the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
A downward counterfactual analysis of flash floods in Germany
Abstract. In this study, we present the results of a counterfactual search for flash flood events in Germany. We used DWD’s RADKLIM product to identify the ten most extreme precipitation events in Germany from 2001 to 2022, and then assumed that any of these top 10 events could have happened anywhere in Germany. In other words, the events were shifted around all over Germany. For all resulting positions of the precipitation fields, we simulated the corresponding peak discharge for any affected catchment smaller than 750 km2. From all the realisations of this simulation experiment, the maximum peak discharge was identified for each catchment.
In a case study, we first focused on the devastating flood event in July 2021 in western Germany. We found that a moderate shifting of the event in space could change the event peak flow at the gauge Altenahr by a factor of two. Compared to the peak flow of 1004 m3/s caused by the event in its original position, the worst case counterfactual of that event led to a peak flow of 1311 m3/s. Shifting another event that had occurred just one month earlier in eastern Germany over the Ahr river valley even effectuated a simulated peak flow of 1651 m3/s.
For all analysed subbasins in Germany, we found that, on average, the highest counterfactual peak exceeded the maximum original peak (between 2001 and 2022) by a factor of 5.3. For 98 % of the basins, the factor was higher than 2.
We discuss various limitations of our analysis, which are important to be aware of: with regard to the quantification and selection of candidate rainfall events, the hydrological model, and the design of the counterfactual search experiment. Still, we think that these results might help to expand the view on what could happen in case certain extreme events occurred elsewhere, and thereby reduce the element of surprise in disaster risk management.
- Preprint
(6952 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
AC1: 'Comment on nhess-2023-224', Paul Voit, 02 Jan 2024
We like to correct the average value quoted in the the section "Conclusions", line 398. Accidentally we mention the median, and not the mean value. Instead of 4.8 the average factor of the worst case scenario is 5.3, as mentioned in the abstract. We apologize for the confusion and will fix this error in the revised version
Citation: https://doi.org/10.5194/nhess-2023-224-AC1 -
RC1: 'Comment on nhess-2023-224', Anonymous Referee #1, 12 Feb 2024
Based on a radar-based precipitation dataset (2001 to 2022) of the German Weather Service, counterfactual studies on flash flood events in Germany are carried out. Counterfactual studies show what happens when an extreme event occurs under different conditions (e.g. at a different location) and can help disaster risk management to better prepare for such events. First, the ten most severe precipitation events in Germany are identified, which then serve as the basis for the counterfactual studies. In particular, the prominent case (floods in July 2021) is discussed in more detail. The TOP10 events are then shifted (depending on the distance from the point of origin) and the potential hydrological response to rare heavy precipitation events is analysed using a hydrological model. One aim is to determine how close the actual historical events are to the worst-case scenarios.
I find it a very nice paper with interesting results. However, these could be presented in a slightly more structured way, some detailed explanations for non-experts would be helpful and some of the highlights could be explained in more detail (especially part 2 of the results).
See attached PDF for details with major and minor comments.
- AC3: 'Reply on RC1', Paul Voit, 14 Mar 2024
-
RC2: 'Comment on nhess-2023-224', Anonymous Referee #2, 14 Feb 2024
The research work presented in the paper “A downward counterfactual analysis of flash floods in Germany” aims at understanding how high precipitation events in Germany could have resulted in stronger hydrological responses if they had occured elsewhere. To do so, the 10 most extreme HPEs were selected and shifted accross Germany, and a simplified hydrological model was used to calculate the quick runoff response on a large number of small catchments.
The results focus first on the Ahr river catchment, then on all subbasins in Germany. They quantify the exceedance of the maximum original peak by the highest counterfactual peak. More generally, the study presents a methodological framework for a systematic downward counterfactual search for flash floods at the national scale.
In my opinion, the paper is very interesting, and evidence of high-quality research. The paper is very well written, the scientific questions are well defined, and the assumptions are clearly outlined and discussed. The proposed work is very useful since counterfactutal thinking is a recent philosophy in flood risk assessment. I only have a few suggestions for the authors to improve the quality of their manuscript. I think some explanations and illustrations could be added.
Abstract: You should define the concept of counterfactual thinking in the abstract, for non-experts.
Line 72 : It would be good to clearly define the time window analysed (e.g. by just adding "between 2001 and 2022" at the end of line 72).
Line 113: It might be difficult to do, but I think that a figure illustrating the method and its different stages would make it much easier to understand.
Line 116 : Maybe you could briefly explain how is calculated the xWEI, and give some orders of magnitude.
Line 123: Can you detail "Clustered based on their neighborhood" ? Or show it on an example if you decide to include an illustration.
Line 133: Can you explain why the upper limit is precisely 750km2 for the catchment size? Could you give the minimum catchment size?
Table 1: Adding information about impacts (e.g. damage) would help understanding the gravity of these events, which are not well-known by international readers.
Figure 5: Could the legend be standardised?
Section 5/ Section 6: In my opinion it would be interesting to discuss the fact that this counterfactual approach is "only" performed from a hydrological point of view (hazard-based). However if you want to go to the bottom of the question "how close actual historical events have already touched upon the worst case scenario", you would need to shift to an impact-based approach.
Citation: https://doi.org/10.5194/nhess-2023-224-RC2 - AC2: 'Reply on RC2', Paul Voit, 14 Mar 2024
Interactive computing environment
Examplary model code and notebooks Paul Voit https://github.com/plvoit/counterfactual_flash_flood_analysis
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
355 | 122 | 21 | 498 | 16 | 17 |
- HTML: 355
- PDF: 122
- XML: 21
- Total: 498
- BibTeX: 16
- EndNote: 17
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
Maik Heistermann
To identify the flash flood potential in Germany, we shifted the most extreme rainfall events from the last 22 years systematically across Germany and simulated the consequent run off reaction.
Our results show, that almost all areas in Germany have not seen the worst-case scenario of flood peaks within the last 22 years. With a slight spatial change of historical rainfall events, flood peaks by the factor 2 or more would be achieved for most areas. The results can aid disaster risk management.
To identify the flash flood potential in Germany, we shifted the most extreme rainfall events...