the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 25 Jun 2025
Atmospheric Rivers as Triggers of Compound Flooding: Quantifying Extreme Joint Events in Western North America Under Climate Change
Abstract. Atmospheric Rivers (ARs) are narrow bands of concentrated moisture that transport water vapor from the tropics to higher latitudes. They are responsible for ~90 % of poleward water vapor transport and play a vital role in water resource management along the North American west coast. While ARs significantly contribute to regional water supplies, they are also major drivers of flooding. This study investigates the extent to which ARs contribute to compound inland flooding (CIF) events where multiple drivers intensify flood risks, namely Rain on Snow (ROS) and Saturation Excess Flooding (SEF) events. Furthermore, the influence of internal climate variability is investigated relative to anthropogenic climate change. Using the CanRCM4 Large Ensemble simulations, we analyze the frequency and seasonality of AR-driven CIF events in Western North American coastal areas, with a focus on understanding how ARs interact with additional factors such as snowpack and soil moisture. ARs are shown to be dominant drivers of CIF events by contributing to the development and intensification of these events. These conditions also shape the seasonality and intensity of AR-driven CIFs. Projections suggest that internal climate variability can significantly contribute to future uncertainty in CIF frequency and intensity, complicating efforts to predict and mitigate these events. The findings underscore the importance of integrating AR-related flooding risks into flood management strategies and infrastructure design to adapt to a changing climate.
- Preprint
(2945 KB) - Metadata XML
-
Supplement
(979 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-2481', Anonymous Referee #1, 16 Jul 2025
General Comment:
This study examines how atmospheric rivers (ARs) contribute to inland flooding by looking at how often ARs, extreme rainfall, and compound events like Rain on Snow (ROS) and Saturation Excess Flooding (SEF) occur together along the western coast of North America. The authors also study how these events change with seasons and how future warming from climate change and natural variability could affect them. While the topic is important and the study includes some new findings, there are serious concerns about the methods used, especially how ARs and compound events are identified and defined. The authors need to address the following points before the manuscript is ready for submission to a scientific journal.
Major comments:
- The authors detect ARs using a fixed IVT threshold of 250 kg/m/s that persists for more than 24 hours at a single grid cell. However, this approach does not consider the typical structure of ARs which are long narrow bands of water vapor stretching thousands of kilometers. Many studies including Guan and Waliser 2015; 2019 have shown that ARs should be identified based on both moisture intensity and their spatial shape such as length and width. Ignoring these features can lead to the inclusion of unrelated weather systems that temporarily meet the IVT threshold but do not have the structure of an AR.
- Using only a point-based threshold without checking for spatial coherence may result in misclassification of short or wide moisture plumes (isolated moisture plumes, mesoscale convective systems, or tropical plumes) as ARs. This is especially concerning because ARs are known for their coherent filament-like structure and influence over large regions. Relying only on how long the IVT exceeds a threshold at a single location does not ensure that a true AR is present. It is unclear how the authors distinguish between real ARs and other systems that happen to bring strong moisture transport for a day or more.
- Although the study references Guan and Waliser 2015;2019, it does not apply their well-known AR detection method which includes geometry and landfall criteria. It is important to ask why the authors cite this work without adopting its key methods. Moreover, the fixed IVT threshold may not work equally well across different regions or seasons. In fact, Ralph et al 2019 and others have suggested that thresholds should be adjusted based on local climate and conditions. There is also no discussion of whether the chosen threshold or duration was tested for sensitivity.
- In recent years several objective and widely accepted methods have been developed to detect ARs. These methods have been compared through efforts like the ARTMIP project. The study would benefit from comparing their results with at least one such established method.
- The authors mention that they selected the Eulerian method for compatibility with CIF detection. However, this seems more like a convenience than a scientific justification. ARs are large scale features and it is important to detect them based on their full structure rather than just point based conditions. This would likely provide a more realistic match to compound flood events which also involve broad spatial atmospheric processes.
- In the introduction, SEF events are described as resulting from pre-existing soil saturation before intense rainfall. However, in the methodology, SEF events are defined based on co-occurrence of soil moisture and precipitation exceeding the 98th percentile on the same day. This contradicts the stated physical mechanism of SEF. It would be more appropriate to consider soil moisture on the day prior to precipitation, which better reflects antecedent saturation conditions driving saturation-excess runoff.
- The methodology does not clarify whether consecutive extreme precipitation days are treated as a single event or as multiple separate events. Multiday rainfall can lead to sustained soil saturation and progressive runoff generation. Defining such events based on their onset date and grouping them accordingly would align better with the hydrological reality and reduce double-counting.
- Similarly, the runoff condition (98th percentile on the same day or the next) should be evaluated over the entire event window, not just individual days, particularly in cases of compound events where runoff can build up over multiple days.
- ROS events are defined as days when both daily precipitation and snowmelt exceed the 98th percentile, with snowmelt contributing at least 20% of total liquid water. However, the approach raises a key question: can sub-98th percentile multiday rainfall still produces substantial snowmelt that satisfies the 20% contribution threshold? If so, limiting detection to only those days with extreme precipitation may exclude physically valid ROS events.
- The ROS definition also appears sensitive to daily co-occurrence of rain and snowmelt. Yet in reality, snowmelt lag and meltwater routing processes might cause peak melt to occur slightly after rainfall, particularly in colder regions. Has this temporal mismatch been accounted for or tested?
- The use of fixed 98th percentile thresholds across all variables, grid cells, and time frames may not equally capture extremes in diverse hydroclimatic settings. Have the authors performed any regional or seasonal sensitivity tests to ensure that this uniform threshold is not excluding impactful events in drier or colder basins?
- It is unclear whether non-snow season months were filtered out before detecting ROS events. Including periods when snowmelt is physically implausible might generate false positives or inflate event counts due to random high precipitation days.
- The method used to associate ARs with CIF events is based on day-of co-occurrence. Given the possible lag between AR landfall and inland flooding, particularly in snow-dominated regions, it would be helpful to test a wider time window (e.g., ±1 or 2 days) when evaluating AR-related CIFs.
Minor Comments:
- The authors may consider briefly mentioning whether there are other notable moisture transport mechanisms in Western North America besides atmospheric rivers. If so, a short explanation in the introduction would provide useful context.
- In the introduction, it is unclear what processes contribute to pre-existing soil saturation prior to Saturation Excess Flooding (SEF) events. Is the saturation driven by earlier low-intensity ARs, local convective rainfall, or antecedent snowmelt before the landfall of stronger ARs? Clarification on the dominant mechanisms would strengthen the physical interpretation.
- While a citation is provided for the bias-corrected dataset, the manuscript does not specify which climate variables were bias corrected. Please list the variables for clarity.
Citation: https://doi.org/10.5194/egusphere-2025-2481-RC1 -
RC2: 'Comment on egusphere-2025-2481', Anonymous Referee #2, 03 Aug 2025
This is a welcome contribution to the literature on atmospheric rivers as triggers of compound flooding. However, it is unclear if sufficient work has been undertaken at this stage to justify publication now.
The authors note that: "If common landfalling locations shift significantly under future warming scenarios, this could explain the difficulty in establishing a statistically robust relationship between ARs and CIF events in later warming periods, as such shifts are not explicitly accounted for in the current methodology. This highlights the need for further research to reduce uncertainties in modeling AR dynamics."
The authors are to be commended for admitting in the conclusions that: "The results carry considerable uncertainty, primarily due to internal climate variability, the exclusion of dynamic factors, sample size limitations, and AR detection methods. Future studies can improve the
methodology by focusing on more characteristics of ARs."For those involved in flood risk decision-making, the paper in its present form is much less informative and useful than it might be if further research would be undertaken to address some of the key uncertainties identified by the authors themselves.
Citation: https://doi.org/10.5194/egusphere-2025-2481-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 427 | 62 | 13 | 502 | 36 | 9 | 19 |
- HTML: 427
- PDF: 62
- XML: 13
- Total: 502
- Supplement: 36
- BibTeX: 9
- EndNote: 19
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1