (with a few general examples):
-
The rationale for choosing ERA5, MERRA-2, and JRA-55 for observations is unclear. JRA-55 is also an older dataset, and it has already been replaced by JRA-3Q.
-
Line 91: It is unclear and confusing what “ensemble mean” for observational data refers to here. The three reanalysis datasets are deterministic and have different systematic biases, so combining them as an ensemble requires a more explicit explanation.
-
The paper uses only four GCMs from CMIP6 without explaining why they were selected over others.
-
In Section 3.1 (Identification and characteristics of CHWEP events), when defining heatwaves and extreme precipitation events, it would help to mention the actual threshold values being used.
-
The definitions of formulas (1)–(8) are not very clear. Some metrics are calculated across all events while others are computed per event, and this is confusing. These formulas are difficult to interpret and need clearer and more accurate notation and description.
-
Some notations and units in the figures are not clear. For example, in Fig. 2a/b, what is the meaning of the units of “time”? Do you mean the total number of events over 1980–2024 or the number of events per year?
-
Some discussions of the results are insufficient or inaccurate. For example, in the discussion of Fig. 3c at line 181 (“Across most mid-to-high latitude areas, the extreme precipitation intensity during CHWEPs exceeds that of single extreme precipitation events”), the authors only focus on the Northern Hemisphere and ignore the Southern Hemisphere mid-high latitudes.
-
Much of the Results section is purely descriptive. The analysis could be more quantitative, for example by providing regional averages or more meaningful values to make the study deeper and more informative.
-
Some figures need to be modified and improved. For example, the colors for “Future SSP2-4.5 single” and “Future SSP5-8.5 single” in Fig. 5 are too similar to distinguish clearly.
-
Although the authors introduce the Wilcoxon rank-sum test and event coincidence analysis in their Methods section, no p-values or other statistical results are shown. Without reporting these outcomes, the reader cannot assess whether the stated differences or assumptions are statistically meaningful.
Citation: https://doi.org/10.5194/egusphere-2025-4289-RC2 -
-
RC3: 'Comment on egusphere-2025-4289', Anonymous Referee #3, 30 Nov 2025
reply
General comments
This study explores historical and future changes in the characteristics of compound heatwave-extreme precipitation events and the statistical linkage between heatwave and extreme precipitation using global reanalyses products and CMIP6 model projections. This topic is relevant given rising climate-driven extremes and the increasing importance of understanding compound hazards. The manuscript is clearly structured and the analyses are potentially valuable. However, in its current form, the analytical framework remains somewhat superficial. The study would benefit from more explicit justification of key methodological decisions, clearer articulation of its scientific novelty, and additional explanation to strengthen transparency and interpretability. Several assumptions are stated but not sufficiently supported, and important methodological steps lack the detail needed to ensure reproducibility. Addressing these issues would substantially improve the clarity, credibility, and overall impact of the work. With these enhancements, the study could make a meaningful contribution to understanding CHWEP events, but substantial conceptual, methodological, and interpretive clarifications are required before the conclusions can be fully supported and before the manuscript meets the standards expected for publication in HESS.
Specific comments
1. Your description of preprocessing of datasets is straightforward. However, I am concerned about the motivation for using ensemble mean as the final reference datasets. Given the uncertainties inherent in global reanalyses, the rationale for preferring ensemble mean over the best-performing reanalyses should be well articulated. Similarly, what is the justification for selecting four specific CMIP6 models? Is this based on their transient climate responses to avoid ‘Hot Model’ problems associated with CMIP6 (e.g. Hausfather et al., 2022)?
Reference
Hausfather, Z., Marvel, K., Schmidt, G.A., Nielsen-Gammon, J.W., Zelinka, M., 2022. Climate simulations: recognize the ‘hot model’ problem. Nature 605, 26–29. https://doi.org/10.1038/d41586-022-01192-2
2. The quantile mapping approach is mentioned only superficially, it is unclear whether temperature and precipitation were corrected separately, future projections were adjusted using historical distributions, and potential risks of overcorrection, particularly for extremes?. Additionally, since CHWEP events potentially rely on joint behavior of temperature and precipitation, it might be helpful if the authors note whether the bias correction approach preserves multivariate dependence and temporal structure. Quantile mapping can distort these if applied independently.
3. “Historical daily maximum temperatures” is vague. Please clarify the baseline climate period used to calculate threshold, and whether the chosen period affects the comparability between historical and future events?. The authors should add a bit more context (e.g. land-atmosphere memory, soil moisture decay timescales) to support the choice of 7-day window. Also clarify whether intensities represent mean values per event or aggregated means across all events.
4. Please clarify whether ECA uses events derived from the earlier definitions or whether counts are based on individual days. ECA may be sensitive to the way the underlying event is defined. In addition, consider clarifying ECA only tests for statistical dependence, not physical causality. The text currently states it can “identify possible causal relationships,..” which may be an overstatement.
5. A brief mechanistic explanation why some patterns (e.g. “hotspots in the Sahara Desert, the Middle East, and Australia”, “CHWEPs are more pronounced in mid-to-high latitudes”) emerge or what they imply would strengthen the narrative. Statements such as “CHWEPs occur more frequently” or “intensity is significantly higher” would benefit from at least approximate magnitudes (e.g., percent differences)
6. The statement that “CHWEP precipitation intensity under SSP2-4.5 exceeds single-event intensity under SSP5-8.5” is interesting, please provide a short explanation e.g. heatwave-induced atmospheric instability?
7. The authors mentioned widely known ideas (e.g., compound events cause more damage than single events) without tying them closely to your specific results or demonstrating how your findings confirm, extend, or challenge this existing knowledge. Also you state that integrated intensities exceed those of single events, but do not explain what this means physically (e.g. more moisture availability? stronger thermal anomalies?). A more explicit comparison would help interpret the significance of CHWEP intensification.
8. ‘..both CHWEP and single events are increasing significantly in frequency, intensity, and duration’, it is unclear which regions or magnitude of change? More details should be added to make the discussion more insightful.
9. ‘..but also reveals a dangerous "rapid transition" phenomenon…’ This is one novel contribution that should be elaborated (currently the discussion is brief).
10. This section is missing the acknowledgment of potential limitations of the methods/results (e.g. uncertainties in climate data, bias correction, sensitivity to thresholds). This is important to strengthen credibility and transparency.
Minor comments
L50-55: It will be useful to connect which category CHWEP belongs to (e.g. temporally compounding). This will help readers who may not be familiar with the terminology.
L70-75: I would be careful with the use of the phrase ‘ the first comprehensive, global scale’. Ensure your phrase is defensible or rephrase to avoid overclaiming.
L75-80: The authors mentioned ‘statistical significance of differences among event types’, however, it is unclear whether this pertains to frequency, duration, temporal lag, etc.?. As I mentioned previously, the authors should consider clarifying that CHWEP events are defined both in terms of temporal thresholds and sequence, if that is indeed the approach.
L85: Use standard citations instead of hyperlinks throughout the texts. Links should be provided in the appropriate ‘data availability’ section
Fig 1: Caption is too short. Consider adding more information.
Citation: https://doi.org/10.5194/egusphere-2025-4289-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 293 | 103 | 19 | 415 | 38 | 10 | 10 |
- HTML: 293
- PDF: 103
- XML: 19
- Total: 415
- Supplement: 38
- BibTeX: 10
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The authors present an analysis of compound heatwave extreme precipitation (CHWEP) events using statistics computed an ensemble of 3 reanalysis data (in the past) and 4 CMIP models (for projection). They first show the differences on several statistics between CHWEP and single heat waves (SH). Then, they show how these statistics vary in the future, under several SSPs.
Overall, I found that this analysis suffers from several methodological issues that are detailed below. The paper is not very well written, with important quantities not properly defined, equations I could make sense of, etc. In my opinion, this work cannot be accepted for publication in HESS, unless completely reworked.
Main comments:
1. Currently, the authors compute the statistics in the past on reanalysis and in the future on climate models. It is well known that climate models are highly biased if not corrected with bias correction methods (see eg François et al., 2020). Is it the case here?
If the answer is 'no', then I don't see how it is possible to draw any definitive conclusions when comparing the future (using CMIP models) and the past (with reanalysis ) as done in Sections 4.2 and 4.3 (Figures 6 - 10 and S3 - S4). If the answer is 'yes', the authors should nevertheless first check that the WRT does not reject the equality of the distribution when the statistics are computed on the models during the past periods. This is a preliminary study that is absolutely necessary. Otherwise, one does not know for sure whether the differences observed are due to climate change or to differences between models and reanalysis.
2. Equations (1) - (8) are unclear. For example, in Eq. (1), there is a sum indexing over $i$, but I don't understand what the index is exactly (and there is no $i$ in CHWEP). Also I don't understand what we are summing exactly. Eq. (3) is even less clear. What is actually computed? The mean (over all events) of the sum of the temperature (withing each events)? In this case, one should see numbers above 100? I am really lost here.
The same kind of criticism could be made for each equation. Note also that these statistics are computed on a 45 year period, across several members of an ensemble. I guess there is some sort of averaging over the years that should be made apparent.
Equation (10) is also very unclear). Why is $P_{rand}$ a coincidence probability?
3. The authors do not explain how the ensembles are taken into account in their study. Are the above statistics computed on each member of the ensemble averaged out? Why is the number of members is different for the climate models than for the reanalysis? Does it pose a problem?
4. The vocabulary is not consistent throughout the paper. For example, according to line 127, $F_C$ and $F_S$ denote frequencies (hence the $F$ letter), but Figure 2(a) and 2(b) show counts. Note also that according to caption, Figures 2c and 2d show heatwave intensity (between 0 and 40 °C). Why "intensity", and not "temperature"? Note also that according to Eqs. (3) and (4), $IT_C$ and $IT_S$ are means over sums, which I expect to be above 100°C?
5. The test used in this study is the Wilcoxon rank-test (WRT), which is consistent under specific assumptions on the two distributions (say $X$ for a statistic computed in the past, and $Y$ for the same statistics computed in the future). The WRT is consistent if the alternative is that $Y$ is stochastically larger than $X$, i.e. $P(Y* > X*) \geq P(X* > Y*)$, where $X*$ and $Y*$ are random values from $X$ and $Y$, respectively. The test on the median (line 139) is consistent with the additional assumption of that alternative is restricted to a shift in location, i.e. $F_Y(s) = \delta + F_X(s)$, with $\delta >0$.
In any case, it is misleading to state that "the WRST is emplyed to assess differences in extreme events characteristics" (line 135-136). The authors should reformulate and be more specific (and narrow).
6. Section is almost impossible to follow; the authors compare quantities that have not been properly defined, such as standard deviations SD (line 242), "ratio of observed to random probability" (line 243). I could not make any sense of this.
Other comments:
Line 110: "The threshold for future periods is determined based on the threshold of historical period" is unclear to me. Does it mean they are equal? If not, what is formula to go from the threshold in the past to that in the futyre?
Caption of Figure 1: "Identification ...events"
Line 167: One could argue that single heat waves are prevalent in all desert regions, where it only rains when it gets cooler
Caption of all Figures: change "frequency" to "count"
Line 242, how are the standard deviations computed?
Caption of Figure 7: reference to panel (c) is missing