the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Nov 2025
Unveiling the Link Between Extreme Precipitation Events and Flood Disasters in China: From 3D Perspective
Abstract. Extreme precipitation events and their triggered flood disasters have received increasing attention owing to their severe threats to human lives and socioeconomic development. However, there is still a lack of research on their evolutionary characteristics and driving factors from a three-dimensional (3D) event-based perspective. Here, we developed a 3D automatic recognition algorithm based on the connected component 3D algorithm. This method was applied to investigate the 3D characteristics of 632 flood-causing precipitation (FCP) events in China from 2000 to 2023. The associated flood disasters and their underlying driving factors were further analysed. The FCP events with larger accumulated magnitudes and affected areas are mainly distributed in the center of Southern China (SC) and Northern China (NC), mostly moving eastward with longer distances and lifespans. FCP-induced flood disasters are more severe in the SC and parts of the NC, while a relatively higher proportion of flood disaster losses are concentrated in the southeastern fringe of the Qinghai-Tibetan Plateau (TP) and southwestern China (SWC). In other words, flood disasters caused by FCP in China exhibit the changing characteristics of "high impact-low losses ratio" in SC and NC and "low impact-high losses ratio" in TP and SWC. Notably, despite the increase in 3D characteristics of FCP events over the past two decades, flood disasters have shown a significant reduction, except for the direct economic losses. Driving factor analysis indicates that the combination of precipitation and environmental factors have the greatest explanatory power for most flood disasters in China, while human activities have a prominent impact on the flood disasters in the center of SC and NC. These findings provide new insights into the characteristics of FCP events and their associated flood disasters from a 3D event-based perspective.
- Preprint
(10121 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 05 Jan 2026)
-
RC1: 'Comment on egusphere-2025-4728', Anonymous Referee #1, 18 Nov 2025
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-4728/egusphere-2025-4728-RC1-supplement.pdfReplyCitation: https://doi.org/
10.5194/egusphere-2025-4728-RC1 -
CC1: 'Comment on egusphere-2025-4728', Qiang Wang, 27 Nov 2025
reply
This manuscript makes valuable contributions to the field of hydrometeorological disaster research. By introducing a 3D event-based perspective to analyze flood-causing precipitation (FCP) events and their associated flood disasters in China from 2000 to 2023, it addresses a key gap in existing studies that often focus on isolated precipitation metrics (e.g., daily intensity) rather than comprehensive spatiotemporal evolution characteristics. The application of the connected component 3D (CC3D) algorithm to identify independent FCP events, combined with geographical detectors to disentangle driving factors, also provides a rigorous methodological framework for quantifying the link between extreme precipitation and flood impacts. Additionally, the identification of regional differentiation patterns (e.g., ‘high impact-low loss ratio’ in Southern/Northern China and ‘low impact-high loss ratio’ in the Qinghai-Tibetan Plateau) offers novel insights for region-specific flood risk management, which is of practical significance for China’s disaster mitigation efforts. However, there are also some shortcomings that deserve improvement.
1. The manuscript adopts a CC3D algorithm for FCP event identification, but it fails to clearly elaborate on the core innovations of this algorithm in the context of the study, for example, the rationale for optimizing the 95th percentile threshold for extreme precipitation or the criteria for judging event independence. Notably, the study solely uses the 95th percentile threshold without comparing it with the fixed threshold (e.g., 16 mm/h, as defined by the China Meteorological Administration) for validation. This omission makes it impossible to verify whether the selected threshold is universally applicable to FCP event identification across different regions (e.g., the arid northwest and humid southern China), potentially leading to under-detection or misdetection of events in specific areas.
2. Population and GDP data use only two time points (2005 and 2020) to represent annual data for 2000–2010 and 2010–2020, respectively. This simplification ignores spatiotemporal heterogeneity in population migration and economic growth (e.g., rapid urbanization in the Yangtze River Delta), which could significantly distort the assessment of human activities’ impact on flood disasters.
3. The manuscript acknowledges that hydraulic engineering (e.g., dams, reservoirs) is an important flood mitigation factor but excludes it from the driving factor analysis due to data limitations. However, large-scale hydraulic projects in China (e.g., the Three Gorges Dam) have significantly altered flood regimes, especially in SC and NC. Omitting these factors may limit the comprehensiveness of the driving factor analysis, and the manuscript should discuss how this omission impacts the interpretation of results.
4. The manuscript reports that FCP 3D features (e.g., accumulated magnitude, lifespan) have increased, while most flood disaster metrics (except direct economic loss) have decreased. The explanation attributes this discrepancy to hydraulic engineering, but no quantitative evidence is provided to support this claim. For example, could the reduction in disaster impacts be quantified by the number or storage capacity of reservoirs built during the study period? Additionally, the increase in direct economic loss is linked to economic development, but a regression analysis or correlation between GDP growth and economic losses would strengthen this argument.
5. The finding of high death rates in the southeastern fringe of TP and SWC is attributed to the lack of flood mitigation infrastructure and flash floods triggered by landslides/debris flows. However, the manuscript does not provide data on infrastructure coverage (e.g., levee density, early warning systems) or the frequency of secondary disasters (landslides) in these regions. Incorporating such data or citing relevant studies would enhance the explanation of spatial loss patterns.
6. What’s 3D characteristic in Abstract?
7. Figure 3 (FCP identification algorithm flowchart) is poorly labeled: terms like ‘26 connectivity tracking’ are not explained.
8. Table 1 (summarizing variables and data sources) is mentioned in the text (Section 2.3) but not included in the manuscript.
9. The term ‘flood-causing precipitation (FCP)’ is used throughout the manuscript but is not formally defined. It should be explicitly defined at the start of the Methods section to avoid confusion with ‘extreme precipitation’ or ‘heavy rainfall’.
10. The calculation method for ‘accumulated affected area’ (a 3D FCP indicator) is vague, whether it refers to horizontal projection area or curved surface area (critical for mountainous regions) is not specified.
11. Page 7, Line 175: ‘the 26-connectivity searching allows that a contiguous precipitation event occurring at a grid on the current hour can move to the adjacent grids in the following hour’ is grammatically incorrect. It should be revised to ‘the 26-connectivity search enables a contiguous precipitation event at a grid in the current hour to move to adjacent grids in the following hour.’
12. Page 12, Line 295: ‘the annually mean values and trend of TC FCP events’ uses an incorrect adverb; it should be ‘the annual mean values and trend of TC FCP events.’
13. Page 15, Line 355: ‘results in economic losses of USD 57.46 billion’ lacks a clear time reference (which flood event?)
14. Figure 1’s caption mentions a study period of 2000–2023, but the subplot label (a) 2024 is confusing.
15. The abstract mentions "632 flood-causing precipitation (FCP) events" but does not specify how these events were derived from the 1,041 flood disaster records. A brief note on the event merging/classification process would improve clarity.
16. The manuscript states that code is available upon request from the corresponding author, but making the code publicly available (e.g., on GitHub) would enhance reproducibility, which is increasingly important in environmental science research.Citation: https://doi.org/10.5194/egusphere-2025-4728-CC1 -
RC2: 'Comment on egusphere-2025-4728', Anonymous Referee #2, 25 Dec 2025
reply
Lines 40–47
I encourage the authors to expand the discussion on changes in extreme precipitation under climate change, with a particular focus on Clausius–Clapeyron scaling and the worst-case scenario of precipitation. Scaling rates (i.e., % K⁻¹) have been reported to increase with precipitation magnitude, and incorporating this background would enhance the importance of the present study, which focuses on extreme precipitation. Please refer to the studies listed below.https://doi.org/10.1016/j.jhydrol.2025.133724
https://doi.org/10.5194/hess-28-1251-2024
Line 61
Please include references to support the statement “Most previous studies”.Lines 86–94
Although the performance of IMERG has been extensively evaluated at the global scale, it can be dependent on region and precipitation system. Furthermore, a spatial resolution of 0.1° is sometimes insufficient to represent extreme precipitation associated with localized convective systems. I encourage the authors to strengthen the justification for the use of IMERG, specifically by addressing its performance in China and for extreme precipitation. The authors are welcome to perform additional analyses in this context; if not, please expand the discussion based on previous studies.Lines 100–102
What about the consistency between the datasets used before and after 2020? I believe that the datasets (e.g., the Meteorological Disaster Yearbook and newspaper-based records) may differ substantially in several aspects. Combining two different datasets in a temporally consecutive manner may affect trend analyses and related results. Could the authors provide justification for combining these datasets and discuss the potential impacts on the conclusions?Caption of Figure 2
Please explain panel (c).Line 160
Did you test the sensitivity of this selected period to your results? I encourage the authors to examine the sensitivity in order to evaluate the robustness of the conclusions.Lines 185–189
It is unclear how the FCP was classified into TC-related and non-TC-related precipitation. Please provide a detailed explanation and justification for this classification.Lines 282–283
Here, the analysis could be expanded based on the Clausius–Clapeyron scaling rate discussed in my earlier comments. It would be of broad interest if the trends shown in Figure 6 could be interpreted in terms of temperature-dependent scaling rates.Lines 366–369
“Accumulated magnitude and lifespan demonstrate higher CC values than accumulated area and moving distance, indicating that precipitation events with large accumulated magnitude and longer lifespan are more susceptible to flood disasters.”
I believe that such conclusions regarding the relative importance of the considered factors cannot be drawn based solely on direct comparisons of correlation coefficients because of potential multicollinearity. Please clarify this point.Lines 445–449
I encourage the authors to further elaborate the discussion on changes in extreme precipitation under climate change, with greater emphasis on scaling rates with rainfall magnitude. Recent studies suggest that scaling rates can depend on rainfall mechanisms (e.g., tropical cyclones versus non-TC events), which may provide additional support for the present findings.
Citation: https://doi.org/10.5194/egusphere-2025-4728-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 206 | 205 | 22 | 433 | 13 | 13 |
- HTML: 206
- PDF: 205
- XML: 22
- Total: 433
- BibTeX: 13
- EndNote: 13
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1