Deep-ocean channel-wall collapse order of magnitude larger than any other documented

The study addresses how and at what scale deep-ocean submarine channel-walls fail, a process that controls sediment, organic carbon, and pollutant transfer to the deep sea and presents geohazards to seafloor infrastructure. While rivers exhibit localised bank failures that influence channel morphology and hazards, submarine channel-wall collapse dynamics in the deep ocean remain poorly constrained. Prior work has focused on proximal slope canyons and channels where levee collapses are generally small and local. The research interprets 3D seismic reflection data from the abyssal plain Hikurangi Channel (water depths ~2600–3600 m) to document and analyse an exceptionally large channel-levee collapse, aiming to refine process models, quantify its morphometry, and evaluate implications for sediment routing and geohazards.

Previous studies of submarine channel-wall and levee failures predominantly concern canyon-to-slope settings within tens of kilometres of canyon heads and typically infer isolated, small-scale piecemeal collapses that partially infill channels and are later cleared by turbidity currents. Large subaqueous landslides on shelves and slopes can remobilise thousands of cubic kilometres, but analogous 규모 from channel-levee failure in deep-ocean channels has been poorly documented. Reported levee-collapse deposits are generally small-scale, affect limited overbank areas, and do not involve erosion of channel-floor high-amplitude reflector (HAR) packages. Comparisons with compiled cases from the literature and the Deep-Marine Architecture Knowledge Store indicate the Hikurangi event exceeds prior channel-related MTDs in thickness, width-to-thickness scaling, and cross-channel extent, and it uniquely occurs far downstream on the abyssal plain rather than on submarine slopes. This gap in observations suggests that large deep-ocean channel-wall failures may be under-recognised due to data limitations and post-event erosion.

The authors analysed 2600 km² of pre-stack Kirchhoff depth-migrated broadband 3D seismic data covering ~150 km of the Hikurangi Channel. Survey resolution is ~25 m horizontally and ~7 m vertically. Seismic interpretation, horizon mapping, surface extraction, and attribute analysis were performed in Petrel. They mapped the top and base of the mass-transport deposit (MTD), calculated preserved volume and thickness distribution, identified seismic facies (chaotic, folded/faulted, and coherent blocks/megaclasts), and delineated kinematic indicators (headwall scarps, ramps, imbricated thrust stacks of HARs). Morphometric measurements (length, area, thickness, volume) were compiled for the MTD and largest megaclasts. Over- and underlying channel forms and overbank sequences were mapped to assess source areas and post-failure incision. To contextualise scale, the team compiled morphometry of other channel-related MTDs from published studies and DMAKS (with supplementary data repository link).

• The mapped MTD is continuous for 68 km, covers >340 km², and has a preserved volume of ~19.3 km³ with maximum thickness ~265 m and median thickness ~50 m.
• Failure spanned the ~2.5 km-wide channel and extended up to ~5 km onto both overbanks, terminating against scalloped headwall scarps interpreted as failure margins, demonstrating a local source.
• Three prominent megaclasts were identified: MC1 and MC2 (adjacent to the left bank) align with overbank stratigraphy and failure-step geometry, indicating retrogressive failure and sliding along decollements; MC3 (right bank, downstream) is a rotated slide block likely sourced from a terrace-bounding weak surface. Megaclast dimensions reach thickness ~120–141 m, long axis up to 4.08 km, and area up to ~3.9 km².
• The MTD eroded pre-existing channel-floor HARs, producing imbricated thrust stacks up to ~40 m thick that were transported across the channel to the opposite margin; in places, underlying HARs are entirely excised beneath the MTD. Two clusters of imbricated HARs indicate opposite transport directions (SE and NE) linked to left- and right-wall failures.
• The coalesced deposit fabric and symmetric preservation across the channel indicate essentially synchronous collapse of both channel-walls along tens of kilometres, a departure from typical piecemeal failure.
• Post-failure, a new channel incised along topographic lows on the MTD, locally scouring to its base, confirming that the original MTD volume exceeded the preserved volume.
• The event represents the largest channel-wall failure yet documented and the first of its scale recognised on an abyssal plain far (>100 km) from slope canyon confluences, requiring a new process model.
• Geohazard implication: erosion depth at least ~40 m, mobilization of channel-floor rafts many tens of metres thick, and overbank remobilization up to twice channel width from margins.
• Sediment routing implication: channel damming trapped at least ~19 km³ of sediment (including organic carbon) within the system, altering downstream flux, analogous to but far larger than known canyon damming cases (e.g., Congo Canyon).

The findings demonstrate that deep-ocean channel-levee systems can undergo quasi-instantaneous, large-scale, synchronous collapse of both channel-walls, generating erosive MTDs that not only inundate overbanks but also rip up and thrust channel-floor deposits. This overturns the prevailing view that channel-wall failures are primarily small, isolated, and slope-confined, and that deep-ocean channels mainly experience piecemeal levee collapses without floor erosion. The mapped kinematics and deposit architecture support a process model involving lateral spreading along weak layers, development of ramps, detachment of megaclasts from different stratigraphic levels, and cross-channel transport sufficient to imbricate HAR stacks. The resultant topography guided re-incision paths and channel stacking, indicating that wall-collapse MTDs can reorganize channel architecture and temporarily dam sediment flux. Although both walls failed essentially coevally, the trigger relationship (simultaneous vs. one triggering the other) remains unresolved. The broader significance includes: revised understanding of abyssal plain channel construction; recognition that deep-water channels can self-generate major geohazards without external slope failures; and the need to reassess fluxes of sediment, carbon, and pollutants considering potential large, channel-internal damming events.

This study documents the largest known channel-wall failure, sourced locally within a deep-ocean channel on the abyssal plain. The event produced a >340 km², ~19.3 km³ MTD with megaclasts up to 4.1 km long and synchronous collapse of both channel-walls, eroding and thrusting channel-floor deposits. These observations necessitate a new process model for large-scale channel-levee collapse and reveal that deep-ocean channels can generate significant internal geohazards and profoundly influence sediment, carbon, and pollutant routing by damming and redirecting flows. Future work should: (1) expand high-resolution 3D seismic surveying of abyssal channels to assess frequency and distribution of such events; (2) integrate geotechnical and pore-pressure analyses to constrain failure planes and triggers; (3) develop quantitative models of channel damming, avulsion potential, and flux sequestration; and (4) refine infrastructure siting and burial guidelines to account for deep-ocean wall-collapse hazards extending beyond channel margins.

• Trigger timing is unresolved: while both walls failed essentially coevally, the study cannot determine if failure was truly simultaneous or if one flank triggered the other.
• Reliance on seismic interpretation: absence of core or well control limits direct lithological and geotechnical confirmation of weak layers and decollement properties.
• Preservation and detection bias: parts of the original MTD were eroded by subsequent channeling; recognition of comparable events elsewhere may be hindered by poorer data quality, limited 3D coverage, or 2D surveys.
• Outcrop and borehole detection challenges mean similar large collapses may go unrecognised, affecting estimates of frequency and generality.
• No direct dating of the failure within the seismic interval limits temporal correlation with potential regional triggers (e.g., earthquakes).