Slope failure dynamics and impacts from seafloor and shallow sub-seafloor geophysical data: case studies from the COSTA project

Holocene and slightly pre-Holocene submarine landslide are found both in high-latitude glacial-dominated margins and in lower latitude, river-dominated margins. This paper constitutes a major assessment on some of the best-studied submarine instabilities in the world. We review and update from original data and literature reports the current state of knowledge of Storegga, Traenadjupet and Finneidfjord slides from the mid-Norwegian margin, Afen Slide from the Faeroe-Shetland Channel, BIG'95 Slide and Central Adriatic Deformation Belt (CADEB) from continental slope and inner continental shelf settings off the Ebro and Po rivers in the Mediterranean Sea, Canary Slide west of the westernmost, youngest Canary Islands and Gebra Slide off the northern tip of the Antarctic Peninsula in the southern hemisphere, i.e. those studied in the Continental Slope Stability (COSTA) project. The investigated slides range in size from the gigantic 90,000 km(2) and almost 3 000 km(3) Storegga Slide to the tiny 1 km(2) and 0.001 km(3) Finneidfjord Slide. Not only do individual submarine landslides rarely involve processes precisely fitting with pre-established categories, mostly based on subaerial research, but also they display complex mechanical behaviors within the elastic and plastic fields. Individual events can involve simultaneous or successive vertical to translational movements including block detachment, block gliding, debris flow, mud flow and turbidity currents. The need for an in-depth revision of the classification criteria, and eventually for a new classification system, based on the new imaging capabilities provided by modem techniques, is more than obvious. We suggest a new system, which, for the moment, is restricted to debris flows and debris avalanches. Volume calculation methods are critically reviewed and the relations between some key geomorphic parameters are established for the selected slides. The assumed volume missing from scar areas does not necessarily match the actual volume of sediment remobilised by an individual event since in situ sediment can be remoulded and eventually incorporated during the slide downslope journey. CADEB, a shore-parallel prodelta detached from its source, is the exception to the good correlation found between across slope width and alongslope length with slide area. Height drop measured from the headwall upper rim to its foot correlates with the debris deposit maximum thickness unless the slide moves into restricted areas, which prevent farther forward expansion of the deposit, such as Gebra and BIG'95. In such cases, "over-thickened' deposits are found. A particularly loose and fluid behavior can be deduced for slides showing an "over-thinned' character, such as the Canary Slide that traveled 600 km. Scar areas and slip planes have been investigated with particular emphasis. Although slide headwalls might present locally steep gradients (up to 23degrees for Storegga Slide), the slope gradients of both the failed segment margins and the main slip planes are very low (max. 2degrees and usually around V and less). An exception is the Finneidfjord Slide (20degrees-5degrees) that occurred in 1996 because of a combination of climatic and anthropogenic factors leading to excess pore pressure and failure. Mechanically distinct, low permeable clayey "weak layers" often correspond to slip planes beyond the slide headwall. Since not only formation of these "weak layers" but also sedimentation rates are climatically controlled, we can state that slide pre-conditioning is climatically driven too. Run-out distances reflect the degree of disintegration of the failed mass of sediment, the total volume of initially failed material and transport mechanisms, including hydroplanning. Commonly, specific run-outs could be attributed to distinct elements, such as cohesive blocks and looser matrix, as nicely illustrated by the BIG'95 Slide. Total run-outs usually correspond to matrix run-outs since the coarser elements tend to rest at shorter distances. Outrunner blocks are, finally, a very common feature proving the ability of those elements to glide over long distances with independence of the rest of the failed mass. In addition to pre-conditioning factors related to geological setting and sedimentation conditions, a final trigger is required for submarine landslides to take place, which is most often assumed to be an earthquake. In high latitude margins, earthquake magnitude intensification because of post-glacial isostatic rebound has likely played a major role in triggering landslides. Although it cannot be totally ruled out, there is little proof, at least amongst the COSTA slides, that gas hydrate destabilisation or other processes linked to the presence of shallow gas have acted as final triggers. (C) 2004 Elsevier B.V. All rights reserved.