the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 Mar 2026
Compound Drivers and Spatial Connectivity led to the Devastating Debris Flood in the Village of La Bérarde, June 2024, French Alps
Abstract. On the evening of June 21, 2024, a debris flood inundated the village of La Bérarde, located at the heart of the Écrins mountain range in the French Alps. More than 200 000 m3 of materials were deposited in place of the village. People were evacuated on time but many buildings were destroyed and buried. The event was understood to be driven by a 10-year return period rain alongside to a 20-year snowmelt, the drainage of a supra-glacial lake, with potentially more internal water storage in the Bonne Pierre glacier. While we do not have direct observation of the supraglacial lake drainage, we found a number of evidences pointing to the role it likely had in destabilizing and triggering sediment transport from the Bonne Pierre fan. This work required an interdisciplinary approach to establish the set of scientific elements to reconstruct the event's chronology and rarity. We found that the combination of moderate magnitude drivers is not sufficient to explain the impacts observed. The location of the village on an alluvial fan directly connected to the source of sediments, was also key to understand the magnitude of the impacts. This event took place in a region particularly sensitive to climate change, where physical processes of the cryosphere at play are subject to alteration in a changing climate (e.g. precipitation amount and phase). The recent paradigm of compound events helps reconsidering the nature of this event and suggests possible approaches in anticipating new up-coming compound events in an era in which the Alps are entering a new paraglacial adjustments. Nevertheless, compound events remain difficult to forecast as they may be generated by diverse set of combination of low to moderate magnitude hazards associated to specific geographical, geomorphological, cryospheric, and meteorological onsets.
- Preprint
(4413 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-971', Lorenzo Marchi, 20 Mar 2026
-
RC2: 'Comment on egusphere-2026-971', Anonymous Referee #2, 31 Aug 2026
- General Evaluation and Contextual Importance
This manuscript presents an extensive, multi-institutional investigation of the catastrophic June 21, 2024 torrential disaster that devastated the village of La Bérarde in the Écrins Massif (French Western Alps). The authors have assembled an exceptionally rich, perishable post-event dataset combining synoptic meteorology (ERA5), quantitative radar precipitation reanalysis (ANTILOPE), detailed snowpack physics modeling (S2M / SAFRAN-Crocus), multi-temporal high-resolution airborne LiDAR and satellite stereo-photogrammetry (IGN LiDAR-HD, Pléiades, Pléiades Neo), regional broadband seismology (EPOS-France), 1D/2D hydrodynamic reconstructions (Telemac-1D/2D), and semi-distributed hydrological modeling (MORDOR-SD). The overarching goal is to reconstruct the multi-hazard sequence, quantify water and sediment fluxes, evaluate driver recurrence intervals, and interpret the event through the lenses of compound climate extremes, sediment connectivity, and paraglacial catchment adjustment.
The core scientific thesis of the paper is compelling and of outstanding interest: the extraordinary severity of the disaster was not driven by a single, off-scale meteorological anomaly or by an explosive glacial lake outburst surge alone. Rather, it resulted from the synchronous compounding of moderate-magnitude hydrometeorological forcings (a ~10-year return period rainfall event superimposed onto a ~20–50-year snowmelt rate over an exceptionally deep, ripe isothermal snowpack), the progressive subglacial drainage of an expanding decadal supraglacial lake (~100,000 m³) and associated cavity storage on the Bonne Pierre glacier, and critical geomorphic conditioning. Crucially, abundant unconsolidated paraglacial sediment stores directly coupled through a steep, confined proglacial channel (Bonne Pierre torrent, slopes 20–40%) delivered over 300,000 m³ of coarse debris directly to the fan apex, whereas the adjacent, much larger Upper Étançons branch acted as a sediment buffer along wide, low-gradient braided reaches.
I wish to emphasize my sincere appreciation for the collaborative effort deployed by the authors to capture and synthesize this perishable event. The work is already very good, scientifically robust in its observational foundation, and highly relevant for the international geomorphological, hydrological, and alpine hazards communities. In an era where mountain catchments globally are adjusting to accelerated cryospheric degradation (recalling also recent events in the Himalayas), La Bérarde provides a textbook benchmark of compound paraglacial hazard. The conclusions of the manuscript could even more explicitly emphasize the wider applicability of these findings for risk assessment in similar high-relief deglaciating valleys. The observations and recommendations detailed below are offered with a purely constructive intent to assist the authors in streamlining the paper's architecture, strengthening the geomorphological process interpretation, and eliminating redundancies. The exact depth to which each point is developed is left to the discretion of the authors and the editor.
- Core Methodological and Geomorphological Critique
A primary aspect that requires substantial revision is the overall structure and narrative flow of the manuscript. In its current state, the text resembles an unedited research report rather than a tightly organized journal paper. The narrative is burdened by significant length and repetitive statements (for instance, the destruction of the village and key sediment volumes are restated multiple times across Sections 3, 4, 5, and 7). Furthermore, the structural sequence is counterintuitive: the geomorphological impacts (Section 3) are presented before the physical drivers (Section 4), and the chronological reconstruction (Section 5.4)—which forms a central pillar of the study—is subordinated at the end of the river response section. I strongly encourage the authors to restructure the manuscript following a logical process chain: Section 2 should establish the study site, geomorphic setting, and historical context; Section 3 should be a dedicated, self-contained Methods section synthesizing the core workflows (S2M, MORDOR-SD, Telemac-2D, LiDAR DoD, seismology, and PyMC GEV modeling), rather than the current brief Section 2.5 that merely delegates all procedures to separate appendices; Section 4 could pool the empirical findings under a unified 'Results' framework (initiating with the full multi-hazard event chronology, followed by meteorological forcings, hydrological/seismic responses, glacial lake evolution, and sediment budgeting); and Section 5 should focus on the overarching scientific discussion (hydraulics, connectivity, paraglacial adjustment, and compound risk governance).
From a geomorphological standpoint, the characterization of catchment morphometry and flow regimes should be grounded in established quantitative metrics. In particular, computing the simple Melton ruggedness number for the Étançons torrent at the La Bérarde fan apex provides could provide valuable clues regarding long-term flow processes (conceptually the same result could be obtained by mean or median IC values, commenting on the spatial arrangement of the values though). With a total basin relief H_b ≈ 2300 m (from the 1700 m fan apex to the 4000 m peaks) and a drainage area A_b = 34.4 km², the Melton number is approximately 0.39. Following regional thresholds for Alpine catchments (e.g., Wilford et al., 2004; Bertrand et al., 2013; Marchi et al., 2019), a Melton number around 0.39–0.40, combined with the fan slope of 7–9%, places the overall Étançons basin right at the transition boundary between bedload/debris-flood dominated systems and debris-flow basins. In contrast, the Bonne Pierre subcatchment (slopes 20–40%, high relief, small area) exhibits a much higher ruggedness typical of debris flow generation. This morphometric duality perfectly mirrors the observed physical behavior: the authors should explicitly articulate the spatial and temporal rheological cascade, wherein an upstream granular debris flow and intense channel scour in the steep (20–40%) proglacial gorge (Parts I and II) transitioned through the confined bedrock canyon (Part III) with substantial liquid dilution from the Upper Étançons, into an intense hyperconcentrated, bedload-dominated debris flood across the unconfined alluvial fan (Part IV).
Regarding the alluvial fan setting and the sediment budget derived from airborne LiDAR differencing, several points merit attention. The area of the La Bérarde alluvial fan should be explicitly stated in Section 2.1 to allow readers to appreciate its modest size relative to the total catchment, and the truncation of the fan toe by the energetic lateral erosion of the Vénéon river should be highlighted immediately when discussing sediment evacuation. In Table 1, the sediment balance indicates -280,000 m³ from Bonne Pierre, -10,000 m³ from Upper Étançons, and +199,000 m³ deposited on the fan, leaving an unresolved deficit of ~91,000 m³. Line 496 simply asserts that the missing 100,000 m³ were transported to the Vénéon. This exported volume must be explicitly formalized in Table 1 as a closed mass-balance term, accompanied by a brief discussion of its morphological impacts on the Vénéon trunk stream (such as bed aggradation, temporary damming, or bar formation). Furthermore, regarding the Level of Detection (LoD) applied to the Difference of DEMs (DoD), the authors adopt a constant LoD of 0.1 m based on literature citations. A constant LoD is a perfectly valid and practical approach when morphogenic signals are massive. However, rather than simply asserting the 0.1 m threshold by citation, the authors should provide data-driven evidence justifying this choice (such as reporting the standard deviation of elevation residuals or vertical RMSE on stable bedrock surfaces). Additionally, it would be geomorphologically valuable to include a comparative map illustrating the raw DoD versus the LoD-thresholded DoD (or a noise mask), allowing readers to visually appreciate the spatial distribution of noise removal across the complex alpine topography.
Spatial sediment connectivity is highlighted in the manuscript title and Section 7.3 as a governing determinant of the disaster. The application of the Cavalli et al. (2013) Index of Connectivity (IC) provides an elegant illustration of why the Bonne Pierre branch acted as an efficient, directly coupled sediment conveyor, whereas the Upper Étançons was effectively decoupled from the fan by wide, low-gradient (4–9%) braided reaches that acted as sediment sinks. The authors could elevate this insight by discussing the interplay between structural and functional connectivity: the Upper Étançons acted as a 'water pump'—it contributed massive liquid runoff from snowmelt and rain without substantial bedload export, which entered the confluence and supplied the excess transport capacity required to flush the massive sediment pulse supplied by the Bonne Pierre torrent through the canyon and onto the fan. This paraglacial response should also be contextualized within the broader literature on postglacial evolution of alpine valleys (e.g., Ballantyne, 2002; Jarman et al., 2011; Brardinoni et al., 2018), highlighting how glacier retreat since the Little Ice Age (~1850) has exposed bounded, highly mobilizable sediment stocks governed by paraglacial exhaustion dynamics. Please also clarify the computational parameters used for IC in Figure 11, including DEM grid resolution, weighting factor computation and at which scale and why, the target definition (fan apex vs. catchment outlet… or hydro-net), and justify the 250 m grid choice.
The glacio-hydrological and seismological analyses provide crucial constraints on the event chronology. The authors present an elegant hydraulic back-calculation in Section 7.1 and Appendix A12 showing that an instantaneous breach (<30 min) of the ~100,000 m³ supraglacial lake is ruled out: empirical debris flow equations and 2D Voellmy routing indicate that a sudden outburst would have generated peak discharges of 1,200–3,000 m³/s and surge depths of 10–17 m on the fan, which starkly contradict post-event flood marks (3.0–3.5 m in the gorge) and the survival of the main bridge (clearance 5.5 m). This hydraulic deduction is reinforced by the observed flow rheology: the fact that the distal channel and fan experienced a sustained debris flood rather than a catastrophic debris flow surge confirms that the subglacial conduit network acted as a hydraulic choke, delivering a prolonged release over several hours. However, when attributing the residual discharge peak (the gap between the MORDOR-SD simulated 66 m³/s and the hydraulic peak of 85–110 m³/s) to the lake and cavity drainage, the authors should acknowledge alternative or compounding hydrological uncertainties, such as radar underestimation of localized orographic convection (Mont Moucherotte radar was down, leaving Mont Colombis at 50 km), rapid runoff from bare rock faces, and sediment bulking inflating stage at the destroyed gauging station. Similarly, the seismological interpretation of the 0.4–0.5 Hz signal at station OGAG (25 km away) warrants caution: while high-frequency noise (2–10 Hz) are correctly referred to fluvial turbulence and bedload collisions in the adjacent Durance River, the 0.4–0.5 Hz band suffers from overlaps with secondary oceanic microseisms; The authors should consider and discuss whether this low-frequency energy (and the corresponding polarization shift at OGAG) could instead reflect discrete, large-scale mass movements—such as major bank collapses, moraine slumps, or bedrock gorge failures along the Bonne Pierre branch—rather than continuous fluvial bedload transport. Acknowledging this potential mixed source (slope failures vs. fluvial dynamics) and the complexities of 3D wave propagation in steep alpine terrain would make the seismological interpretation much more robust.
Regarding the statistical return periods in Section 6.2 and Appendix A14 (estimated at ~7–15 years for rainfall, ~22–57 years for snowmelt, and ~153–157 years for total runoff), the authors rightly note that fitting a univariate GEV to aggregated runoff ignores seasonality. In the Western Alps, intense Mediterranean rainstorms typically occur in autumn, whereas peak snowmelt is confined to late spring. The co-occurrence of a 10-year rainstorm over a ripe, 90th-percentile snowpack in late June is an exceptionally rare joint event. The authors should frame this limitation clearly, presenting the 153-year estimate as a simplified baseline and highlighting multivariate copula or joint probability approaches as the appropriate conceptual frontier for compound mountain hazards.
- Specific Recommendations and Concluding Remarks
A series of specific textual, editorial, and typographical corrections will help polish the manuscript. In the title and throughout the text, replace 'evidences' (lines 6, 592, 608, 621, 796) with 'evidence' (uncountable noun), and in line 20 replace 'colluding in time and space' with 'co-occurring' or 'synchronizing' (colluding implies deliberate human conspiracy). In lines 14–15, correct the grammatical agreement to 'an era in which the Alps are entering a phase of renewed paraglacial adjustment'. In the introduction (lines 39–44), shorten the description of the Chamoli event as the citation to Shugar et al. (2021) is sufficient, and delete the redundant sentence 'The detached mixture of rock, ice and water collapsed...'; ensure a clear physical distinction between massive rock-ice avalanche cascades and hydrometeorological-thermokarst compound triggering over paraglacial sediment stores, perhaps mentioning the 2023 South Lhonak/Sikkim GLOF (Sattar et al., 2025). Figure 2 (the historical painting by Abbot Guétal) provides cultural context but does not convey essential scientific information; it could be moved to the site description or relegated to the Appendix. In Section 3.1, it would be valuable to add brief operational details on the warning procedures, civil protection coordination, and the timing of the helicopter evacuation that successfully prevented loss of life, as this is a key takeaway for disaster risk management.
Catchment and subcatchment areas must be reported consistently and explicitly: in lines 103–106, please provide the absolute drainage areas (in km²) for the Upper Étançons and Bonne Pierre subcatchments rather than stating that the former is '2.5 larger', and harmonize the 34.4 km² in line 100 with the 33.8 km² in Table 2 and the 8.7 km² and 3.3 km² values in line 503. Correct the typographical error 'alluvion fan' to 'alluvial fan' (line 130) and 'progacial cone' to 'proglacial cone' (lines 124–128). In lines 139–143, justify the physical rationale for selecting a 48-hour integration window for precipitation and melt (explaining whether it matches storm duration, moving the explanation from lines 211–213 here). In Section 5.1, clarify the hydraulic control and failure threshold at the Les Étages (ETA) station (exact timestamp and stage at 2.87 m / ~80 m³/s where bed aggradation invalidated the rating curve). In the hydrological modeling description, report the Kling-Gupta Efficiency (KGE) alongside the Nash-Sutcliffe Efficiency (0.86) and describe how the -0.6°C/100m temperature lapse rate was verified against high-altitude loggers. In line 594, replace 'we may incriminate the role of the GLOF' with 'we can attribute an important role to the GLOF'. In the figures, increase the font size and contrast on cross-sections AA' and BB' in Figure 4 to ensure elevation labels are legible in print.
In conclusion, this manuscript documents an extraordinary alpine disaster with an invaluable interdisciplinary empirical foundation. The work is already very strong, and the recommendations provided in this report are intended solely as constructive suggestions to assist the authors in refining the structural cohesion, geomorphological depth, and conceptual clarity of their study. By addressing these points, the authors will further strengthen an already important paper, turning it into a benchmark reference for compound and cascading hazards in deglaciating mountain environments.
The depth and detail with which each of the raised points may be addressed are left to the authors’ discretion; on a "base-level straightforward scenario" several specific observations may well be bypassed or addressed simply by adding concise clarifying commentary directly in the text.Declaration on AI tools: Google Gemini 3.7 was used as an assistive tool exclusively for text editing, stylistic polishing, and syntactic refinement of the reviewer's original scientific comments and evaluation.
Sincerely,
Anonymous Referee
Citation: https://doi.org/10.5194/egusphere-2026-971-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,364 | 975 | 114 | 2,453 | 139 | 131 |
- HTML: 1,364
- PDF: 975
- XML: 114
- Total: 2,453
- BibTeX: 139
- EndNote: 131
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comment
This manuscript provides a comprehensive analysis of a complex event that resulted in an extreme debris flood in a small catchment of the Fench Alps in June 2024.
The results point out that the severity of the event stems from its compound nature, i.e. the co-occurrence of different drivers. Such a result is undoubtedly relevant for the interpretation of cascading events in alpine catchments under current climate changes. The conclusions could stress the wider interest of these findings beyond the interpretation of this study case.
This work could become a valuable contribution to NHESS after moderate revision aimed at improving the structure and organization of the manuscript.
I would like to invite the authors to perform a thorough revision of paper structure and organization aimed at making it smoother and easier to follow. The present manuscript, because of its structure, looks like a research report instead of a journal paper. Moreover, although an important part of the work is presented in Appendices, the text is very long, and it has some redundancies. For instance, the Introduction includes a rather long text and some figures that would be more appropriate in the description of the study site. I would recommend checking the entire manuscript for repetitions and redundant text.
Specific comments
Figure 2 does not provide significant information: it could be expunged or moved to the Appendix.
The methods applied to the study of this event would deserve to be described in a separate section, instead of being part of Section 2, which also discusses site and data (Section 2.5 is very short and does not present all methods applied in the study). All relevant methods should be described or referred to by indicating the Appendices where they are described.
Section 2.1
The Melton ruggedness number of the Étançon at La Berarde could be computed: its value, together with the fan slope, would provide some clues about long-term flow processes (debris flow, debris flood, mixed processes?) in the distal part of the torrent (see Bertrand et al., 2013 for thresholds between debris flows and fluvial processes). Of course, no ultimate answers on the type of flow processes that occurred during a specific event can be expected from these simple morphometric indices, but they could be of some interest in site description.
Reference: Bertrand, M., Liébault, F., & Piégay, H. (2013). Debris-flow susceptibility of upland catchments. Natural Hazards, 67(2), 497-511.
Could Sections from 3 to 6 be pooled under the general heading “Results”? This would imply a three-levels heading, but it would result in a better-defined structure of the work.
Section 3.1. Some details on the warning procedures, which permitted rescuing people and avoiding loss of lives, should be reported.
“Drivers of the event” (now Section 4) should be described before “Impact of the event” (now Section 3).
The reconstruction of event chronology, now in section 5.4 and part of the section on river response, should be one of the first reported results. Moreover, event chronology should not be limited to fluvial processes, and it should cover the entire spectrum of the processes involved in the studied event.
Lines 41-43. “The detached… mobile debris flow”. This sentence is redundant: the references to the Himalayan event are sufficient.
Line 72. Baud-Bovy, 1907 is not on the refrences list.
Lines 103-106. I wish to suggest reporting the areas of upper Étançons and Bonne Pierre catchments. The current sentence “Its [upper Étançons] catchment is about 2.5 larger than Bonne Pierre Torrent” does not clarify the extent of these two subcatchments.
Line 130 “alluvion fan” should be corrected to “alluvial fan”.
Lines 130-138. Some rephrasing and more details are necessary. What is the alluvial fan area? Reporting it would permit appreciation that the fan is small in comparison to the catchment size (line 130). The truncation of the fan toe by the Vénéon river should be reported immediately after the second sentence, which mentions the evacuation of sediment by the Vénéon river.
Lines 139-143. Why has a 48-hours duration been considered? Does it correspond to the duration of the rainstorm and snowmelt that triggered the 2024 event (see lines 59 and 211.213)? Please specify. The authors could consider moving here the text at lines 211-213.
Lines 163-164. A reference to the Glacioclim program is necessary.
Line 242. “leading to many damages in the Bérarde village”: damage to the village has already been mentioned. No need to say it again.
Lines 639-641. Two possible references to papers that propose different interpretations on the formation of a cluster of megafans in the eastern Alps:
Jarman, D., Agliardi, F., & Crosta, G. B. (2011). Megafans and outsize fans from catastrophic slope failures in Alpine glacial troughs: the Malser Haide and the Val Venosta cluster, Italy.
Brardinoni, F., Picotti, V., Maraio, S., Bruno, P. P., Cucato, M., Morelli, C., & Mair, V. (2018). Postglacial evolution of a formerly glaciated valley: Reconstructing sediment supply, fan building, and confluence effects at the millennial time scale. Bulletin, 130(9-10), 1457-1473.
Lines 609-620 and A12
The authors note that the observed flow depth at La Bérarde is much smaller than would have been assuming 200,000 m3 debris flow triggered by the sudden drainage of the superglacial lake. Could we argue that the also the type of flow processes observed in the distal part of the channel (i.e. a debris flood instead of a debris flow at La Berarde) indicate that a sudden lake drainage is highly unlikely?
Line 1327 (Torricelli law)
If the authors wish to cite the original work of E. Torricelli, which is probably not strictly necessary, the year of publication (1644) should be reported.
The reference would be Torricelli, E. Opera geometrica, typis A. Masse et L. de Landis, Florentiæ, 1644.