Submarine channels are the largest conveyors of sediment on Earth, yet little is known about their stability in the deep-ocean. Here, 3D seismic data from the deep-ocean Hikurangi channel-levee system, offshore New Zealand, reveal the largest channel-wall failure yet documented. Collapse of both channel-walls along a 68 km stretch created a mass-transport deposit of 19 km 3 , containing 4 km long blocks. Channel-walls typically collapse piecemeal, but here synchronous failure of both channel-walls and landslide erosion of the seafloor is documented, requiring a new process model for channel-wall failure. Mass-failure on this scale poses an under-appreciated risk to seafloor infrastructure both within channels and over regions extending twice the channel width into their overbank. Hitherto, channel-wall failures of this size are unrecognised in abyssal plains; its scale changes our understanding of how channel-levee systems are constructed and how they conduct sediment, carbon and pollutants into the deep-ocean.

Abstract 3D seismic reflection data from the Hikurangi trench-axial channel-levee system, offshore New Zealand, reveal the largest channel-wall failure yet described. Wholesale collapse of both channel banks along a 68 km stretch created a mass transport deposit (MTD) containing blocks up to 4 km long, with a preserved volume of 19 km3. Subsequent deposits have erosional contacts into the MTD, confirming that its original volume was even larger. Channel-walls typically collapse locally, yet synchronous failure of both channel flanks is documented here, implying a catastrophic external trigger. For most of its length, the MTD overlies deposits imaged as high-amplitude reflectors, representing the pre-failure channel floor. Locally these older channel deposits were excavated, transported across and down channel, and then deposited in imbricated stacks, such that the volume of displaced materials exceeded that of the initial failure. Metadata studies of other channel-wall derived MTDs confirm this to be the largest channel-wall collapse documented. MTDs initiated by large-scale levee-collapse like that documented here pose an under-appreciated risk to seafloor infrastructure both within channels and over regions extending at least twice the channel width into their overbank areas.

Deep-water channels can be bound by overbank deposits, resulting from overspilling flows, which are often ornamented with sediment waves. Here, high-resolution bathymetry, backscatter, and 2D and 3D seismic data are integrated to discern the controls on flow processes on the overbank areas of the Hikurangi Channel. Qualitative seismic interpretation and quantitative analyses of sediment wave morphologies and distributions are conducted through the shallowest 600 m of stratigraphy up to the seafloor. Four outer-bend wave fields are present throughout the studied stratigraphy on the landward margin (left margin looking down-channel) only. Originally closely spaced or combined, these fields evolved to become spatially separated; two of the separate wave fields became further subdivided into distinct outer- and inner-bend fields, whose constituent waves developed distinct differences in morphology and distribution. Sediment wave character is used to interpret the direction and strength of overbank flow. Nine controls on such flow and associated deposition are identified: flow versus conduit size, overbank gradient, flow tuning, Coriolis forcing, contour current activity, flow reflection, centrifugal forcing, interaction with externally derived flows, and interaction of overspill from different locations. Their relative importance may vary across parts of overbank areas, both spatially and temporally, controlling wave field development such that: (1) outer-bend wave fields only develop on the landward margin; (2) the influence of centrifugal force on outer-bend overbanks has increased through time, accompanying a general increase in channel sinuosity; (3) inner-bend wave fields on the landward margin form by the interaction of Coriolis-enhanced inner-bend overbank flow, and outer-bank flow from up-channel bends; (4) inner-bend fields on the oceanward margin form by the interaction of axial flow through wave troughs, and a transverse, toward-channel flow component. This work has implications for interpreting overbank flow from seafloor and seismic data, and for palaeogeographic reconstructions from outcrop data.

The type and volume of sediment entering subduction zones affects the style of plate-boundary deformation and thus sedimentary and tectonic cycles. Because submarine channels significantly increase the transport efficiency of turbidity currents, their presence or absence in subduction trenches is a primary control on trench fill. To date, comprehensive architectural characterization of trench-axial channels has not been possible, undermining efforts to identify the factors controlling their initiation and evolution. Here, we describe the evolution of the Hikurangi Channel, which traverses the Hikurangi Trench, offshore New Zealand. Analysis of two- and three-dimensional seismic data reveals that the channel was present only during the last ~3.5 m.y. of the ~27 m.y. of the trench’s existence; its inception and propagation resulted from increased sediment supply to the trench following amplified hinterland exhumation. To test if the controls on the evolution of the Hikurangi Channel are universal, multivariate statistical analysis of the geomorphology of subduction trenches globally is used to investigate the formative conditions of axial channels in modern trenches. Terrigenous sediment supply and thickness of sediment cover in a trench are the dominant controls; subsidiary factors such as trench length and rugosity also contribute to the conditions necessary for trench-axial channel development. Axial channels regulate sediment distribution in trenches, and this varies temporally and spatially as a channel propagates along a trench. The presence of a trench-axial channel affects plate-boundary mechanics and has implications for the style of subduction-margin deformation.

Monitoring of modern deep-water channels has revealed how migrating channel-floor features generate and remove stratigraphy, improving understanding of how channel morphologies relate to their deposits. Here, seafloor and subsurface data are reconciled through an integrated study of high-resolution bathymetry and three-dimensional seismic data imaging a ca 150 km stretch of the trench-axial Hikurangi Channel, offshore New Zealand. On the seafloor, terraced channel-walls bound a flat, wide, channel-floor, ornamented with three scales of features that increase then decrease in longitudinal gradient downstream, and widen downstream: cyclic-steps, knickpoints and knickpoint-zones (in increasing size). Mass-transport deposits derived from channel-wall collapse, are bordered by wide and flat reaches of channel-floor upstream and by knickpoint-zones (reaches containing multiple knickpoints) downstream. In the subsurface, recognition of ten seismofacies and five types of surface enables identification of four depositional elements: channel-fill, sheet or terrace, levee, and mass-transport deposits. Integration of subsurface and seafloor interpretations reveals that knickpoint-zones initiate on the downstream margins of channel-damming mass-transport deposits; they migrate and incise through the mass-transport deposits and weakly-confined deposits formed upstream, as the channel tends towards equilibrium. Downstream of a knickpoint-zone, a flat channel-floor is bounded by newly-formed terraces. Knickpoints migrate by eroding upstream and depositing downstream, generating filled concave-up (cross-sectional) surfaces in their wake. Within knickpoint-zones, knickpoint-generated surfaces are re-incised by subsequently-passing knickpoints to produce a composite bounding surface; this surface does not delineate the morphology of any palaeo-conduit. The Hikurangi Channel's subsurface architecture records the localized erosional response to mass-transport deposit emplacement via knickpoint-zone migration, showcasing how transient seafloor features can build channelized stratigraphy. This model provides an additional mechanism to conventional models of channel deposit formation through 'cut-and-fill' over long stretches of channel. These findings may aid subsurface interpretation in systems lacking a contemporary self-analogue or with poor data coverage.

Information on acquisition, processing, and interpretation of seismic data, and the statistical methodology and sensitivity analysis.

ABSTRACTThe inception and evolution of channels in deep-water systems is controlled by the axial gradient and lateral confinement experienced by their formative flows. These parameters are often shaped by the action of tectonic structures and/or the emplacement of mass-transport deposits (MTDs). The Arro turbidite system (Aínsa depocenter, Spanish Pyrenees) is an ancient example of a deep-water channelized system from a bathymetrically complex basin, deposited in an active tectonic setting. Sedimentologic fieldwork and geologic mapping of the Arro system has been undertaken to provide context for a detailed study of three of the best-exposed outcrops: Sierra de Soto Gully, Barranco de la Caxigosa, and Muro de Bellos. These locations exemplify the role of confinement in controlling the facies and architecture in the system. Sedimentologic characterization of the deposits has allowed the identification of fifteen facies and eight facies associations; these form a continuum and are non-unique to any depositional environment. However, architectural characterization allowed the grouping of facies associations into four depositional elements: i) weakly confined, increasing-to-decreasing energy deposits; ii) progradational, weakly confined to overbank deposits; iii) alternations of MTDs and turbidites; iv) channel fills. Different styles of channel architecture are observed. In Barranco de la Caxigosa, a master surface which was cut and subsequently filled hosts three channel stories with erosional bases; channelization was enhanced by quasi-instantaneous imposition of lateral confinement by the emplacement of MTDs. In Muro de Bellos, the inception of partially levee-confined channel stories was enhanced by progressive narrowing of the depositional fairway by tectonic structures, which also controlled their migration. Results of this study suggest that deep-water channelization in active tectonic settings may be enhanced or hindered due to: 1) flow interaction with MTD-margin topography or; 2) MTD-top topography; 3) differential compaction of MTDs and/or sediment being loaded into MTDs; 4) formation of megascours by erosive MTDs; 5) basin-floor topography being reset by MTDs. Therefore, the Arro system can be used as an analog for ancient subsurface or outcrop of channelized deposits in bathymetrically complex basins, or as an ancient record of deposits left by flow types observed in modern confined systems.

Exhumed basin margin‐scale clinothems provide important archives for understanding process interactions and reconstructing the physiography of sedimentary basins. However, studies of coeval shelf through slope to basin‐floor deposits are rarely documented, mainly due to outcrop or subsurface dataset limitations. Unit G from the Laingsburg depocentre (Karoo Basin, South Africa) is a rare example of a complete basin margin scale clinothem (>60 km long, 200 m‐high), with >10 km of depositional strike control, which allows a quasi‐3D study of a preserved shelf‐slope‐basin floor transition over a ca. 1,200 km 2 area. Sand‐prone, wave‐influenced topset deposits close to the shelf‐edge rollover zone can be physically mapped down dip for ca. 10 km as they thicken and transition into heterolithic foreset/slope deposits. These deposits progressively fine and thin over tens of km farther down dip into sand‐starved bottomset/basin‐floor deposits. Only a few km along strike, the coeval foreset/slope deposits are bypass‐dominated with incisional features interpreted as minor slope conduits/gullies. The margin here is steeper, more channelized and records a stepped profile with evidence of sand‐filled intraslope topography, a preserved base‐of‐slope transition zone and sand‐rich bottomset/basin‐floor deposits. Unit G is interpreted as part of a composite depositional sequence that records a change in basin margin style from an underlying incised slope with large sand‐rich basin‐floor fans to an overlying accretion‐dominated shelf with limited sand supply to the slope and basin floor. The change in margin style is accompanied with decreased clinoform height/slope and increased shelf width. This is interpreted to reflect a transition in subsidence style from regional sag, driven by dynamic topography/inherited basement configuration, to early foreland basin flexural loading. Results of this study caution against reconstructing basin margin successions from partial datasets without accounting for temporal and spatial physiographic changes, with potential implications on predictive basin evolution models.