Giant Waves in Lituya Bay, Alaska; USGS PP 354-C

Nature and Cause of the Wave of July 9, 1958

Steven Dutch, Natural and Applied Sciences, University of Wisconsin - Green Bay
First-time Visitors: Please visit Site Map and Disclaimer. Use "Back" to return here.

Nature And Cause of the Wave

From the foregoing evidence the nature, sequence, and approximate time of events associated with the July 9, 1958, wave in Lituya Bay are interpreted as follows:

Beginning at about 10:16 p.m. the southwest side and probably most of the bottom of Gilbert and Crillon Inlets moved northwestward and possibly up relative to the northeast shore at the head of the bay, on the opposite side of the Fairweather fault. Observations of the surface breakage along the Fairweather fault 6 to 10 miles southeast of Crillon Inlet indicate that the displacement occurred in several pulses and that the total movement was about 21 feet horizontally and 3 feet vertically (Tocher and Miller, 1959). Intense shaking in Lituya Bay continued for at least 1 minute according to the account of William A. Swanson, and possibly as much as 4 minutes according to Howard G. Ulrich. Slides and avalanches started in the mountains at the head of the bay within a minute after the shaking was first felt. Not less than 1 minute nor more than 2-1/2 minutes after the earthquake was first felt a large mass of rock slid from the northeast wall of Gilbert Inlet. The initial movement of this rock mass, with attendant clouds of rock dust and avalanching snow and ice, may account for the "moving glacier" observed by Swanson. The impact of the large rockslide on the surface of the water caused the "deafening crash" heard by Ulrich and caused a huge sheet of water to surge up over the spur on the opposite side of Gilbert Inlet. The sudden displacement of a large volume of water as the rock mass plunged into Gilbert Inlet set in motion a giant gravity wave with a steep front, traveling at high velocity and with its greatest force directed initially about due south. The gravity wave, probably supplemented by the surge of water over the spur southwest of Gilbert Inlet, struck first against the steep cliffs on the south side of the bay in the vicinity of Mudslide Creek; the maximum force of the wave was then reflected and refracted toward the north shore a little farther out the bay, and again back to the south shore near Coal Creek. Variations in the height and intensity of the gravity wave as it moved out the bay, as recorded in the trimlines, may have been caused also by the interaction of diagonally refracted waves, by seiche wave motion, and by reflection of waves from the narrow entrance. Estimates by Ulrich and Swanson of the time elapsed from the first sighting of the wave front until it reached their boats indicate that the crest of the gravity wave moved out the bay at an average speed between 97 and 130 miles per hour. After the giant wave passed, the water in the bay was set into turbulent wave motion and continued to surge from shore to shore for 25 minutes or more.

According to R. L. Wiegel (written communication, Mar. 31, 1959), the wave speed as calculated from the estimated time elapsed is in good agreement with the theoretical speed as calculated from the formula C= sqrt(g (d+H)) where g is the acceleration due to gravity, d is the depth of water below sea level, and H is the height of the wave above sea level. He states

If the water depth averaged between 400 and 500 feet and the wave height averaged between 200 and 300 feet the wave would travel at a theoretical speed of about 100 miles per hour. If the water depth were taken as a conservative 400 feet and the wave height at a conservative 100 feet the theoretical wave speed would be about 86 miles per hour.

From evidence observed and photographed from an airplane on July 10, the writer with Kenneth Loken as pilot, concluded that water had risen to a height of about 1,800 feet on the spur southwest of Gilbert Inlet and caused destruction of the forest to the sharp trimline across this spur (Daily Alaska Empire, 1958b; Seismol. Soc. America Bull., 1958, p. 406). This conclusion was based on the following evidence: (a) The "washed" appearance of the bedrock below the trimline on the spur; (b) the sharp and even appearance of the trimline, and its similarity to and continuity with the trimline known to have been caused by water action farther out the bay; (c) at the highest point on the trimline, where the 1,800-foot altitude was estimated from the airplane altimeter, about 30 large trees were turned upslope and back into the forest. The roots of some of the upturned trees were bare and white, as though they had been washed out rather than merely pulled out of the soil (pl. 7A).

The initial report of wave damage to 1,800 feet above a water surface was widely doubted both on theoretical grounds and on the basis of aerial observations and study of photographs by others. This figure is more than 8 times the maximum height attributed to a [tsunami] 179) and nearly 8 times the maximum height reached by the largest of the slide-generated waves in Norway. Brazee and Jordan (1958), from study of aerial photographs and evaluation of reports of field observations, including those of the writer and Don Tocher, concluded that the spur southwest of Gilbert Inlet "has been denuded to a height of 1,800 feet either by avalanche, wave action or a combination of the two." Jordan later stated (written communication, Dec. 29, 1958) "More information is now available and it seems that landsliding is the major activity for any elevation above 300 feet or so," and this view is expressed also in an announcement of plans for a field investigation of Lituya Bay by the U.S. Coast and Geodetic Survey (Daily Alaska Empire, 1959). T. N. Davis, from aerial observations in Lituya Bay on July 12, 1958 first attributed the destruction of trees at high altitude on the spur southwest of Gilbert Inlet to "earthslide" (paper read at Alaska Science Conf., Sept. 2, 1958), but on reexamination of his photographs he found a few trees stripped of bark high on the slope and now believes that this damage to the trees is more likely due to action of high velocity water than to slide action alone (written communication, Apr. 6, 1959).

After examining the area of the high trimline again from the air and on the ground later in the summer, it is still the writer's conclusion that water was primarily responsible for destruction of the forest cover. Examination on the ground confirmed that trees just above the highest point on the trimline, at 1,720 feet altitude as remeasured by a hand-carried altimeter, had been washed out and overturned by water. At this point on the crest of the spur the water rose about 20 feet higher than the highest overturned trees and flowed across the ridge and at least a quarter of a mile down the opposite side into the forest, leaving rocks and driftwood on the moss. It is true that rockslides either accompanied or closely followed the earthquake on the northeast side of the spur. Cracks trending parallel to the scar were seen in the forest on the crest of the spur, just above the trimline. Comparison of the 1958 oblique photographs taken after the earthquake with the 1948 vertical photographs show, however, that the 1958 slides occurred mainly in old landslide or rockslide scars, and that the volume of new sliding was small. Moreover, the trimline which the writer believes was formed by water, cuts across the tracks of these slides (pl. 4B). After the water had dashed over the spur there was minor sliding from the unstable scarp at the trimline. The conspicuous streaks of debris left by small slides on the otherwise washed, bare bedrock surface of the southwest face of the spur (pl. 7B) provide further convincing evidence against landsliding or avalanching as the primary cause of the destruction here. Also, along the margin of the trimline on the southwest face of the spur from the low point to an altitude of about 700 feet the trunks of many large trees knocked down but not washed out by the water are oriented parallel to the trimline, with their tops turned to the west (pl. 7B). These trees, if felled by avalanching or sliding, should be preferentially oriented parallel to the gradient of the surface.

Small slides occurred, presumably at the time of the earthquake, on the south side of Lituya Bay between Mudslide Creek and Crillon Inlet. The area affected by new slides is much smaller than is shown by Brazee and Jordan (1958, fig. 3). The trimline formed by the wave continues across this area, between slide scars, at altitudes ranging from 500 to 600 feet (fig. 16).

The large mass of rock that plunged into Gilbert Inlet from the northeast wall during the 1958 earthquake is referred to as a rockslide in this report, although it is near the borderline between rockslide and rockfall as defined in two classifications of landslides (Sharpe, 1938, p. 76-78; Varnes in Eckel, 1958, p. 20-32 and pl. 1). This rockslide as stated on page 63 probably caused the 1958 giant wave at Lituya Bay. The rockslide occurred in an area of previously active sliding and gulleying to an altitude of about 3,000 feet on a slope averaging 40�. The rocks in this area, as mapped by D. L. Rossman (written communication, 1957), are mainly amphibole and biotite schists; bedding and schistosity strike about N. 50� W. and dip steeply northeastward, into the slope.

The new slide area on the northeast wall of Gilbert Inlet, as shown on figure 16, was plotted by transferring the outer limits of the new scar by inspection from oblique photographs taken after July 9, 1958, to the vertical photographs taken in 1948, and thence by photogrammetric methods to the map. The dimensions of the slide on the slope are reasonably accurate, but the thickness of the slide mass normal to the slope can be estimated only roughly from the data and photographs now available. The main mass of the slide, as outlined on figure 16, is a prism of rock that is roughly triangular in cross section, with dimensions of 2,400 feet and 3,000 feet along the slope, a maximum thickness of about 300 feet normal to the slope, and a center of gravity at about 2,000 feet altitude. From these dimensions and an assumed specific gravity of 2.7, the volume and weight of the rock mass are, respectively, 40 million cubic yards and 90 million tons. It is highly probable that this entire mass plunged into Gilbert Inlet as a unit at the time of the earthquake, although the only known fact is that it fell between about noon on July 7 and about 10 a.m. on July 10.

The writer went to Lituya Bay in 1958 with a strong belief that fault displacement was the most likely mechanism for generating the giant waves originating in the fault zone at the head of Lituya Bay. The magnitude of the slide on the northeast wall of Gilbert Inlet was not fully realized from the aerial inspection on July 10, and it was first considered to be only a minor factor in the generation of the 1958 wave. Tocher (written communication, Aug. 1, 1958), however, suggested avalanching of rock or ice from the northeast wall of Gilbert Inlet as a possible generating mechanism before he was informed that a rockslide had occurred there. Arguments advanced by Tocher, information obtained later in the field and from the literature on similar waves elsewhere in the world, and the model studies made by Wiegel all have contributed to the writer's present acceptance of the rockslide as the major, if not the sole cause of the 1958 giant wave. Among the arguments against fault displacement as an important contributing mechanism to the generation of this wave, the following seem most significant: (a) Eyewitness reports of a lapse of 1 to 21/2 minutes between the onset of the earthquake and the first sighting of the wave at the head of the bay: (b) The predominantly horizontal movement along the Fairweather fault, as indicated by ground breakage a few miles southeast of Lituya Bay. If the fault trace lies near the northeast side of Gilbert and Crillon Inlets, nearly the entire area under water at the head of the bay moved relatively northwestward and possibly up; wave motion resulting from this displacement should be directed toward the northwest and southeast side of the bay and (or) toward the head of the bay. (c) Vertical displacement of the bottom of the bay along the Fairweather fault probably would generate waves as a line source. An eyewitness account and the configuration of the trimlines, however, indicate radial propagation from a point source in Gilbert Inlet.

The comments of R. L. Wiegel on the nature and cause of the wave follow (written communication, Mar. 31, 1959)

It is a well documented fact that waves with large energy content are generated impulsively by such varying mechanisms as underwater seismic disturbances, islands exploding, atomic bombs, and large masses of water added suddenly to a body of water. The characteristics of waves generated by such mechanisms depend upon the disturbing force and the rate at which it is applied. The resulting waves may be oscillatory in character, nearly solitary in form, a complex multicrested non-linear wave existing entirely above the initial undisturbed water surface, or a bore (Prins, 1958a, 1958b).

 The size of the slide, the water depth, and the general dimensions of Lituya Bay indicated that a wave similar to a solitary wave should form, but with a complex "tail" to the wave. A rough model was constructed at the University of California, at a 1:1,000 scale. Motion pictures were taken of the model tests and measurements were made of the water surface time histories at two points. Observations of the effects of various types of slides in the model indicated that the prototype must have fallen almost as a unit, and very rapidly. If the slide occurred rapidly then a sheet of water washed up the slope opposite the landslide to an elevation of at least three times the water depth. At the same time a large wave, several hundred feet high, moved in a southerly direction, causing a peak rise to occur in the vicinity of Mudslide Creek. This same wave swung around into the main portion of Lituya Bay, due to refraction and diffraction. The movements of the main wave and the tail were complicated within the bay due to reflections and due to the effect of bottom hydrography. One further wave characteristic was noticed when large waves were obtained, and this was that the crest appeared to move at a nearly uniform velocity across the bay even though the water depths at the edge were considerably less than the water depth in the center of the channel. It is believed that this phenomenon is associated with the phenomenon studied by Perroud (1957) 2 The model study movies showed that the wave elevation was higher along the edges of the bay than in the center.

The action of the wave over the center of Cenotaph Island and at La Chaussee Spit are due to shoaling effects which have not been studied in detail for solitary, or similar, waves.

The energy in a solitary wave 100 feet high in water 400 feet deep with a channel width of 8,000 feet can be computed using an equation given by Ippen and Mitchell (1957). It is about 6X1012 foot-pounds. The potential energy of the landslide was about 3.5X1014 foot-pounds. Hence, only about 2 percent of the potential energy of the slide went into the main wave. This is of the same order of magnitude as obtained by model studies of a similar type of disturbance (Wiegel, 1955).

Comparable Waves In Other Parts Of The World

Waves similar to the 1958 giant wave in Lituya Bay have been generated by the sliding of part of a mountain into Shimabara Bay in Japan, by the sliding or falling of large masses of rock into a lake and several fiords in Norway, by the avalanching of a hanging glacier into a bay in Alaska, and by landslides into a lake in Washington. References and significant data on several such localized waves that have come to the writer's attention are summarized on table 1. An exhaustive search of the literature no doubt would reveal many other such occurrences in parts of the world where steep or unstable slopes are adjacent to bodies of water. Earthquakes acted as a triggering mechanism for the slide in Japan, but no earthquake was reported at the time of the 1905 wave in Alaska or at the time of any of the large waves in Norway and Washington. Some waves that accompanied earthquakes in uninhabited or sparsely inhabited areas and were attributed to tectonic movement, as for example the 1899 wave in Yakutat and Disenchantment Bays and Russell Fiord in Alaska (Tarr and Martin, 1912, p. 46 17) may have been generated instead by slides or avalanches triggered by the earthquakes. On the other hand one interpretation of the April 2, 1868 tsunami on the south coast of the Island of Hawaii as the result of a mudflow (Omori, 1907, p. 144) is not correct, according to G. A. Macdonald (written communication, Apr. 15, 1959).

TABLE 1. Data on localized giant waves generated by falling or sliding of solid masses

Location, date, and time of occurrence Generating mechanism Nature of water body, velocity and height of waves Effects of waves References
Shimabara Peninsula, Kyushu Island, May 21, 1792, about 8 p.m. During period of intense earthquakes and volcanic activity about 700 million cu yds of rock and soil to a maximum altitude of 1,700 ft on the east flank of Maye-yama slid 1-3/4 miles down a slope averaging 10�, and plunged into the sea along a front 3 miles wide. Shimabara Bay, length about 60 miles, average width 10 miles, maximum depth 210 ft near the slide; opens into East China Sea at southwest end. At Shimabara 3 waves in rapid succession, the second and largest wave rising on land to a maximum height of about 33 ft. Trees as much as 9 ft in diameter felled, buildings destroyed. More than 15,000 people were killed, most of them by the waves. Wave destruction extended about 50 miles along the shores of the bay. Omori (1907); Ogawa (1924, p. 219-224, pls. 6, 7).
Langfjord, Feb. 22, 1756--- About 15.7 million cu yds of bedrock and soil to a maximum altitude of 1,312 ft on the fiord wall at Tjelle slid down a slope averaging 25� or more, and plunged into the fiord. Landslide may have been triggered by heavy rainfall. Langfjord (fiord), length about 20 miles, average width 1.5 miles, maximum depth about 1,100 ft; opens into Norddalsfjord to west. Three waves observed, rising to a maximum height of 130 ft on shore opposite the slide. Vegetation, soil, buildings, and boats destroyed, 32 people killed. Effects of the waves were noticed as much as 25 miles from slide. Jorstad (1956).
Leon Lake, Jan. 15, 1905, about 11 p.m. About 450,000 cu yds of bedrock and talus to a maximum height of 1,640 ft on Ravnefjell (Raven Mountain) fell and slid down a slope averaging 65�, and plunged into lake. Loen Lake, length 7 miles, average width 0.6 mile, maximum depth 436 ft. Wave 10 ft high in middle of lake; rose to maximum height of 131 ft on shore opposite the slide and to 19 ft at the far end of the lake, 4.8 miles from the slide. Vegetation, soil, buildings, and boats destroyed; iron steamboat 48 ft long was carried 820 ft and stranded 56 ft above lake level; 61 people killed. Holmsen (1936, p. 173-177, figs. 2, 3); Bugge (1937, especially figs. 1, 8, and 10); Brigham (1906); Holtedahl (1953, p.10441045).
Loen Lake, Sept. 13, 1936, 5 a.m. About 1.3 million cu yds of bedrock to a maximum height of 2,625 ft on Ravnefjell fell at the same locality as the 1905 slide. Slide about 1,300 ft wide at lakeshore. Loen Lake, see above. Wave appeared 3-6 ft high in center of lake; it rose to a maximum height of 230 on shore opposite the slide, and to 50 ft at the far end of the lake. Vegetation, soil, buildings, boats and bridges destroyed, 73 people killed. Remains of stranded steamboat carried on up to 164 ft above lake level. Holmsen (1936, p. 183-186, photograph opposite p. 176); Bugge (1937, figs. 8 and 10, p. 357); Holtedahl (1953, p. 1045-1046).
Loen Lake, Sept. 21, 1936, in evening. Rockslide or rockfall from Ravnefjell. Loen Lake, see above. Wave rose to maximum height of 49 ft on shore. Boats used for rescue work were damaged. Holmsen (1936, p. 186).
Loen Lake, Nov. 11, 1936, at night. Rockslide or rockfall from Ravnefjell; volume as large as that on Sept. 13, 1936. Loen Lake, see above. Wave rose to about the same height as on Sept. 13. Nothing left to destroy Holmsen (1936, p. 190), supplement, in German.
Tafjord, April 7, 1934, 3 a.m. Overhanging rock mass of nearly 2 million cu yds volume fell from maximum altitude of 2,395 ft on fiord wall with an average slope of 45�, and plunged into the fiord along a front 750 ft wide. Rockfall triggered by melting of ice in fractures. Tafjord (fiord), length about 5.6 miles, average width 0.7 mile, maximum depth 700 ft; opens into Norddalsfjord to west. Three waves of increasing height were observed at several places. Water rose to maximum height of 204 ft about 650 ft from the slide margin, to 122 ft on shore opposite the slide, and to 3 ft above normal high-tide line about 31 miles from the slide. Approximately measured velocities range from 13.4 to 26.8 miles per hour. Vegetation, soil, buildings, and boats destroyed, 44 people killed along fiord within 2 miles of the slide; extensive damage to boats and docks as much as 31 miles from the slide. Kahldol and Kolderup (1937); Holmsen (1936, p. 177-183, figs. 4 and 5); Bugge (1937, especially figs. 4,5 and 6); Holtedahl (1953, p. 1046).
Norddalsfjord, across from Stranda, 1938.  Landslide from Skafjell. Norddalsfjord (fiord). Three waves reported. Not described in reference. Jorstad (1956, p. 330). Incidental mention only; no detailed description found.
United States
Disenchantment Bay, Alaska, July 4, 1905. Fallen Glacier, a hanging glacier about 3,500 ft long and 1,200 ft wide, avalanched from an altitude of 1,000 ft down a slope averaging about 16�, and plunged into bay along a front 0.5 mile wide. Disenchantment Bay, length about 10 miles, average width 3 miles, maximum depth 942 ft; opens into Yakutat Bay to south and into Russell Fiord to east. Waves 15-20 ft high observed for half an hour on Russell Fiord 15 miles from the avalanche; water rose to maximum height of 115 ft about 2.5 miles from the avalanche. Unconsolidated deposits eroded, bushes broken off or washed out; area uninhabited.     Tarr (1900, p. 67-68). According to Indian legend falling glaciers in this area generated similar waves at least twice before; reportedly 100 Indians were killed by a wave about 1845. 
Reed Terrace area near Kettle Falls, Columbia River valley, Washington; from April 8,1944, to Aug. 19, 1953. Landslides in terrace scarps underlain by bedded unconsolidated deposits. Narrow segments of the scarp on slopes averaging about 23� suddenly gave way and slid into the lake. Debris came down from maximum height of 210 ft above water level. Franklin D. Roosevelt Lake, average width 5,000 ft, maximum depth 160 ft at slide area. Waves were generated by at least 11 different slides; the largest wave rose to maximum height of 65 ft on opposite shore, and was observed 6 miles up the lake. Observed velocity of one series of waves was about 45 miles per hour. Vegetation destroyed, unconsolidated deposits eroded; barges and boats broke loose from dock 6 miles from slide area. F. O. Jones and W. L. Peterson (written communications, Mar. 16 and May 7, 1959); Jones in Eckel (1958, figs. 31, 32, p. 40-41).
Mouth of Hawk Creek near Lincoln, Columbia River valley, Washington, July 27,1949. Landslide in terrace scarp underlain by bedded unconsolidated deposits. A narrow segment of the scarp on a slope averaging about 31' suddenly gave way and slid into the lake. Debris came down from maximum height of 340 ft above lake level. Franklin D. Roosevelt Lake, in bay about 1,200 ft wide and 120 ft deep at slide area. Wave rose 65 ft on shore opposite the slide. Trees knocked down. F. O. Jones and W. L. Peterson (written communications, Mar. 16 and May 7, 1959).
East side of Columbia River valley north of Kettle Falls, Washington, Feb. 23, 1951. Debris slide in bedded unconsolidated deposits and talus from maximum height of several hundred feet above lake level. Franklin D. Roosevelt Lake. Not described in references. Jones in Eckel (1958, fig. 23 on p. 33); W. L. Peterson (written communication, May 7, 1959).

Return to Professor Dutch's Home Page

Created 3 August 2004, Last Update 17 November 2011

Not an official UW Green Bay site