Climate Extremes
Earthquakes
Floods
Climate Change
Volcanoes
Tsunamis
Indian
Ocean Tsunami
Landslides
Disaster Studies & Management
Rapid Environmental Impact
Assessment in Disaster Response
Intranet
| Home Contact us Gallery Site Map Resource Centre Search | ||
 |
||
| About us People News Publications Education & Training Events | ||
| About Tsunamis | Project Pages | Indian Ocean Tsunami |
More about Mega-Tsunamis
Attention is currently focused on the Cumbre Vieja volcano on the Canary Island of La Palma ( images ; Q & A ; papers), the western flank of which has started to slide seawards. This volcano rises 6 km from the ocean floor - with only the top 2 km emerging above sea level - has grown to this size over the past 125 000 years or so, and has had seven eruptions in the past five centuries. During the last eruption right at the summit of the volcano, in 1949, violent earthquakes accompanied the formation of a fault-fissure along the crest of the volcano that displaced the west flank of the volcano downwards and towards the ocean.Geological mapping of the fault and the rocks around it have defined its full extent at the surface, over 4 km long and with a maximum down-to-the-west offset of around 4 m, and also revealed that it was the first such fault to have cut the volcano in the time represented by the many different lava flows and ash beds exposed near the summit. Radiometric dating of the rocks, undertaken at the Climate and Environmental Science Institute at Gif-sur-Yvette, Paris, subsequently showed that it was the first such fault to have ruptured the surface of the volcano for at least 25,000 years. The fault rupture was therefore a very unusual event, unlike the episodes of fissuring that accompany every eruption of the volcano, which supported the idea that it represented the first appearance of a large deep structure at the surface. As faults typically grow in the subsurface and only rupture the surface when they have grown to a large size, it appeared likely that the rupture was just the surface representation of a fault that extended deep beneath the surface.
This idea was supported by detailed mapping of the volcano. Oceanic island volcanoes such as the Cumbre Vieja typically do not have large summit craters. Instead, they normally have arrays of small volcanic vents, known as volcanic rift zones, running down the flanks from the summit in a three-pointed star pattern. This geometry results from the stresses generated by rising magma deep beneath the volcano, that cause it to inflate and then split along cracks beneath the rift zones. The mapping showed that this star pattern had been present on the Cumbre Vieja as well, from its first stage of growth until between 15,000 and 8,000 years ago. In that time period, first the northwest rift zone and then the northeast rift zones became extinct: but after 8000 years ago the third, south rift zone propagated north through the summit and now bisects the volcano. Most recently, largely in the historic eruptions, new fissure systems have developed running downslope on the western flank of the volcano. This pattern suggests that the volcano is splitting apart and that as the western side moves towards the sea it has begun to bulge and fracture. This fundamental change in the structure of the volcano requires that a weak zone be present underlying the whole of its western flank. The obvious candidate for this zone of weakness is a curved, seaward-dipping fault, perhaps 15 to 25 km wide and extending from the crest of the volcano to as much as 10 km offshore. The geometry of this fault is comparable to the shape of past collapse scars, suggesting that eventually the mass of rock above, with a volume of at least 200 km3 and perhaps up to about 500 km3, will slide into the ocean in a giant landslide.
It is unlikely, however, that the collapse is imminent. Theoretical studies, by Derek Elsworth of Penn State University, of how these landslides are triggered indicate that the forces generated by intrusion of magma into the volcano, ranging from the direct pressure of the rising magma to (perhaps the most significant) pressurisation of trapped groundwater as it is heated by the magma, are necessary to trigger collapse. Elsworth and I have analysed the time taken for these forces to build up and we predict that collapse of a volcano like the Cumbre Vieja is most likely to occur several days to several months after the start of an eruption. As at Mount St. Helens, the collapse is likely to be preceded by progressively accelerating deformation of the unstable flank. Thus, there will be plenty of short-term indications that a collapse may be about to occur, although successful interpretation of these will require detailed monitoring of the volcano. Furthermore, although we cannot say whether the volcano will fail in its next near-summit eruption (like that in 1949; a small eruption in 1971 at the very southern end of the island seems to have had relatively little effect, probably because the magma did not rise so high in the volcano) or only after several more eruptions have progressively weakened it, since eruptions of the Cumbre Vieja occur at intervals of a few decades to as much as a few centuries the year-to-year probability of failure is relatively low. The "half-life-to-failure" of the volcano, if things continue as they are, might be as much as 5 000 years - but could be much less.
These theoretical conclusions have received support from geodetic monitoring of the Cumbre Vieja, carried out between 1994 and 1998 by Bill McGuire and his students with the aim of detecting any ongoing creep of the flank of the volcano. Whilst the flank moved by at least 4 metres in the space of a few days during the 1949 eruption, in the 5 years of geodetic monitoring no movement was detected outside the errors of the method used, of around a few centimetres. Similarly, the geological evidence is that no movement of more than a few centimetres has occurred at the surface fault rupture (although this does not exclude movement at depth) in the 5 decades since 1949. Thus - again assuming that the factors controlling the behaviour of the volcano remains uniform - there seems no significant risk that the volcano will collapse spontaneously.
The bad news concerns what is likely to happen when it does. With one questionable exception, all the historic collapses of small island volcanoes (in volcanic arcs, not the oceanic island type) have been catastrophic, in the sense that a single large mass of rock fails and slides away into the ocean at high speed. Two collapses in particular, those of Oshima-Oshima (Japan) in 1741 and of Ritter Island (New Guinea) in 1888, produced large regional scale tsunamis, with wave runups of at least 15 m and perhaps in excess of 30 m up to 100 km distant and with significant tsunami damage recorded up to 1000 km away. Each collapse produced a landslide with a volume of 3 to 4 km3: less than 1% of the volume of the landslides produced by typical oceanic island volcano collapses!
The great run-out distances of these giant landslides over the ocean floor, of up to 200 km, and their tendency to go over rather than around obstacles, implies that they too involve catastrophic volcano failure leading to landslides with velocities of up to 100 m/s (360 km/h or over 200 miles per hour) or more. It has been proposed that the Hawaiian landslides produced immense tsunamis that have laid down the coral boulder and gravel deposits that occur hundreds of metres above sea level on the sides of some of the islands. This suggestion proved controversial, as the Hawaiian islands rise and fall relative to sea level as the oceanic crust flexes under the enormous weight of the growing volcanoes and so these deposits could be raised beaches. The Atlantic volcanic islands and much of the length of the continental margins around this ocean, on the other hand, are much more stable and so the occurrences of recent marine fossil bearing deposits high above sea level that we are currently investigating are much more likely to be unequivocal indicators of the past occurrence of giant tsunamis. Some of these deposits are truly spectacular: for example, 123 000 year old deposits in the Bahamas that we have proposed to have been produced by tsunamis from the El Golfo I collapse in the Canaries, which is of the same age, include boulders weighing up to 2000 tons that were displaced up to 20 metres above sea level and at least half a kilometre inland.
Another approach to estimating the scale of the tsunamis that might be produced by a future collapse of the Cumbre Vieja volcano is to use computer modelling. Computer models of tsunamis produced by earthquakes have been developed to a high degree of sophistication and have been tested extensively by comparison with historic tsunamis. Landslides, by comparison with earthquakes, are much more complex tsunami sources and correspondingly more difficult to model in detail. Nevertheless, the sheer scale and implied speed of landslides from oceanic island volcano collapses - with total energies comparable to those released by the impact of moderate-sized asteroids, 200 to 500 metres in diameter - indicates that they should produce very large waves. In fact, fast landslides from oceanic island volcanoes are likely to be much more efficient tsunami sources than either the slower sediment slides that occur on the continental slopes and are also noted tsunami sources, or asteroid impacts in the ocean, which convert most of their energy into heat, light and vaporised rock and water. A variety of theoretical analyses and laboratory-scale experiments, notably those carried out by Hermann Fritz and colleagues at ETH-Zurich, indicate that initial wave heights of several hundred metres to over a kilometre are reasonable over a range of oceanic volcano collapse parameters (size, speed, thickness and so on). These waves will of course decrease in height as they move out from their source, through geometric spreading and a process known as wave dispersion, and this has lead to the (perhaps wishful?) suggestion that the waves will lose most of their energy quite rapidly and therefore not present a major hazard at distances of more than some hundreds of kilometres from source.
A computer model of a future collapse on La Palma (click here to go to pdf of paper) by Steven Ward of the University of California at Santa Cruz and Simon Day of BHRC, indicates that this is not the case. Although wave dispersion effects do significantly modify the waves, nevertheless the results indicate that these will retain a significant proportion of their energy as they propagate outwards from the Canaries (where their initial heights are around a kilometre, in agreement with the other independent predictions by Fritz and others) towards the USA, Europe and northern Brazil. Tsunamis travel at high speeds in the deep ocean ---- as fast as passenger jet aircraft ---- and then slow down and pile up, increasing their height, as they enter shallow water. The upshot of the model is that it predicts that between 6 and 9 hours after the collapse of the Cumbre Vieja, tsunami waves with amplitudes of around 50 metres will strike the entire western seaboard of the Atlantic: these values are consistent with the size of the giant boulders and other deposits in the Bahamas, lending support to the model. Hours before the waves arrive in America, the coasts of the Canaries and of western Africa and Europe will have been swept by waves that have refracted around the submarine flanks of La Palma. This last is a complex process, and so it is difficult to predict the size of waves that will strike Europe in particular: but the model predicts that the waves in the Canaries may run up to several hundreds of metres above sea level on the steep slopes of the islands.
A series of output images from a computer model showing predicted propagation of tsunami waves across the Atlantic from a possible future collapse of the Cumbre Vieja volcano. Red bands denote wave crests, blue bands denote wave troughs; yellow dots indicate heights of waves above/below initial sea level at particular spots. Note that these are mostly deep water wave heights and will amplify by factors of 2 to 5 or more as they approach coastlines (from Ward & Day, 2001 in the press).
Much remains to be done in understanding the mechanisms and effects of these oceanic island volcano collapses and the resulting tsunamis, and still more in monitoring the volcanoes and predicting future collapse occurrences. Nevertheless the work so far suggests that the phenomena may form a third member of a group of rare but very large geophysical events, the other two of which are giant explosive volcanic eruptions, and asteroid and comet impacts, and may be more frequent than either. All three thus present a comparable threat in terms of loss of life averaged over very long time periods. This long-term average mortality rate is arguably greater in all three cases than the average rate of loss of life in many types of 'everyday' event that are considered to present a significant risk, for example air or rail crashes.
For further information contact Dr. Simon Day at: s.day@ucl.ac.uk
| > | Natural Hazards for Insurers Certificate Course |
| > | Masters in Geophysical Hazards |
| > | BUHRC Workshops |
| > | Tropical Cyclones |
| > | UK & European Weather |
| > | North Atlantic Oscillation (NAO) |
