TSUNAMI SOURCE PARAMETERS OF SUBMARINE EARTHQUAKES AND SLIDES M. D. TRIFUNAC and M. I. TODOROVSKA University of Southern California, Civil Engineering Department, KAP 216D, Los Angeles, CA 90089-2531, USA Abstract The nature of the tsunami sources (displacement, duration, area and volume) is reviewed for selected past earthquakes, slumps and slides, from the point of view of tsunami generation. It is concluded that slides and slumps differ significantly form earthquakes in: (1) the velocity of spreading, (2) the balance of the uplifted material, and (3) the vertical displacements per unit horizontal area. Keywords: Submarine slides, slumps, earthquakes, tsunami, near-field 1. Introduction Tsunami are dispersive gravity water waves, propagating at long periods with phase velocity cT = gh , where h is the ocean depth and g is the acceleration due to gravity (Kajiura, 1963). The velocities of earthquake faulting are about 3 km/s, an order of magnitude larger than cT. Therefore, most classical studies of tsunami generation by earthquakes have neglected the details of wave propagation in the fluid during the source time, and have assumed that the initial conditions for tsunami generation can be approximated by specifying the initial disturbance of the ocean surface to be the same as the vertical offset of the ocean floor (Todorovska and Trifunac, 2001). For most tsunamigenic earthquakes, the average runup amplitudes (corrected for geometric spreading) correlate well with the surface wave magnitude Ms (Pelayo and Wiens, 1992), but for some may be 1 to 2 orders of magnitude larger than the average trend. These have been named “tsunami earthquakes”, and it has been suggested that they are caused by a long faulting process, so that Ms underestimates the true value of the seismic moment (Kanamori, 1972). Very long source times may result from multiplicity of the earthquake source, slow dislocation velocities in the accretionary wedge (Pelayo and Wiens, 1992; Tanioka et al., 1997), submarine landslides triggered by an earthquake, slumps (Hawaiian chain, Hading Bay and Riangkroko, and Storegga), repeating submarine volcanic explosions (1883 Krakatau: Yokoyama, 1987) and changes in the atmospheric pressure. Submarine slides and slumps that cause tsunami may be triggered by earthquakes, volcanic eruptions, storm waves, or by gravitational loading and instability (Hampton et al., 1996). This paper reviews the source durations, displacement amplitudes, areas and volumes of selected past earthquakes, slumps and slides that have or may have generated a tsunami. The aim is to describe and quantify the differences between submarine slides and slumps on one side and tsunamigenic earthquakes on the other, from the viewpoint of tsunami generation. 121 122 Trifunac and Todorovska 2. The Nature of Tsunami Sources 2.1 VELOCITIES OF SPREADING The seismological studies of tsunamigenic earthquakes (Kanamori, 1972; Comer, 1982) capture the long periods of the wave motions, and allow the earthquake source to be represented by a point. In contrast, the high frequency recordings of strong motion in the near-field (0.1 to 25 Hz) can resolve fine time and space variations in the source time function, and can be used in inverse studies to interpret the motions on the fault surface (Trifunac, 1974). At high frequencies, a long lasting rupture of a large earthquake often appears as a sequence of multiple events. The Nicaraguan tsunami earthquake of September 2, 1992, is an example of a multiple event (Ide et al., 1993). The generation of tsunami by vertical displacement of the ocean floor depends on the characteristic size (length L, and width W) of the displaced area and on the time, τ, it takes to spread the motion over the entire source region. The ratio L/τ defines the average spreading velocity, c (assuming unilateral faulting). During “ordinary” earthquakes ( 1 < c < 10 km/s), the slip propagates with velocities approaching the shear-wave velocity in the medium, and reductions in the overall average value of c can result from multiplicity of the source and delays associated with breaking of barriers. The “slow” earthquakes ( 0.1 < c < 1 km/s) may consist of one or several high velocity rupture events (thus producing the usual train of high frequency waves), with the long delays between the successive events accompanied by a slip that can contribute large amplitude low-frequency excitation. Examples of such “slow” earthquakes are the June 6, 1960, Chile earthquake which ruptured as a series of earthquakes for about an hour (Kanamori and Stewart, 1979), and the February 21, 1978, Banda Sea earthquake (Silver and Jordan, 1983). Figure 1 shows seismic moment (M0), magnitude (Richter, ML, surface wave, Ms, and moment, Mw, magnitudes), and a moving source “volume”, all used as measures of the size of the source versus different measures of the source duration. Here “volume” refers to $ for earthquakes, where is the average dislocation amplitude and A is the fault area, or to (inferred) slide volume. The Grand Banks, 1929, and the Papua New Guinea, 1998, slides are shown plotted versus two different measures of “size”: (1) magnitude of the “earthquake” that triggered the slide or amplitudes of the seismograms recorded during the sliding process, and (2) estimates of their volumes, respectively 185−550 km3 (Driscoll et al., 2000; Hasegawa and Kanamori, 1987) and 4 km3 (Heinrich et. al., 2000). For these two events, the “size” measured in terms of the volume of the displaced material is two to three orders of magnitude “larger” than the size measured in terms of $(via M0 and M). The left three shaded zones represent approximate boundaries between “silent” (0.01 < c < 0.1km/s), “slow”, and “ordinary” earthquakes, assuming M ≈3.94 + 1.94L for c = 0.1, 1 and 10 km/s (Trifunac, 1993). The solid lines are for unilateral faulting, while the top (left) ends of the gray zones are for symmetric bilateral faulting. The velocities of underwater landslides and gravity currents are one to two orders of magnitude smaller than the velocities associated with earthquake dislocations. Based on a succession of broken underwater cables, Bjerrum (1971) estimated the mean gravity current speeds to be 3−7 m/s. From data on broken underwater cables in the Var river delta, Hamilton and Wigen (1987) computed gravity current velocities along a 0.5° slope of 6 m/s. From the timing of breaks in underwater cables during 1929 Grand Banks slide, Kuenen (1952) estimated the gravity current Tsunami source parameters of submarine earthquakes and slides 123 velocities to be 5−30 m/s. The Grand Banks underwater slide had maximum velocities of 20−30 m/s, along slopes of 2°. The maximum velocity of the first Storegga slide was estimated to be 35 m/s (Harbitz, 1992). Jiang and LeBlond (1992) simulated underwater landslide velocities of 20−45 m/s on inclines 2−12°. The velocities of turbidity currents are 4−8 m/s (Shepard, 1963; Bowen et al, 1984; Reynolds, 1987). τC - Beroza & Jordan (1990) T-wave duration, Okal & Talandier (1986) Pelayo & Wiens (1992) 26 /s km 0 .1 ea r th q Slo w Hasegawa & Kanamori (1987) 1960 Chile 1960 Peru Kanamori & Stewart (1979) 4 km 3 1978 Banda Sea 10 0 s 1 c - 3 0 = 10 m/ km s /s 10 1 Volume - km 3 7 c= m/ c= es u ak 1992 Nicaragua 1994 Java 1896 Sanriku 1946 Aleutian 1929 Grand Banks Slide 185 km 3 10 2 MS log M 0 - dyne cm hq km 550 km 3 8 28 Ea rt 10 c= MW 10 3 1k 10 4 /s 9 uak s es 30 itie lo c en u rr avi ty c 10 -2 Papua New Guinea Heinrich et al. (2000) Gr 10 -3 22 0.1 t ve t ea Sile n 5 1 10 100 1m uak rth q ML 24 c= es 1979 Eltamin 10 -1 /s 1979 South of Panama 6 Turbidity currents 1-8 m/s Shepard (1963) Bowen et al. (1984) Reynolds (1987) 1000 10000 Duration - s Figure 1. Earthquake moment M0, magnitudes (ML, M, and Mw) and source volume, versus different measures of source duration: (1) duration of T-waves (solid small circles); (2) characteristic source duration, τc (open small circles); (3) duration of three tsunami and one tsunamigenic earthquakes (crosses); (4) approximate durations of Sanriku, 1896, Aleutian, 1946, Nicaragua, 1992, and Java, 1994, events (solid large circles); (5) the overall duration of two complex events: 1929 Grand Banks and 1960 Chile (horizontal bars). The shaded zones represent approximate boundaries between silent, slow and ordinary earthquakes. The large solid and dashed circles show the estimated duration of the Papua New Guinea slide. The grey zone at the bottom represents gravity currents, and the grey cross-hatched zone turbidity currents. 124 Trifunac and Todorovska In Figure 1, the velocities of underwater slides and of turbidity currents are compared with those of “tsunami” earthquakes and “slow” earthquakes. 4 2 10 -1 Seward - 1964 10 0 10 -1 10 0 10 1 10 2 Area - km 2 10 3 10 10 4 0 Landslide area 10 20 Percent 10 2 10 1 N. California P.N.G. - 1998 10 3 Munson Nygren 10 4 Hawaiian slumps 10 G. Banks 1929 10 1 Hawaiian Debris avalanches 5 1 6 Magnitude 8 10000 5 1000 100 Water depth, h - m 1929 Grand Banks Slide Hasegawa & Kanamori (1987) 10 100 Area - km 2 cT cT 0.5 Silent Slow 1000 10000 Tsunami, faulting and slide velocities - m/s Ea rth qu ak ef a u lt a rea Tanioka & Satake (1996) Satake (1994) 1896 Sanriku 1992 Nicaragua cT 100 50 c T Johnson & Satake (1997) 1946 Unimak Island 20 c T Heinrich et al. (2000) 10 c T 1/2 5 cT 1998 Papua ) (gh New Guinea cT= 2 cT How the vertical movements at the source displace the surface of the ocean will depend on the ratio c/cT. For c/cT > 20 to 50, the water amplitudes above the source will approximately be equal to the vertical amplitudes of the permanent displacements at the bottom. For c~cT, wave focusing may occur above the spreading edge of the uplift. For c < cT , the amplitudes of water waves above the source will be “small”, and at the instant the source motion has ceased may cover much larger area than the area of the source. Figure 2 (left) illustrates five examples, three for c > cT and two for c < cT . The range of applicable velocities c is plotted versus the depth of water above the source area. Figure 2. (left) Comparison of long period tsunami velocities (diagonal lines) with velocity zones of "silent" and "slow" earthquakes, versus depth of the water layer in meters. Also shown are the estimated velocity and depth ranges for three tsunamigenic earthquakes (1946 Unimak Island: 1992 Nicaragua: and 1896 Sanriku) and two submarine slides (1929 Grand Banks and 1998 Papua New Guinea). (right) Comparison of: (1) the distribution of landslides by their area (bottom), (2) the areas of major debris avalanches and slumps of the Hawaiian Islands (shown by short solid lines), (3) the areas of selected submarine slides, and (4) the source areas of earthquakes versus their magnitude (top). 2.2 SOURCE AREA For earthquake source areas larger than about 100 km2 (earthquake magnitudes greater than 6 to 6.5; see Fig. 2) the wave energy begins to be transmitted through the water with less dispersion (Comer, 1982). Most submarine landslides (∼90 percent) are smaller than 100 km2 (Fig. 2), and so most undersea slope failures occur without being noticed. However, submarine slides that are close to the coastline, and such that a “large” volume of material is Tsunami source parameters of submarine earthquakes and slides 125 moving over a “small” area, can cause highly localized and devastating tsunami disasters (e.g. Valdez and Seward, on 27 March 1964; Hampton et al., 1993). The largest known slope failures have taken place off the coast of the Hawaiian Iislands (Figs 2 and 3). Some were longer than 200 km, had volumes of 5 × 103 km3 (Figs 2 and 3), and may have generated cataclysmic tsunami amplitudes, assuming that all the debris moved at once. In 1929, a magnitude 7.2 earthquake triggered a submarine landslide at Grand Banks. Part of the failed material transformed into a turbidity current, which traveled more than 700 km, breaking communication cables in its way (Trifunac et al., 2001b; 2002b). 10 4 Nuuanu, Hawaii 10 3 East Breaks (west) Munson-Nygren 10 2 East Breaks (east) P.New Guinea 1998 Alika-1 and Alika-2 Valdez 27 Mar. 1964 N California 8 Nov. 1980 Volume - km 3 Puerto Rico - Insular slope Grand Banks 1929 10 1 10 Papua New Guinea 1998 Heinrich et al. (2000) 0 10 -1 10 -2 Seward 27 Mar. 1964 Grand Banks 1929 Hasegawa and Kanamori (1987) Volume of typical slide 10 -4 4 Volume of ocean floor uplifted by earthquakes Hatori (1966) uA = M 0 / µ 10 -3 5 6 7 8 Magnitude Figure 3. Comparison of selected submarine slides by their volume (in km3), with u A (= average dislocation × fault area), and the volume of uplifted ocean floor by earthquakes, plotted versus earthquake magnitude. 2.3 SOURCE VOLUME The largest known volume of a submarine mass movement is about 5 × 103 km3 and is associated with the Nuuanu debris avalanche, northwest of the Hawaiian island Oahu. Its estimated age is 1.4 to 2.6 million years. The Grand Banks slope failure of 1929 may have displaced 5.5 × 102 km3. The smallest volume illustrated in Fig. 3 is that of the Seward landslide of 1964, with only 5.4 × 10-4 km3 to 5.4 × 10-3 km3 (Hampton et al., 1993). The “volume” of crustal displacement ( uA) associated with earthquakes, can be related to the seismic moment M0 and crustal rigidity µ as uA = M 0 / µ , where u is the average dislocation amplitude and A is the fault area. Because of different geometries of earthquake sources and source depths, the “effective” volume of the displaced ocean floor is smaller. It can be estimated from data on vertical movement in the source regions of well documented tsunami events (Hatori, 1966), and from an estimate of the associated source areas. In Fig. 3, the range of observed values versus magnitude, for 6.5 < M < 8.5, is shown by a gray band (parallel to uA , but smaller, by one to two orders of magnitude). This figure shows that the largest submarine landslides are capable of displacing volume of sediments at ocean floor two orders of magnitude larger than one displaced by large (M = 8.5) earthquakes. It 126 Trifunac and Todorovska also shows that small landslides with volumes of 10-3 to 10-2 km3 can produce powerful, though localized tsunami. 2.4 RELATIVE VERTICAL DISPLACEMENT The spectral amplitudes of tsunami at long periods are determined by the volume of the uplifted sea floor above the source region. The near-field tsunami amplitudes, however, depend on the average amplitude of the uplift and on its variations in time and space in the source area (Trifunac et al., 2001a,b; 2002a). Figure 4 (left) shows a comparison of log10 ( u/A ) for eleven submarine mass movements ( u ) is the average vertical displacement of the ocean floor) and for earthquakes that produced tsunami. It is seen that u / A - km -1 100 10 -1 10 -2 10 -3 10 -4 10 -5 H L 10 -6 es Earthquak 10 -7 6 7 Magnitude 8 0 Scheidegger (1973) average for subaerial landslides Edgers and Karlsrud (1982): upper bound -1 log 10 (H/L) 101 Alika-1 0.1-0.3 MY Alika-2 0.1 MY Grand Banks 1929 102 Papua New Guinea 1998 Seward, Alaska, 1964 (all) Munson-Nygren (segment) Puerto Rico - Insular slope (E) (W) East Breaks (mass floor) N California 8 Nov. 1980 Sur submarine slide, 1500 yrs ago Nuuanu, 1.4-2.6 MY 103 -2 -3 -6 Tristan de Cunha Seward 25.0 4.8 quic k cla y sl ide( ssub aer ial 6.0 Alika - 2 2.0 1.5 Nuuanu 0.5 1.5 ) 2.5 volcanos 3.5 non-volcanos 8.6 slope steepness (degrees) 2.0 -4 -2 0 log 10 Volume - km 3 2 8.6 4 Figure 4. (left) Comparison of u/A (average vertical displacement of the ocean floor / area of the uplifted region) for selected submarine slides (points and vertical bars, plotted versus arbitrary magnitude coordinate), and earthquakes (grey zone). (right, top) Runout model of submarine landslides. (right, bottom) H/L versus volume of failed mass, for a compilation of known submarine landslides (after Hampton et al., 1996). Also shown are the upper bound values from Edgers and Karlsrud (1982), the lower bounds for quick clay slides (subaerial), and the average trend for subaerial landslides proposed by Scheidegger (1973). most of the time this ratio is one to five orders of magnitude larger for submarine landslides than for earthquakes. For most, particularly for the largest and older slides, it cannot be assumed that all material slid during one event. It is possible that what is seen today is the end result of repeated mass movements. On the other hand, both the Papua New Guinea and Seward slides not only have the largest u/A (for this sample) but also represent single events. An exception (in Fig. 4, left) is u/A for the Grand Banks, 1929, event. 2.5 RELATIVE SOURCE LENGTH When the velocities of moving landslides become comparable to the long period tsunami Tsunami source parameters of submarine earthquakes and slides 127 velocity gh , the parameter that governs the amplification of near-field water waves by focusing is the ratio of slide length over water depth, L/h. Trifunac et al. (2001a,b) show L/h for thirteen submarine slides and debris avalanches. For 9 of their 13 examples, this type of amplification is possible. 2.6 RUNOUT Figure 4 (right-top) shows a schematic cross-section of a submarine slide, which travelled downslope, stopping at distance L and level H below its origin. Figure 4 (right-bottom) summarizes known ratios of H/L versus the volume of failed mass (Hampton et al., 1996). When known, the slope steepness in degrees is also indicated. Figure 4 further shows the average value of H/L for subaerial landslides from Scheidegger (1973), and an upper bound from Edgers and Karlsrud (1982). All the data fall above the trend for quick clay slides (Locat and Demers, 1988). For the Hawaiian submarine slides, the runout data are close to the trend for subaerial landslides (Hampton et al., 1996) and have H/L ratio larger than those for the other submarine slides. 3. Conclusions We presented a summary of the physical characteristics of tsunami sources. The main differences between earthquake and tsunami sources are that (1) the earthquake rupture usually spreads with larger velocities (2-3 km/s versus 0.001 to 0.1 km/s), and (2) the balance of the total uplifted material for slides is usually small while for earthquakes it can be considerable (0.01 of M0/µ). This affects the long period limit of the spectra of tsunami amplitudes, proportional to the total uplifted volume of the ocean floor. 4. References Beroza, G.C. and T.H. Jordan, 1990. Searching for slow and silent earthquakes using free oscillations, J. Geophys. Res., 95(B3): 2485-2510. Bjerrum, L., 1971. Subaqueous slope failures in Norwegian fjords. Norwegian Geotechnical Institute Bulletin, 88: 1-8. Bowen, A.J., Normark, W.R., and Piper, D.J.W., 1984. Modeling of turbidity currents on navy submarine fan, California: continental borderland, Sedimentology, 31:169-185. Comer, R.P., 1982. Tsunami generation by earthquakes, Ph.D. Thesis, Earth and Planetary Sciences, Mass. Inst. of Tech., Cambridge, Mass. Driscoll, N.W., Weissel, J.K. and Goff, J.A., 2000. Potential for large-scale submarine slope failure and tsunami generation along the U.S. mid-Atlantic coast, Geology, 28(5): 407-410. Edgers, L. and Karlsrud, K., 1982. Soil flows generated by submarine slides – case studies and consequences, in Proc. Third International Conference on the Behaviour of Off-Shore Structures, edited by C. Chryssostomidis and J.J. Connor, Hemisphere, Bristol, Pa. 425-437. Hamilton, T.S., and Wigen, S.O., 1987. The Foreslope Hills of the Fraser Delta: implications for tsunamis in Georgia Strait, Science of Tsunami Hazards, 5, 15-33. Hampton, M.A., Lemke, R.W. and Coulter, H.W., 1993. Submarine landslides that had a significant impact on man and his activities: Seward and Valdez, Alaska, in “Submarine Landslides: Selected Studies in the U.S. Exclusive Economic Zone, U.S. Geological Survey Bulletin 2002, U.S. Dept. of the Interior, Denver, Co., 80225, 123-134. Hampton, M.A., Lee, H.J. and Locat, J., 1996. Submarine landslides, Reviews of Gephysics, 34, 33-59. Harbitz, C.B., 1992. Model simulations of tsunamis generated by the Storegga Slides, Marine Geology, 105: 1-21. 128 Trifunac and Todorovska Hasegawa, H.S. and H. Kanamori, 1987. Source mechanism of the magnitude 7.2 Grand BanksEearthquake of November 1929: double couple or submarine landslide? Bull. Seism. Soc. Amer., 77: 1984-2004. Hatori, T., 1966. Vertical displacement in a tsunami source area and the topography of the sea bottom, Bull. Earthq. Res. Inst., Tokyo Univ., 44: 1449-1464. Heinrich, P., Piatanesi, A., Okal, E., and Hebert, H., 2000. Near-field modeling of the July 17, 1998 tsunami in Papua New Guinea, Geophysical Research Letters, 27(19): 3037-3040. Ide, S., F. Imamura, Y. Yoshida and K. Abe, 1993. Source characteristics of the Nicaraguan Tsunami earthquake of September 2, 1992, Geoph. Res. Letter, 20(9): 863-865. Jiang, L and LeBlond, P.H., 1992. The coupling of a submarine slide and the surface waves which it generates, J. Geophys. Res., 97(C8): 12731-12744. Johnson, J.N. and K. Satake, 1997. Estimation of seismic moment and slip distribution of the April 1, 1996, Aleutian Tsunami Earthquake, J. Geophys. Res., 102(B6): 11,765-11,774. Kajiura, K., 1963. The leading wave of a tsunami, Earthquake Res. Inst., Tokyo Univ., 41: 535-571. Kanamori, H., 1972. Mechanism of tsunami earthquakes, Phys. Earth. Planet. Inter., 6: 346-359. Kanamori, H. and G.S. Stewart, 1979. A slow earthquake, phys. Earth Planet. Inter., 18: 167-175. Kuenen, P.H. (1952). Estimated size of the Grand Banks turbidity current, American J. of Science, 250: 874-884. Locat, J. and Demers, D., 1988. Viscosity, yield stress, remoulded strength and liquidity index relationships for sensitive clays, Can. Geotech. J., 25, 799-806. Okal, E.A. and J. Talandier, 1986. T-wave duration, magnitudes and seismic moment of an earthquake – application to tsunami warning, J. Phys. Earth, 34: 19-42. Pelayo, A.M. and D.A. Wiens, 1992. Tsunami earthquakes: Slow thrust -faulting events in the accretionary Wedge, J. Geoph. Res., 97(B11): 15,321-15,337. Reynolds, S., 1987. A Recent Turbidity Current Event, Hueneme Fan, California: Reconstruction of Flow properties, Sedimentology, 34, 129-137. Satake, K., 1994. Mechanism of the 1992 Nicaragua Tsunami earthquake, Geophys. Res. Letters, 21(23): 25192522. Scheidegger, A.E., 1973. On the prediction of the reach and velocity of catastrophic landslides, Rock Mech., 5, 231-236. Shepard, F.P., 1963. Submarine Canyons, in Hill, M., ed., The Sea. John Wiley, New York, 480-506. Silver, P.G. and T.H. Jordan, 1983. Total-Moment Spectra of Fourteen Large Earthquakes, J. Geophys. Res., 88, 3273-3293. Tanioka, Y., L. Ruff and K. Satake, 1997. What controls the lateral variation of large earthquake occurrence along the Japan tranch? The Island Arc, 6, 261-266. Tanioka, Y. and Satake, K., 1996. Fault parameters of the 1896 Sanriku Tsunami Earthquake Estimated from Tsunami Numerical Modeling, Geophys. Res. Letters, 23(13): 1549-1552. Todorovska, M.I. and Trifunac, M.D., 2001. Generation of tsunamis by slowly spreading uplift of the sea floor, Soil Dynam. and Earthq. Eng., 21(2): 151-167. Trifunac, M.D., 1974. A three-dimensional dislocation model for the San Fernando, California, Earthquake of February 9, 1971, Bull. Seism. Soc. Amer., 64: 149-172. Trifunac, M.D., 1993. Long period Fourier amplitude spectra of strong motion acceleration, Soil Dynamics and Earthquake Eng., 12(6): 363-382. Trifunac, M.D., Hayir, A. and Todorovska, M.I., 2001a*. Near-field tsunami waveforms from submarine slumps and slides, Dept. of Civil Eng. Report No. CE 01-01, Univ. of Southern California, Los Angeles, California. Trifunac, M.D., Hayir, A., and Todorovska, M.I., 2001b*. Tsunami waveforms from submarine slides and slumps spreading in two dimensions, Dept. of Civil Eng. Report No. CE 01-06, Univ. of Southern California, Los Angeles, California. Trifunac, M.D. Hayir, A. and Todorovska, M.I., 2002a. A note on the effects of nonuniform spreading velocity of submarine slumps and slides on the near-field tsunami amplitudes, Soil Dynamics and Earthquake Eng., 22(3):167-180. Trifunac, M.D. Hayir, A. and Todorovska, M.I., 2002b. Was the Grand Banks event of 1929 a slump spreading in two directions? Soil Dynamics and Earthquake Eng., 22(5): 349-360. Yokoyama, I., 1987. A scenario of the 1983 Krakatau Tsunami, J. of Volcanology and Geothermal Res., 34: 12313. * Available from http://www.usc.edu/dept/civil_eng/Earthquake_eng/