the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 13 Jul 2026
Cascading Wildfire–Atmospheric River Hazards: Postfire Flood Risk after the 2025 Eaton Fire
Abstract. Wildfires and extreme precipitation events increasingly interact under a warming climate, yet their combined hydrologic impacts remain poorly characterized. In fire-prone regions such as Southern California, atmospheric rivers (ARs) can both promote wildfire through antecedent vegetation growth and trigger severe flooding in postfire landscapes. This study evaluates the compound flood risk associated with a strong AR event occurring after the 2025 Eaton Fire by examining both its likelihood and hydrologic consequences. Historical AR records were analyzed to assess the probability of sequential-year strong AR events, and an event-based hydrologic modeling framework (AGWA/KINEROS2) was used to characterize watershed response under prefire and postfire conditions using precipitation analogous to the February 2024 AR. Results indicate that strong AR events (Scale ≥ 3) occur with a mean annual frequency of 0.51 and a 38 % probability of recurrence in consecutive years, supporting the plausibility of a postfire AR scenario. Simulations show that wildfire-induced landscape changes amplify downstream flood response, increasing runoff volume by approximately 6 % and sediment yield by about 50 % in channels immediately below the burned area. Sequential storm pulses further enhanced postfire impacts, with the second flood peak increasing by ~30 % relative to prefire conditions. These findings demonstrate that wildfire and AR events act as interconnected hazards whose cascading effects can intensify flood and erosion risks. By linking event probability with watershed-scale hydrologic response, this study provides a framework for assessing postfire flood hazards under compound climate extremes and underscores the need for integrated risk management in fire-affected regions.
- Preprint
(2527 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2665', Anonymous Referee #1, 01 Aug 2026
-
RC2: 'Comment on egusphere-2026-2665', Anonymous Referee #2, 21 Oct 2026
Summary
The authors present a study that examines the hydrologic impacts of wildfire in Southern California, using a K2 model forced with precipitation totals similar to a 2024 atmospheric river (AR) event. To contextualize the study, they present the historical occurrence of atmospheric rivers (via CW3E AR catalog) and their sequential occurrence. This context suggests that it would be plausible for an atmospheric river to occur just after the 2025 Eaton fire. The modeling shows a spatially heterogeneous response across the watershed, with reaches and contributing areas near the burn area affected more than the more distal segments that integrate flow from unburned areas. The paper is framed as a study of "cascading wildfire–atmospheric river hazards," using the 2025 Eaton fire as a setup to ask what would happen if that area were hit by an atmospheric river similar in magnitude to the most recent AR in 2024. The effects measured by the model are flow volume and sediment yield.
Overall, this is a worthwhile modeling effort on a timely and important topic. The comments below are intended to help sharpen the scope, methods reporting, and the central framing around cascading hazards. My recommendation is for major revisions or an invitation to resubmit after substantial reworking.
Major comments
- Framing and scope of the "cascading hazard" narrative
My main suggestion concerns the scope and framing. The historical characterization of successive AR events supports the conclusion that another AR in 2025 with a magnitude similar to 2024 is plausible. However, the "plausibility" being established is for one AR following another AR, based on the historical occurrence of two atmospheric rivers in succession, rather than for an atmospheric river following a fire. Because the historical analysis did not include fires, it does not directly speak to the fire–AR sequence that is central to the paper. It would strengthen the study to include the successive occurrence of fire with ARs, which seems more central to the cascading-hazard question the paper is posing.
The paper mentions the increased fuel from the 2024 setup and the prolonged ~8-month dry period preceding the Eaton fire as part of the lead-up antecedent conditions, but these conditions don’t appear to be incorporated into the analysis, except for the initial condition of soil saturation = 0.001. If the antecedent fuel conditions are not part of the modeling, it would be helpful to clarify why the successive-AR history is included, or to connect that history more directly to the fire–AR sequence. It may also help to better chronicle the Eaton fire sequence for the reader (2024 AR fuel loading → ~8 months of dry conditions → January 2025 Eaton fire → hypothetical subsequent AR).
Overall, the historical occurrence analysis seems disjointed from the modeling portion of the paper. The only connection between the two is that the rainfall totals are “plausible”.
- Reporting of model parameterization
The paper notes that the K2 parameterization was automated by AGWA for post-fire conditions, and that soil saturation was set to 0.001, but it does not report the other key parameters. It would be very helpful to present the parameters that represent the increased hypothetical fuel load and the post-fire soil and vegetation conditions, at least the ranges in a table for pre-and post- fire scenarios, or shown in a figure. That would include the roughness coefficient, saturated hydraulic conductivity (K), and canopy settings. Were these parameters set to approximate the Eaton land cover conditions post-fire? And how did the parameters differ between pre- and post-fire conditions? Also, it is my understanding that the AGWA-K2 workflow needs a different set of parameters for the soil erosion/ sediment yield outputs. Those should be reported as well.
- Calibration and uncertainty, especially for sediment yield
The authors do address the model's calibration limitations by using volumetric totals from a neighboring streamgage located at a dam spillover. This is a substantial limitation, but using volumetric totals to approximate calibrated flow may be acceptable. The concern is sediment yield: it is presented as one of the main findings, yet it is less clear how much of the model output for this variable is meaningful given the calibration approach. Some discussion of the confidence in the sediment-yield results would strengthen the paper. Understanding that post fire sediment calibration is difficult, one could instead answer questions like: How do the changes in sediment yield pre-and post-fire compare to similar studies? Or even the sensitivity of yields to the parameters that were used in the soil erosion portion of the K2 model.
- Storm duration is framed as an evaluated effect but not varied
The paper describes evaluating the effect of storm duration on post-fire runoff amplification. Demonstrating this would require testing different storm durations and comparing their runoff effects, whereas a single storm duration was used here. The historical context added in Discussion Section 4.1 states that the implemented rainfall durations were "plausible" and notes that larger durations are likely under future real-world conditions, but it does not vary duration within the model. I would suggest either reframing this contribution or, ideally, running additional scenarios so that storm duration is genuinely investigated.
- Experimental design and the strength of the contribution
The study would be considerably stronger with some experimental variation in initial conditions, for example: with and without the additional fuel from the 2024 fire; different storm durations and pulse timings; and varying burn severity by adjusting the initial parameterizations that represent canopy cover, soil infiltration capacity, and the elapsed time between the fire and the AR. As currently framed, the study is an uncalibrated (or partially calibrated) model at a single location examining post-fire hydrologic conditions that could plausibly occur as an AR following a fire. Because the historical context examines successive ARs rather than fire–AR pairings, and because the antecedent fuel conditions are discussed but not incorporated into the analysis, the central cascading-hazard argument is not yet fully supported by the analysis presented. Additional scenario variation would go a long way toward establishing a novel contribution to how we think about the cascading effects of these two hazards.
Minor comments
- Discussion section: The discussion does not yet contextualize the results relative to prior work and does not mention the sediment-yield findings. Expanding it to interpret the results, and to address sediment yield specifically, would strengthen the paper.
- Precipitation event types: California ARs account for the majority of annual precipitation, but it may be worth noting whether other rainfall event types could produce totals comparable to the AR used here, and whether that affects the generality of the findings.
- Spatial pattern of results: The finding that reaches closer to the burn area experience more severe effects is intuitive and has been fairly well studied. It may help to frame this result in relation to that existing body of work, or to emphasize what is new here.
- Discussion of uncertainty seems incomplete: The last sentence in the discussion of sediment yield uncertainty is incomplete (line 336).
- Table 1 and References: The references for the data sources are incomplete. In the table, you should reference the release year of the dataset (to replace the n.d.). In the references section, the version number should be included. Also add citations to the datasets used in the body text in section 2.3.1.
- Line 26 and 27: “United States” is mentioned twice.
- Line 224: Comparable contributing area- what do you mean by comparable? Is it the same relative percentage of the watershed? The two hatched areas are very different in size.
- Line 242: the post-fire amplification is reported out- this is visible in Figure 5. Figure 5 also hints at a flashier response post-fire, but this was not reported out.
Citation: https://doi.org/10.5194/egusphere-2026-2665-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 212 | 88 | 27 | 327 | 21 | 24 |
- HTML: 212
- PDF: 88
- XML: 27
- Total: 327
- BibTeX: 21
- EndNote: 24
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Please see the attached file. Thank you