"Landslide" is something of a misnomer for many of the events in this category, since they may involve the rapid transport of a fragmented mass of rock or sediment by mechanisms more akin to avalanches or fluid flows; although in other cases the movement of a coherent mass on a basal slip surface or zone is sufficiently rapid to generate tsunami. The term "submarine mass movement" is therefore also sometimes used but the more familiar term "landslide" will be used here in a loose sense. Whatever the process of mass movement involved, two distinct tsunami generation mechanisms can be distinguished. Submarine landslides involve the movement of sub - seafloor material from one place to another, generating a reverse flow of water ("dipole mechanism") FIGURE ILLUSTRATING THIS. Subaerial landslides entering the sea, on the other hand, push water laterally as they do so ("monopole mechanism") FIGURE and only generate a reverse flow once they are completely submerged and continue to move. The lateral push as these landslides enter the water is a much more efficient mechanism for generating tsunami: for example, in the 1994 Skagway tsunami, generated by the collapse of a pier and its foundations into a fjord, only 10% of the displaced mass was initially above water level but computer modelling indicates that the lateral push as this mass entered the water doubled the height of the wave produced, to over 40 feet, over that which would be expected from the movement of the 90% of the landslide which was initially underwater.

Wave size increases with landslide volume, not only because of the obvious reason that the greater the landslide volume, the greater the energy released by unit amount of movement, but also because larger landslides experience less basal friction and turbulent drag in relation to their size and therefore move further and faster.

Wave - generating efficiency increases with speed, typically to the point where the landslide velocity equals the velocity of the waves that it produces, at which wave resistance alone will prevent further acceleration. This velocity depends, as discussed below, (SECTION CROSS - REFERENCE) on water depth and wavelength of the dominant wave frequency, which in turn will be broadly dependent upon landslide size. Landslides in shallow water will tend to produce larger waves, but for large landslides at least the effect will be reduced somewhat by their continued acceleration in deeper water.

Coherent sliding blocks (commonly referred to as slump blocks) usually move more slowly than avalanche - like mass movements. However, the very largest blocks, with dimensions of kilometres to tens of kilometres, in ancient submarine landslides moved as far (and, by inference, as fast) as the rest of the rock avalanches with which they are associated. Geological evidence indicates that this is because these blocks trap a layer of pressurised mud and debris beneath them upon which they effectively float by a "hovercraft" effect and so slide without basal friction.

The greater the slope, the greater the component of gravitational force acting down it and so the greater the accelerating force. Subaerial landslides have the additional feature that their initial acceleration is in air (which is less dense and much less viscous, and so offers less resistance to movement) and so they may be travelling at much higher speed when they enter the water than an equivalent submarine landslide would after the same amount of movement, by a factor of 2 to 3.

The distribution of submarine and initially subaerial tsunamigenic landslides is complex owing to the wide variety of rock and sediment types which they affect (SECTION CROSS - REFERENCE). In recent years the development and use of scanning sonar mapping and imaging devices has led to the identification and mapping of a wide variety of submarine landslide types and the recognition that they can occur on a vast scale; and also to the reinterpretation of many facets of the onshore geology of volcanic islands and uplifted and eroded continental margin sediment sequences.

In which masses of sediment with volumes of thousands to tens of thousands of cubic kilometres are disturbed by earthquakes, rapid sediment deposition or the release of gas hydrates and move down the relatively gentle slopes (typically 0.5 - 5 degrees) of the so - called passive continental margins as particulate flows. Passive continental margins, such as the Atlantic margins of Europe, North America and Africa, are characterised by a lack of major deformation and infrequent earthquakes, allowing the accumulation of thick sequences of unstable sediment. Notable examples of these events include the three Storegga slides off the coast of Norway between 20 000 and about 8 000 years ago, the last of which produced notable tsunami deposits from north Norway to at least as far south as northern Scotland. DIAGRAM (OR PHOTO FROM AD).

Which have affected many groups of islands but are best known from the Hawaiian and Canarian archipelagoes DIAGRAM SHOWING EXTENTS OF DEPOSITS AND COLLAPSE SCARS. The combination of thick sequences of weak volcanic rocks forming steep slopes (5 to 25 degrees) and a variety of perturbing forces ranging from earthquakes to the destabilising effects of heated and pressurised groundwater within the volcanoes result in major landslides, with volumes of hundreds to thousands of cubic kilometres. Past landslides from these islands form avalanche deposits extending hundreds of kilometres over the flat ocean floor. They contain massive slide blocks such as the Tuscaloosa "seamount" block off Oahu in the Hawaiian islands: this is over 2 km high, has a mass of the order of 10¹² (one million million) tons and probably attained a peak velocity of the order of 100 metres per second during the landslide in which it was emplaced. These landslides occur on relatively steep slopes and, most importantly, normally begin as subaerial landslides which leave huge landslide scars on the islands PHOTO: they are therefore particularly efficient tsunami sources. Deposits left by past collapse - generated tsunami are found hundreds of metres above sea level on adjacent islands (PHOTO, AGAETE DEPOSIT IN GRAN CANARIA). More controversially, such collapses have been linked to deposits left by giant tsunami on the opposite sides of oceans: if these connections are correct then these are by far the largest tsunami generated by terrestrial events.

On the steep slopes (5 to, rarely, 40 degrees) between continental margins and the deep ocean trenches above subduction zones. Submarine landslides, with volumes of up to several hundreds of cubic kilometres, are common at certain oceanic trenches (most notably the East Japan, Peru and Puerto Rico trenches) as a result of earthquake - triggered instability of sedimentary rock sequences tilted towards the trench by progressive deformation. Although some involve rapid movement of debris avalanches, the dominant type of mass movement appears to be sediment slides or slumps. Ongoing theoretical studies and marine geological surveys indicate that the 1998 Sissano (Papua New Guinea) tsunami, with a maximum runup of about 20 metres, was generated by a relatively small (about 50 km³) slump landslide of this type. If the slump had evolved into a rapid avalanche - type landslide, it could have generated a tsunami with runups of up to 90 metres! However, although relatively common, these landslides are relatively inefficient tsunami sources because they are entirely submarine and are usually initiated in relatively deep water.

The margins of coral reef platforms, such as the Bahamas and the Australian Barrier Reef, form submarine cliffs up to 4 km high. Prehistoric collapses of these, with volumes of tens to hundreds of cubic kilometres, have produced notable debris avalanche deposits around them and can be inferred to have been efficient tsunami sources (steep slopes, avalanche - type movements, and initiation near sea level).

Most of these types of giant event are known only from the geological record and appear to occur relatively infrequently (on average over periods of geological time): however, as discussed later (SECTION CROSS - REFERENCE) prediction of their frequency in the near future is complicated by evidence that their occurrence is influenced by environmental factors and cannot be assumed to be uniform over time.

Other, smaller landslide events have generated a significant proportion of historic tsunami, perhaps around 20% in total. These include relatively small continental slope sediment failures (such as that triggered by an earthquake off Newfoundland in 1929, which produced a tsunami on the coasts of Newfoundland and Nova Scotia which claimed over 20 lives along a very sparsely populated coastline); trench slope failures (such as the 1998 Papua New Guinea tsunami, as noted above, but probably also the 1993 Sanriku tsunami in Japan which claimed 3000 lives); and, possibly, carbonate platform collapses (the 1867 St. Croix earthquake in the Caribbean, a relatively small event which produced a large tsunami along the margin of the Virgin Islands carbonate platform). Other historically recorded landslide - generated tsunami have been produced by slightly different types of landslide which have more limited maximum volumes:

The ash - rich stratovolcanoes which form above subduction zones (and which define, for example, the circum-Pacific "ring of fire") are much smaller than oceanic island volcanoes but are significantly steeper and undergo much more frequent, but smaller, lateral collapses. Many of these collapses have entered the sea and generated tsunami which are particularly dangerous when the volcanoes concerned are located around, or within, enclosed bays. A notable example is that of a small (about 0.2 cubic kilometres) collapse on the flank of Unzen volcano, Japan, in 1792: this produced a debris avalanche which buried the town of Shimabara, killing about 10 000 people, and then entered Kagoshima bay and produced a tsunami which killed a further 5 000 people around the bay. Other, somewhat larger collapses of island volcanoes in the open sea have also produced notable tsunami: that of Ritter Island (Papua New Guinea) in 1888 produced tsunami with runups of 15 m on coastlines around 100 km distant and of 5 m at Rabaul, over 400 km distant AERIAL PHOTO OF RITTER TODAY. These collapses are efficient tsunami sources because they typically start above sea level and accelerate rapidly on the steep slopes of the volcanoes.

The steep fronts of river deltas, especially where these are built out into bodies of relatively deep water such as fjords, are particularly prone to collapse because they are composed of weak and poorly consolidated sediments. Collapses may occur as a result of loading by new layers of sediment during floods; loading by construction or other human activities; or the shaking induced by major earthquakes. Perhaps the most notable recent example was at Valdez, Alaska during the 1964 Alaska earthquake: a large section of the delta upon which the town was built collapsed into the fjord and generated a tsunami with maximum runup of over 60 metres. Subsequent investigations showed that the delta front was so unstable that the remaining part of the town was abandoned and a new town built at the side of the delta. Delta front collapses in fjords are a particular problem because the deltas are in general the only available areas of flat ground and therefore tend to be intensively developed. The damage from the tsunami is therefore often compounded by the loss of whatever was located on top of the collapsed area. The most notable historical example of this type of event involved the sandbar upon which the 17^th Century town of Port Royal (Jamaica) was built: when part of this collapsed during an earthquake in 1692, the resulting tsunami completed the destruction of the town and also damaged adjacent areas.

Rock avalanches and slides on steep slopes around enclosed bodies of water, most especially fjords and mountain lakes, have produced some of the most spectacular tsunami events recorded. A large rock avalanche from the side of Lituya Bay in Alaska, on 9^th July 1958, produced a surge of water which travelled up the opposite side of the bay to a height of 540 m above sea level, the highest ever recorded.

FIGURE ---- MAP OF LITUYA BAY from USGS PP432C

The surge of water at Lituya Bay, and comparable events caused by other rock slope failures, are commonly quoted as the highest tsunami runups recorded. However, these events are not strictly comparable to tsunami generated in the open sea, their extreme runup heights being a reflection of the confined bodies of water in which they occur. Any wave which escapes from narrow fjords such as Lituya Bay swiftly dissipates. As a result, statistical analysis and interpretation of tsunami catalogues containing such events should be carried out with care. Nevertheless, these events have the potential to cause very severe local damage along the shores of the fjords or lakes, and also downstream in the case of naturally or artificially dammed lakes. Landslide - generated tsunami in the latter are particularly dangerous when they overtop the dams and cause catastrophic floods downstream (as at Vaiont, Italy, in 1960).