the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Sep 2025
Present and future trends of extreme short-term rainfall events in Germany, by downscaling convective environments of ERA5 and a CMIP6 ensemble
Abstract. For the four main quadrant regions of Germany we study the possibility of projecting the occurrence of extreme convective rainfall events, as monitored by the CatRaRE database, into the future, based on CMIP6 projections of corresponding convective environments. We characterize such environments by using the atmospheric profile derivates convective available potential energy (cape) and convective inhibition (cin), along with model-simulated convective precipitation (cp). The convective environments are linked to the small-scale CatRaRE events by classifying the corresponding ERA5 fields according to the concurrent occurrence of such events. Classifiers are conventional machine-learning procedures along with more modern deep learning schemes. Positive centennial trends are identified for both antagonists cape and cin, serving as a source for large uncertainties of the corresponding CatRaRE-type trends. Their full distribution is analyzed using ANOVA, based on the factors of event severity, greenhouse gas emissions, climate models, classifiers, and region. Beyond all uncertainty, positive trends outweigh the negative ones for all regions.
- Preprint
(2520 KB) - Metadata XML
-
Supplement
(422 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-3584', Anonymous Referee #1, 04 Jan 2026
-
AC1: 'Reply on RC1', Gerd Bürger, 09 Feb 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-3584/egusphere-2025-3584-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Gerd Bürger, 09 Feb 2026
-
RC2: 'Comment on egusphere-2025-3584', Anonymous Referee #2, 07 Apr 2026
In this manuscript the authors aim to explore the potential for projecting the occurrence of extreme convective rainfall events into the future over Germany using a set of statistical classifiers trained to link convective environments (given by CAPE, CIN and convective precipitation) from reanalysis and climate model simulations with observed precipitation events. The validation of the classifiers indicates some skill in correctly classifying observed events and trends for the present conditions, especially for more conventional machine-learning procedures. Applying the classifiers on future scenarios of convective environments from CMIP6 global models result in increasing trends of the occurrence of small-scale convective precipitation in all four quadrants of Germany, although with large spread. The uncertainty is analyzed through application of ANOVA, revealing that most of the uncertainty is associated with the choice of GCM, followed by the choice of SSP scenario.
General comments:
The manuscript is very well organized and structured and clearly written. The scientific questions are stated clearly, and the method is for the most part well presented and explained. Although, since the details of the methodology were described in a previously published paper, I felt the need to read that one as well (which was also an interesting read!). The topic is interesting and the study represents, in my opinion, an interesting illustration of how statistical tools can be used to derive trends and changes in local phenomena, such as extreme precipitation, based on large-scale fields which are represented with greater confidence in coarse scale climate models. Below follows specific, mostly minor, comments and suggested changes. If these are appropriately addressed, I would find this manuscript ready for publication.
Specific comments:
L13-14: “with an estimated damage…”. Suggestion to change to “with an estimated cost due to flood damages ….” or similar.
L15: “a warmer atmosphere” → “a warmer Earth atmosphere” (since it may not apply to all atmospheres, even though the moisture-holding capacity increases).
L18-20: Is this sentence referring to the use of very high-resolution convection-permitting models? If so, I suggest the sentence be modified slightly since these models do not concern the implementation of convection processes, but rather enabling the removal or turning off of convection parameterizations and treating (deep) convection explicitly.
L24: I’m not sure I fully agree with this, or maybe I misunderstand. I’m under the impression that conventional statistical downscaling approaches have not been used historically, since they haven’t shown capability of generating realistic high-resolution convective rainfall fields with correct spatio-temporal patterns and extreme values. Hence, the promising avenue of ML methods…
L36-38: Do you plan to extend the analysis using the new CMIP6/EURO-CORDEX suite when these downscalings have been published? Would be interesting to see the impacts on the classification of using input data with significantly higher spatial resolution.
L56: Even though the reader most likely understands what P00 represents, it’s worthwhile to define it explicitly in the text.
L64: overcome → reach (?)
L79: “same resolution” → “same temporal resolution” (if you refer to the temporal resolution)
L127: What is the main reason for the reduced or non-existing skill for the P99 severity level? Sampling size? Would be good if you could elaborate on this further. Is it still worthwhile to continue consider P99 given the small skill?
L134: cf. 2.3 → cf. Sec. 2.3
L144: The “base rate” expression is used frequently through the manuscript, however, it hasn’t been clearly defined. Is it the frequency of occurrence of a certain class; either observed, based on ERA5-fields or simulated in GCM historical simulation? Please clarify.
L156-158: Remarkable difference between the two GCMs, not only in trends but also in the inter-annual variability. Could you also add the ERA5-driven classification in the same figure (Fig. 7)? Perhaps just the variability (standard deviation?), so you will get an indication of how well the GCM represent this aspect (I guess you cannot expect the trends to be similar to ERA5-classification). How does the variability in these GCMs compare with the other GCMs?
L165: Regarding the different model behavior; have you looked at the temperature and humidity estimates and trends from the GCMs, both over the very local target area and on the larger (European?) scales? Could this give a clue on the different behavior?
L166: “Examples analogous to 7 are Figs. S5-S8” → “Examples analogous to Fig. 7 are presented in Figs. S5-S8”
L172-173: Have you analyzed specifically the changes in intensity and frequency of convective rainfall events, or do you infer this somehow from Fig. 8? The normalized cp in Fig. 8 is derived from cp intensity, right, not frequency?
L174: What do you mean, with respect to cin/cape changes, by “… in line with the law of clausius-clapeyron and the global warming narrative”? Decrease in rainfall frequency but at the same time more intense rainfall when it occurs? Consider rephrasing.
L176: “More examples are Figs. S9-S12” → “More examples are presented in Figs. S9-S12”
L176: Also, you have mainly discussed results for SW region in this section. It would be nice to include some information on the results over the other sub-regions as well (even though you refer to the SI). Do you see the same model behavior or are there any other conclusive results or take-home messages?
Figures & Tables
The figures and tables are overall very nice and clear! A few minor comments below:
- Figure 5: Why are the (stacked) bars plotted from top (prob 1) – down (prob 0)? Wouldn’t it be more intuitive to have them originate from prob 0? Or am I misreading this?
- Figure 7: As mentioned in comment above, it would be nice to have some metric of the spread (variability) from ERA5-driven classification in the plots as well. Perhaps same in Fig. 8 (if possible)?
- Figure 10: In the figure text/caption, it would be good to explain again what the “base rate” represents.
- Figures 11, 12: A minor thing; these figures are quite large (in printed manuscript version), while the content of both is rather minimalistic. You could easily make these smaller and put them together in one figure, or even merge into one panel with semi-transparent (and colored! Unless the economy is an issue :)) bars.
Citation: https://doi.org/10.5194/egusphere-2025-3584-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,221 | 161 | 29 | 1,411 | 58 | 26 | 34 |
- HTML: 1,221
- PDF: 161
- XML: 29
- Total: 1,411
- Supplement: 58
- BibTeX: 26
- EndNote: 34
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
see PDF attached