Tsunami Sediments and Flood Run-Up

The Second Storegga Slide Tsunami

Evidence for a regionally extensive layer of marine sediment within isostatically-uplifted coastal sediments in eastern Scotland has been known for over 20 years (Sissons and Smith, 1965) and, until recently, was attributed to a major storm surge (Smith et al., 1985). More recently, it was suggested that the marine sediments were deposited by a tsunami associated with the Second Storegga Slide (Dawson et al., 1988). Later research has shown that this sediment layer is, in fact, very complex in terms of both its sediment composition (ranging from a coarse gravel to a clay/silt) and its sedimentary structures. In most cases the base of the sediment layer rests upon an eroded peat surface. Frequently, the sediment accumulation contains intraclasts of peat that have been eroded and transported in suspension during the tsunami inundation. In addition, the deposit is often characterised by the presence of large numbers of marine and brackish water diatom species of which the most common is the marine diatom Paralia sulcata, (Ehrenberg) Cleve although other species include Cocconeis Acute Ehrenberg, Diploneis spp., Grammatophera oceanica Ehrenberg Crun and Hyalodiscus stellinger Bail (Smith et al., 1985). Very large numbers of diatoms (60-80%) are broken and thus may indicate that the sediments were deposited under high-energy conditions. An alternate view is that the proportions of diatom species that are present within the deposit reflect the preservation of more robust centric (spherical) species (for example Paralia sulcata) and the destruction of delicate pennate (elongate) species during tsunami transport (S. Dawson, pers. comm.).

The sediment layer exhibits little evidence of having been sorted. The sand layer reaches a maximum thickness of up to 1.4m in parts of NE Scotland. In general the sand layer occurs within coastal sediments located above sea level in the Scottish mainland while in peripheral areas (e.g. Orkney, Shetland and northern England), decreased crustal uplift in conjunction with a rise in sea level has resulted in the occurrence of the tsunami layer below present day sea level. Indeed, farther south near Amsterdam, where the rise in sea level has been associated also with land subsidence, the layer appears to have been deposited in association with a relative sea level of -15m (De Groot, pers. comm.). In Scotland, the altitude of the estuarine surface beneath the layer has been used to infer regional patterns of relative land uplift that have taken place since 7,000 years ago as a result of differential crustal rebound caused by the melting of the last ice sheet. If the different values so far obtained are described using trend surface analysis, the differential uplift of these areas since the time of the tsunami can be demonstrated (Figure 2). The contoured surface shown in Figure 2 has been used as the basis from which calculations of tsunami runup can be made. It should be noted that the estimate of tsunami runup is not affected by the amount of crustal subsidence or uplift that has taken place since 7,000 years ago nor by the position of relative sea level at that time.

In Scotland, the tsunami sediment layer has been radiometrically dated at over 15 separate sites (Smith et al., 1991) with the 14C ages showing a distinct clustering at circa 7,000 years. Stratigraphic investigations have shown that it is possible to calculate the run-up of this tsunami flood for each of the sites investigated. These show typical run-up values of between +4 and +6 m (cf. Smith and Dawson, 1990). Large local variations in coastal flood run-up are to be expected, however, due to the effects of wave resonance and amplification within individual inlets. In general, however, the estimates of tsunami run-up based on the maximum altitudes of marine sediment layers exhibit a good agreement with theoretical calculations of flood run-up based on the numerical models of Harbitz (1991, 1992) and Henry and Murty (1992). However, the calculations of Harbitz (1992) probably underestimate the amount of coastal runup. Harbitz (1992) provided estimates of the wave period and wave amplitude associated with the tsunami caused by the Second Storegga Slide using the assumption that the slide moved at an average velocity of 35 metres per second. He showed that wave period ranged between 2.05 and 2.83 hours while the values for wave amplitude ranged between 1.2 and 3.3m.

Palaeotidal Regime

The impact of tsunami waves on any coastline will of course be greatly affected by the state of the tide at the time they strike. This will be especially marked when the tidal range is at its greatest. Along the North Sea coast of the United Kingdom, tidal range is generally in the region of two to three metres, rising in estuaries. Thus, a circa five metre high tsunami wave will have much greater impact if it strikes the coast during high tide rather than during a low tide, where, in any case, the energy of the waves are more likely to be dissipated across a broader intertidal zone. Although the interval of time between individual tsunami waves may be half an hour or more, it is only the first few waves which are of any great size. Consequently, the flood impact of a series of tsunami waves along a particular stretch of coastline may take place within circa one or two hours. In the North Sea, the tidal changes rotate about amphidromic points and circulate about the basin in an anti-clockwise direction at the present time. For example, at present it is high tide in the Forth estuary when it is low tide in the Shetland Isles.

Detailed stratigraphic evidence from Scotland of the sand layer attributed to the Second Storegga Slide (e.g. Morrison et al., 1981; Smith and Cullingford, 1985) provide the basis for an estimate of the tidal conditions when the Second Storegga Slide tsunami struck the coastline of eastern Scotland. Comparison of the maximum altitude of the estuarine sediments beneath the tsunami layer with the maximum values for the altitudinal limit of deposits of the tsunami layer at several sites in this area suggest that the tsunami occurred near high tide in an area between Montrose and the Forth estuary, since in those areas the difference between the two values is greatest.

As described above, it seems that the Second Storegga Slide tsunami reached its highest levels along the coastline of central eastern Scotland, and it seems reasonable to suppose that in this area the waves struck during high tide. The numerical model of the Storegga tsunami developed by Henry and Murty (1992) showed that the tsunami took circa 6 hours to reach the Montrose-Forth area of eastern Scotland from western Norway. Therefore, if present tidal conditions are similar to those of circa 7,000 years ago, it would appear that the tsunami caused by the Second Storegga slide struck the coastline of the Shetland Isles during low tide and that approximately 6 hours later it flooded the coastline of eastern Scotland during high tide.

A More Recent Destructive Tsunami?

There is a growing body of evidence that a more recent high-magnitude destructive tsunami has also taken place in the northeast Atlantic. In the Shetland Islands, a widespread layer of fine to medium sand occurs within coastal blanket peat mosses on the Shetland mainland and on Yell (Birnie, 1993). The layer is commonly no greater than 10 cm in thickness, tapering landward. Particle size analysis by Shi (pers. comm.) indicates a single fining upwards sequence, and detailed borings around Sullom Voe oil terminal show that the layer rises inland to a maximum altitude of +9.3 metres (Ordnance Datum) OD. Radiocarbon dates from peat at the upper and lower contacts with the layer at Basta Voe and Garth's Voe, on the eastern shores of Sullom Voe indicate that the layer accumulated at circa 5,700 years ago (Smith et al., 1993) (Table 1). It is likely that relative sea level at the time the layer was deposited was perhaps several metres (perhaps as much as 10m) below Shetland OD. Thus the evidence from Sullom Voe indicates a run up of at least +10 metres for this event and possibly as high as +20m. The possible widespread nature of this layer and its age suggest that it was deposited as a result of a more recent tsunami. If true, this would imply run-up values at least as great if not considerably higher than those attributable to tsunami deposits associated with the Second Storegga Slide. The apparent restricted distribution of the circa 5,700 years tsunami deposit to the coastline of northern Scotland and the astonishingly high coastal runup values (10-20m) can most easily be explained by a slide-generated tsunami that was associated with considerably shorter wavelengths of propagated waves than the extremely long wavelength phenomena associated with the 7,000 years tsunami (cf. Harbitz 1992).

Table 1: Radiocarbon dates of tsunami sediments, Shetland Isles

Numerical modelling of the Tsunami generated by the Second Storegga Slide

Detailed research on this subject has recently been undertaken by Harbitz (1991, 1992) who has developed a mathematical model based on the hydrodynamic shallow water equations for numerical simulation of water waves generated by the Second Storegga Slide. Harbitz has solved the equations using a finite difference technique and has shown that the likely tsunami run-up values are greatly dependent on the average velocity of the landslide as well as the shear stress at the interface between the water and the slide body. He has shown that a landslide moving at an average velocity of 35 m/sec would produced average flood run-up values of between +3 and +5m along the eastern coast of Greenland, Iceland, Scotland and the western coast of Norway. By contrast, a landslide moving with a velocity of 50ms-1 is likely to have produced runup at the coast in excess of +20m. This value for flood runup is considerably higher than the empirically measured values of tsunami runup along the northern coastline of Scotland. Owing to the good correspondence between the measured runup values and the predicted values for a 35ms-1 slide velocity, the latter is considered to be a realistic means of tsunami generation.

Harbitz concluded that there is likely to have been a very marked initial drawdown of water (possibly in excess of -10m) prior to the arrival of the first major tsunami wave (possibly circa 10m high) along the west Norwegian coast. Tsunami runup of up to +10m (equivalent to a water level rise of +20m) follows within the space of two hours (Figure 1). The data of Harbitz suggest that the Second Storegga Slide was associated with two major tsunami waves as well as several minor water level fluctuations. More recently, Henry and Murty (1992) have developed a different numerical model of the Storegga Slide and have derived similar tsunami run-up values to those of Harbitz. These data suggest that there is a broad compatibility between the preliminary results of the landslide-generated tsunami models and the empirical results on tsunami run-up for the Scottish coast as suggested by Dawson et al. (1988). These results should be treated with caution, however, since, the run-up values take no account of the influence of nearshore bathymetry on regional variations in former tsunami run-up. Nor do they include the effect of tidal variations during the period of tsunami propagation. In theory, this effect could add or subtract circa 3m to the tsunami runup above the position of mean sea level on the day that the tsunami took place.

Despite the sophistication of the numerical models there are certain limitations that affect the mathematical calculation of runup. A feature of the tsunami generated in the numerical model of Harbitz (1992) is the high value (circa 200 km) for the tsunami wavelength. This means that, owing to scale differences, the model treats the coastline as almost a vertical wall against which runup takes place. In order to compensate for this inaccuracy, Harbitz defined a series of locations offshore yet close to the coast where the waves were forced to deform over a hypothetically evenly sloping seabed surface (Table 2).

Table 2: Second Storegga Slide (after Harbitz 1992). Tsunami height offshore, nearshore seabed slope and runup,

Thus it was possible to calculate tsunami height offshore (Table 2). The chosen water depths are sufficiently shallow and imply that, owing to the large wavelengths involved, the waves have already been significantly deformed before being forced to move over an inclined seabed surface. The combined effects of these computations means that whereas the calculated values for tsunami height offshore are reasonably accurate, limited reliability can be placed on the computations of runup.

Time series analysis of tsunami height at the three offshore sites are shown in Figure 1 where one can observe that the average duration of the most dramatic water level oscillations is in the order of 6 hours.

A weakness of the Harbitz model of the Second Storegga tsunami is that it does not make any correction for tidal changes that may have occurred during the progress of the tsunami. The inshore transformation of the tsunami hydrodynamics described by Harbitz is as realistic as present technology allows. To date there have been no nearshore simulations of the Storegga tsunami for any specific coastal area. This is a research priority.

Potential Damage to Offshore Oil Structures from Extreme Water Level Changes

There is an extremely serious risk of damage to offshore oil structures due to extreme variation in the level of the ocean surface during the period of time immediately prior to the propagation of the first tsunami wave. It is well-known, e.g. Harbitz (1991) that sea floor disturbance, whether it be from a seabed fault or from an underwater slide, is accompanied by strong vertical water motion. In general, when a fault or an underwater slide takes place, water moves into the area from which seafloor material has been displaced. Thus in the case of the Second Storegga Slide, the initial water motion accompanying the landslide is into the area defined by the landslide scarp slope. In any tsunami these movements of water in the open ocean are accompanied by a draw-down of water at the coast. In general terms, the volume of ocean water moved into and over the area of seabed disturbance is approximately equal to the volume of water displaced seaward in response to lowering of the ocean surface and the draw-down of water at the coast.

Thus, when a tsunami takes place the primary effect is a movement of tremendous volumes of water into the ocean area located above the sea floor disturbance. Owing to the effects of momentum transfer, ocean water tends to pile-up in an ocean area circa several tens of km2 and the level of the ocean surface in this area can increase its elevation dramatically. This accumulation of surplus water over a restricted ocean area is followed by collapse and it is the collapse of this water column that constitutes the primary mechanism by which tsunami waves are propagated out from a point source in the ocean.

It should be noted from this discussion that the precise altitude above the ocean surface to which water may rise is dependent upon the rate and dimensions of seabed displacement. Thus, a landslide that is generated on the seabed that moves relatively slowly will produce smaller volumes of surplus water over the area of sea disturbance as compared with an extremely rapid event. In the case of the Second Storegga Slide we do not know how fast the landslide moved. The only way that crude attempts can be made to reconstruct the former landslide velocity is by comparing the geological observations of runup at the coast with those produced by a mathematical model. Harbitz (1991) tuned the landslide model to correspond with geological estimates of runup at the coastline and was able to estimate a likely average landslide velocity of 35m/s but it was always an impossible task to estimate the speed of initial slope failure in the area now defined by the scarp slope. These two parameters, once defined, produce the dimensions of the tsunami waves, the rate of slope failure the first factor that comes in to play when the tsunami is initially generated, but the landslide velocity also being important since it defines the rate at which the seabed continues to be subject to mass displacement at a time when the tsunami waves begin to be propagated outwards from the area of initial ocean disturbance.

It is virtually impossible to estimate the rise in the water level in the area of the ocean that is subject to the initial accumulation of water. We choose a value of +100m to be included in this report, but this value could be significantly higher or lower dependent upon the processes described above. We therefore identify an area of ocean which is at greatest risk from water level changes and it should be noted that water level changes described for this area are likely to be several times higher than the water level changes associated with propagation of individual tsunami waves. The question therefore in terms of safety is to identify the area of ocean most likely to be subject to such extreme water level changes should a future disaster occur. Although this is extremely difficult to accomplish with any degree of precision, it is logical to conclude that the possible areas of extreme water disturbance are likely to occur above any future scarp slope activated by a slide. Since the greatest accumulations of soft sediments occur on the continental slope west of Norway, we identify this area (as opposed to the continental shelf or abyssal regions) as the area in which oil installations are at greatest risk. Since the upper level of the scarp slope associated with the Storegga Slides occurs at circa 500m water depth, we identify a zone of the Norwegian Sea beyond the continental shelf at a water depth range between circa 500-750m as the area at greatest risk.

It is difficult to imagine what the effects of such extreme water level changes would be at an offshore oil installation. The first effect following the triggering of an underwater slide in the future would be a draw-down of the ocean surface surrounding a central area into which water would move. We envisage that in the first few minutes of such an event the water levels would increase by several tens of metres thus exerting immense strains on the structures as well as causing their temporary submergence and possible collapse.

Potential Damage to Offshore Oil Structures as a Result of Underwater Slumps and Slides

One of the characteristic features of the Norwegian Sea is the widespread occurrence of underwater slump and slide sediments. These were first described by Kenyon (1987) and more recently newly-identified underwater slides have been identified along the continental margin NE of the Faeroe Islands (van Weering et al., 1998). In addition, evidence for long-term instability in the Storegga Slide region on the continental slope of W Norway has recently been summarised by Evans et al. (1996). The continental slope region of the Norwegian Sea and Greenland Sea is therefore well-known as an area characterised by underwater slide deposits and points to the occurrence in this area in the recent geological past of a series of submarine slope failures. The area is therefore potentially susceptible to underwater slope failures and therefore there is a danger that an oil rig whose foundation piles are constructed in landslip sediments are susceptible to movement in the future. Assessment of the geotechnical properties of these sediments is beyond the scope of this report, but is presently an area of experimentation undertaken by engineering geologists at various academic institutes in Norway. Suffice it is to state here that there is a real risk that future sediment failures could take place on the continental slope west of Norway and lead to the collapse of individual oil rigs. By contrast, those oil rigs located on the continental shelf west of Norway would appear to be less susceptible to submarine sediment slumping and sliding.

Triggering Mechanisms

At present, contrasting views have been expressed on the likely mechanisms that would lead to the occurrence of a future submarine slope failure. First there is the obvious hazard from offshore earthquakes. Information on past offshore earthquake frequency is described in a later section. However, it should be simply stated here that there is a possibility that a future offshore earthquake of sufficient magnitude could lead to a future slope failure. The second mechanism is less well understood and involves the release of methane gases (clathrates) as a mechanism by which seabed sediments may be subject collapse and failure. The process of clathrate production in underwater sediments has been widely discussed in the geological literature for many years, and although detailed discussion of these processes is beyond the remit of this report, it should be noted that gas release should be considered seriously as a possible mechanism for seabed sediment failure.