STUDY OF TSUNAMI ATTACKS ON NEIGHBORING COUNTRIES OF CASPIAN SEA CAUSED BY A PROBABLE SUBMARINE LANDSLIDE

Transcription

1 Journal of Coastal Research SI ICS2011 (Proceedings) Poland ISSN STUDY OF TSUNAMI ATTACKS ON NEIGHBORING COUNTRIES OF CASPIAN SEA CAUSED BY A PROBABLE SUBMARINE LANDSLIDE M. Soltanpour and E. Rastgoftar Dept. Civil Eng. Dept. Civil Eng. K. N. Toosi University K. N. Toosi University of of Technology, Tehran Technology, Tehran , Iran ,Iran soltanpour@kntu.ac.ir ehsan_rastgoftar@sina.kntu.ac.ir ABSTRACT Soltanpour, M. and Rastgoftar, E., Study of tsunami attacks on neighboring countries of Caspian Sea caused by a probable submarine landslide. Journal of Coastal Research, SI 64 (Proceedings of the 11th International Coastal Symposium), Szczecin, Poland, ISSN Subduction zone earthquakes have been considered as the tsunami source in most of the previous studies of Caspian Sea tsunamis and the result of a submarine landslide has been less investigated. A probable tsunami generated by a submarine landslide, located in Derbent Basin of the middle part of Caspian Sea, is studied in this research. The effect of tsunami on surrounding countries is simulated using GEOWAVE, a combination of TOPICS and FUNWAVE models. Numerical results show that tsunami waves, propagating out from landslide location in circular rings, are first amplified approaching the coasts of close neighboring countries; but damp rapidly when they travel to far distances. The high tsunami waves are observed only along the coastlines of countries in the vicinity of the landslide and the probability of tsunami attack on other coasts of Caspian Sea countries is relatively low. This can be attributed to the limited far-field effect of submarine landslides tsunamis due to their radial damping and dispersion. Future studies to determine other probable landslide locations in Caspian Sea are important for the risk assessment of generated tsunamis on neighboring countries. ADDITIONAL INDEX WORDS: Tsunami Simulation, risk assessment, GEOWAVE, Derbent Basin INTRODUCTION The public awareness of tsunamis has been intensified following the destructive Indian Ocean tsunami caused widespread damage and more than 225,000 fatalities. Similar to the other coastal regions around the world, the increase of the population along the coasts of Caspian Sea highlights the urgent need to assess tsunami hazards in the region. There have been a number of investigations on Caspian earthquake tsunamis during past years. The collected information achieved by historical events, regional seismicity and numerical models shows that coseismic tsunamis in the Caspian Sea have repeatedly happened in the past and their occurrence in future are probable. Although the past historical tsunamis have not been destructive (Dotsenko et al., 2002), seismic activity is not the only possible cause of tsunami generation in the Caspian Sea. Underwater landslides, explosions of mud volcanoes and other factors can probably produce locally destructive tsunamis. These non-earthquake sources of locally destructive tsunamis have been less studied due to limited existence of reliable information and low recurrence of these events in the region. In the autumn of 2004, using a super high-resolution narrowbeam parametric profiler, the fine structure of the uppermost sediments of the Caspian Sea was studied for the first time. Analyzing the data of profiling across the western slope of Derbent Basin, located in the middle part of Caspian Sea and in the vicinity of the Dagestan coast, indicated that submarine landslide processes proceeded on the region during the Neopleistocene Holocene, which may have been kept their activity up to present (Levchenko et al., 2008). Most submarine slopes are inherently stable. Elevated pore pressures (leading to decreased frictional resistance to sliding) and specific weak layers within stratified sequences appear to be the key factors influencing landslide occurrence. Elevated pore pressures can result from processes such as rapid sedimentation, earthquake shaking or possibly due to melting of gas hydrates contained within the sediments. Historical evidences suggest that the majority of large submarine landslides are triggered by earthquakes (Masson et al., 2006). Seismological analysis and historical events show Derbent Basin has highest seismic activity among the regions of Caspian Sea (Dotsenko et al., 2002). The existence of the transient factor to make a submarine landslide reveals the relatively high probability of occurrence of a submarine landslide in the region. GEOWAVE, a combination of TOPICS and FUNWAVE models, is an integrated tsunami simulation numerical model. Using GEOWAVE model, the probable tsunami caused by a submarine landslide in Derbent Basin is simulated in this study to assess tsunami hazards on the coasts of Caspian Sea. Tsunami generation is first simulated by TOPICS model. The propagation of tsunami will then be investigated employing FUNWAVE. SUBMARINE LANDSLIDES Submarine landslides, or submarine mass failures, are one of the main agents through which sediments derived from land (mainly 1195

2 Coastal Modelling carried by rivers) and from the continental shelf (e.g. through erosion and transport by ocean currents and storms), are transferred across the continental slope to the deep ocean. Although subduction zone earthquakes are the commonest source of tsunamis around the world, submarine mass failures have also resulted to considerable tsunamis. While earthquake tsunamis, studied for the last 50 years, are now relatively well understood, knowledge of the generation and propagation of submarine mass failure tsunamis is instead still fragmentary. The importance of tsunamis generated by submarine mass failure was only recognized following the 1998 Papua New Guinea terrible tsunami, where waves up to 15 m high affected a 20 km segment of coast and killed 2200 people (McSaveney et al., 2000). Tsunamis generated by subduction zone earthquakes or submarine mass failures have fundamental differences. Rapture dimensions determine the source areas for earthquake tsunamis resulting to vast source areas, compared to the areas affected by landslide tsunamis. On the other hand, tsunamis generated by subduction zone earthquakes have a linear source and propagate perpendicular to the source fault but landslide tsunamis propagate radial due to their point source. The small source area of landslide tsunamis also leads to the generation of shorter waves in comparison to the waves caused by earthquakes tsunamis. The dispersion of short waves and also radial spreading decrease the far-field effects of landslide tsunamis in contrast to tsunamis of seismic origins. However, shorter waves are more prone to coastal amplification with higher local effects. Unlike tsunamis generated by earthquakes, submarine landslide tsunamis generated in shallow waters are more destructive compared to those generated in deep water. This is due to the higher energy that can be converted from the slide to the water in shallow areas. Moreover, shallower waters are usually closer to the coasts and thus a shorter available distance exists for radial damping. The time of the initial wave generations are also different in these two types of tsunamis. Earthquake tsunamis are generated instantaneously, so the final seabed vertical displacements are immediately transferred to initial sea surface elevations. However, since the movements of landslides are normally sub critical, a landslide tsunami leaves the generation region more rapidly than the duration of landslide motion. Thus, the timing of the landslide movement becomes important for the generation of submarine landslide waves. TSUNAMI GENERATION TOPICS, tsunami generator model of GEOWAVE, can simulate multiple tsunami sources with different generation mechanisms. For submarine landslides, the initial free surface elevation and water velocities in TOPICS are derived from multivariate, semiempirical curve fits as a function of non-dimensional parameters characterizing the landslide (e.g., density, geometry, etc.) and the local bathymetry (e.g., slope, depth, etc.). Relevant nondimensional parameters are selected based on the numerical experiments, first carried out with 2D fully nonlinear potential flow model of Grilli and Watts (1999). The curve fits were later modified based on the results of the more recent 3D model of Grilli et al. (2002). Two idealized types of submarine mass failures moving over plane slopes are considered in these models, representing the extreme cases of a general probable submarine mass failure motion. These two types are underwater slides, i.e. translational failures, and slumps, i.e. rotational failures. For underwater slides, which probable submarine landslide in Derbent Basin of Caspian Sea appears to be similar to it, the landslide is idealized as a mound with elliptical cross-section translating along a straight slope θ (Figure 1). The mound is specified with maximum Figure 1. Definition sketch of the simulation domain for underwater slides (Watts et al., 2003). thickness T in the middle, total length b along the down-slope axis, total width w along the cross-slope axis, and an initial submergence d at the middle of the landslide. Expressing the Newton s first law by the balance of existent forces for the center of mass motion and performing 32 underwater slide numerical simulations by Grilli et al. (2005), covering a wide range of governing parameter values, led to construct predictive equations for 2D tsunami amplitude, minimum surface depression above the middle of the initial slide position, and characteristic wavelength based on curve fitting results. Since more than half of all tsunamigenic submarine landslides do not satisfy the 2D criteria established in 2D model, 3D simulation were performed by Grilli et al. (2002) to propose an analytical method to specify initial 3D tsunami elevations. It was concluded that underwater slide tsunami features are primarily a function of submarine mass failure volume (b,w,t), angle of slope, and initial submergence in these models. Table 1 shows the required TOPICS parameters to simulate tsunami generation based on landslide characteristic and local bathymetry. Data reported from typical widths of the past landslides around the world (e.g., McAdoo et al., 2000; Hutton and Syvitski, 2004) show that underwater slides are often narrow compared to their length, with typical width w =0.25b (Grilli et al., 2005). Unfortunately, there is not an accurate estimation for the total width of the landslide in the region. This parameter has been considered as a variable in this study assuming 1,000, 2,000 and 3,000 meters for the total width of the landslide. The initial wave s amplitude is linearly proportional to landslide width, as illustrated in Figure 2. Despite the differences of initial free surface elevations, the calculated wavelengths, corresponding to the Table 1 landslide, are about 30 kilometers for all three cases. This is due to the fact that the wavelength, determined by the travel time which itself is a function of initial submergence and landslide length, is independent of landslide width. The assumed widths of the landslide and other parameters were introduced to TOPICS model. Consequently, TOPICS provided the initial free surface of the landslide tsunami at characteristic time (t 0 = 484s), after that the landslide starts its motion (Figure 3). Table 1: Parameters of the Caspian Sea probable submarine landslide. Estimated parameter Value Slope (θ) 2 Initial submergence 415 m Length 5300 m Maximum thickness 1100 m 1196

3 Soltanpour and Rastgoftar Figure 2. Variation of initial free surface elevation with width of the landslide (w), corresponding to the Table 1 landslide. TSUNAMI PROPAGATION AND INUNDATION The outputs calculated from TOPICS model are introduced as the initial conditions to the tsunami propagation model. In the case of a landslide tsunami, the generated outputs are free surface elevation and water horizontal velocities, while no initial horizontal velocities are assumed for the tsunamis resulted by earthquakes. Figure 3. Initial free surface elevation corresponding to the Table 1 landslide with an assumed landslide width of 2,000 m. Earthquake tsunamis are most commonly described by the shallow water equations, i.e. the simplest type of depth integrated long wave equations, since their wavelengths, in the order of hundreds of kilometers, are much larger than the ocean depth, in the order of few kilometers. Assuming the vertical acceleration of water particles to be negligible compared to the gravitational acceleration, the hydrostatic pressure approximation is used in long wave theory. Moreover, the shallow water equations disregard frequency dispersion. In spite of these simplifications, these equations are generally enough accurate for the modeling of earthquake tsunamis. However, landslides produce shorter tsunami waves in comparison to those generated by earthquake tsunamis, as mentioned before. These tsunami waves are nonhydrostatic and the vertical velocities cannot be neglected, in contrast to the waves generated by earthquake tsunamis. Moreover, the shorter wavelength indicates the necessity of the dispersive model application. A different approach is necessary for the numerical modeling of landslide tsunamis. GEOWAVE simulates tsunami propagation and inundation using the long wave propagation model FUNWAVE based on fully nonlinear Boussinesq equations developed by Wei et al. (1995), considering the effects of nonlinearity and frequency dispersion. Another advantage of choosing a Boussinesq wave propagation model is that the horizontal velocities are no longer constrained to have a constant value over the water depth (Watts et al., 2003). Since Boussinesq equations become invalid in the surf zone, because of not including the wave breaking, FUNWAVE applies a simple eddy viscosity-type formulation to model the turbulent mixing and dissipation caused by wave breaking. Some additional eddy viscosity terms are introduced to momentum conservation equations. Onset and cessation of breaking in each point of model domain is determined by η t, variation of free surface with respect to time, which is calculated from mass conservation equation. For simulation of wave run-up, the model uses the slot method of Tao (1983, 1984). This technique assumes the beach is porous, or it contains narrow slots. The porous beach allows the water level to be below the beach elevation and propagate within the land. Run-up occurs when the elevation of the groundwater rises above that of the land. Slot method calculates the maximum run-up height with about a 10% diminution error. This error arises because first the slot should be filled before water could cover the dry land. A slightly different formulation in slot method, proposed by Kennedy et al. (2000), is applied by FUNWAVE to reduce water mass losses. However, using slot method leads to a small alteration in mass conservation. Using finite difference technique and a composite 4 th -order Adams-Bashforth-Moultan scheme, the governing equations are solved in the modeling domain area (UTM coordinates). Surface elevation and horizontal velocities are calculated at 427,200 ( ) grid points for all time steps of simulation. Considering the size of the mesh, i.e decimal degree, the time step of dt=3.65s was determined by the model to satisfy the stability conditions. Figure 4 displays the propagating tsunami waves computed by FUNWAVE model. It is observed that tsunami waves propagate out from the landslide location in circular rings. The waves amplify approaching the coast of landslide neighbouring countries but they highly damp propagating more distances to reach far countries. In order to have a better view of waveforms at different locations, point stations were defined along the Caspian Sea coastline. Figure 5 shows the location of assumed stations to get the time series of tsunami, where the calculated wave heights are presented in Figure

4 Coastal Modelling Figure 4. Computed tsunami waves propagating at (a) 0.5, (b) 1, (c) 2, and (d) 3 hours after the landslide motion. Figure 5. Location of numerical wave stations (A-K). 1198

5 Soltanpour and Rastgoftar part of Dagestan, north of Azerbaijan, and near the regions of Aqtau in Kazakhstan, other coasts of Caspian Sea are safe against the mentioned landslide tsunami. However, for an integrated assessment of Caspian Sea tsunami hazards, the possibility of the occurrence of submarine landslides in other parts of Caspian Sea should be investigated. Figure 6. Wave height time series for numerical wave stations; w (width of the landslide) = 1,000 meter (...), 2,000 meter (---), and 3,000 meter ( ). It is observed that the increase of the landslide width results to higher wave heights at the coastlines, as expected. Moreover, the stations closer to the landslide receive relatively higher waves. Figure 6 reveals that the generated tsunami can cause a considerable run-up along the coastlines of adjacent countries such as southern part of Dagestan, north of Azerbaijan, and Aqtau of Kazakhstan. Northern parts of Azerbaijan are first hit by tsunami waves, just after about 35 minutes. Considering that the landslide movement is towards east direction and the trough of the tsunami wave is created behind the landslide, falling water is first observed along the coastlines locating west of the tsunami source, i.e. stations D, F and J. This can be trusted as a useful natural warning sign to the local communities. However, the coastlines locating east of tsunami source first experience the rising tsunami waves. It should also be mentioned that station G, located in west of Caspian Sea, is not behind the landslide and it will first experience a high tsunami wave. It can also be observed that the tsunami waves along the coastlines of the countries far from the landslide location are small even for the case of w=3,000 m. Therefore, the danger of this tsunami at the coastlines of Iran, southern part of Turkmenistan and Azerbaijan, and North of Kazakhstan is not high and these coasts will not experience a remarkable inundation. CONCLUSIONS A probable tsunami generated by submarine landslide, located at the middle part of Caspian Sea in the vicinity of the Dagestan coast, was simulated to investigate tsunami hazards on the neighbouring countries. GEOWAVE numerical model was employed to simulate tsunami generation and propagation. The required model inputs were estimated based on landslide characteristic and local bathymetry. Since there is not an accurate estimation for the total width of the landslide, this parameter was considered as a variable based on the typical widths of the past landslides. Model results revealed that this probable landslide tsunami is capable of generating high waves and considerable run-ups along the coasts of countries in the vicinity the landslide. However, the danger of a major tsunami attack on the other coasts due to this landslide, e.g. Iran, is very low and there will not be a major inundation. This can be attributed to the limited far-field effect of the submarine landslides tsunamis, because of their radial damping and frequency dispersion. In summary, except southern LITERATURE CITED Dotsenko, S.F.; Kuzin, I.P.; Levin, B.V., and Solovieva, O.N., Tsunamis in the Caspian Sea: Historical events, regional seismicity and numerical modeling. Local Tsunami Warning and Mitigation,: Proceedings of the International Workshop, Grilli, S.T., and Watts, P., Modeling of waves generated by a moving submerged body: Applications to underwater landslides. Engineering Analysis with Boundary Elements, 23(8), Grilli, S.T.; Vogelmann, S., and Watts, P., Development of a 3D numerical wave tank for modeling tsunami generation by underwater landslides. Engineering Analysis with Boundary Elements, 26(4), Grilli, S.T.; Watts, P.; Tappin, D.R., and Freyer, G.J., Tsunami generation by submarine mass failure. II: Predictive equations and case studies. Journal of Waterway, Port, Coastal, and Ocean Engineering, 131, Hutton, E.W., and Syvitski, J.P., Advances in the numerical modeling of sediment failure during the development of a continental margin. Marin Geology Journal, 203(3), Kennedy, A.B.; Chen, Q.; Kirby, J.T., and Dalrymple, R.A., Boussinesq modeling of wave transformation, breaking, and run-up. I:1D. Journal of Waterway, Port, Coastal, and Ocean Engineering, 126(1), Levchenko, O.V.; Verzhbitskii, V.E, and Lobkovskii, T., Submarine landslide structures in Neopleistocene deposits on the western slope of the Derbent Basin of the Caspian Sea. Oceanology Journal, 48(6), Masson, D.G.; Harbitz, C.B.; Wynn, R.B.; Goldsmith, P.; Pedersen, G., and Lovholt, F., Submarine landslides: processes, triggers, and hazard prediction. Philosophical Transactions of the Royal Society A. Journal, 364, McAdoo, B.G.; Pratson, L.F., and Orange, D.L., Submarine landslide geomorphology, US continental slope. Marin Geology Journal, 169, McSaveney, M.J.; Goff, J.R.; Darby, D.J.; Goldsmith, P.; Barnett, A.; Elliot, S., and Nongkas, M., The 17 July 1998 tsunami, Papua New Guinea: evidence and initial interpretation. Marin Geology Journal, 170, Tao, J., Computation of wave run-up and wave breaking. Internal Report, Danish Hydraulics Institute, 40 pp. Tao, J., Numerical modeling of wave run-up and breaking on the beach. Acta Oceanologica Sinica, 6(5), Watts, P.; Grilli, S.T.; Kirby, J.T.; Freyer, G.F., and Tappin, D.R., Landslide tsunami case studies using a Boussinesq model and a fully nonlinear tsunami generation model. Hazards and Earth System Sciences Journal, 3(6), Wei, G.; Kirby, J.T.; Grilli, S.T., and Subramanya, R, Fully nonlinear Boussinesq model for free surface waves. Part 1: Highly nonlinear unsteady waves. Journal of Fluid Mechanics, 294,