Table 1
The principal information about nine earthquakes that were used in our numerical modeling of tsunami waves in the Bering and Chukchi seas.
№
|
Earthquake
|
Mw
|
Date
|
Lat.
|
Lon.
|
Depth (km)
|
Type of source
|
Reference
|
1
|
Aleutian Islands
|
8.6
|
01.04.1946
|
53.49°N
|
162.832°W
|
15
|
Subfaults1
|
Johnson and Satake, 1997
|
2
|
Kamchatka
|
9
|
04.11.1952
|
52.62°N
|
159.78°E
|
22
|
Subfaults1
|
Johnson and Satake, 1999
|
3
|
Andreanof Islands
|
8.6
|
09.03.1957
|
51.5°N
|
175.63°W
|
25
|
Subfaults1
|
Johnson et al., 1994
|
4
|
Chile
|
9.5
|
22.05.1960
|
38.14°S
|
73.41°W
|
30
|
Seafloor displacement
|
Moreno et al., 2009
|
5
|
Alaska
|
9.2
|
27.03.1964
|
60.91°N
|
147.34°W
|
25
|
Seafloor displacement
|
Suleimani et al., 2020
|
6
|
Rat Islands
|
8.7
|
04.02.1965
|
51.25°N
|
178.72°E
|
30
|
Subfaults1
|
Johnson and Satake, 1996
|
7
|
Tohoku-Oki
|
9.1
|
11.03.2011
|
38.3°N
|
142.37°E
|
29
|
Seafloor displacement
|
Hayes, 2011
|
8
|
Haida Gwaii
|
7.8
|
28.10.2012
|
52.79°N
|
132.1°W
|
14
|
Seafloor displacement
|
Fine et al., 2015
|
9
|
Commander Islands
|
7.7
|
17.07.2017
|
54.44°N
|
168.86°E
|
10
|
Single fault
|
USGS
|
1The seismic model was created using inverse tsunami modeling. |
I. 1946 Aleutian. The Aleutian earthquake Mw=8.6 of 1 April 1946 is believed to be one of the most unusual seismic events of the 20th century. This earthquake generated a very strong trans-Pacific tsunami that hit the coast of Alaska, the Aleutian Islands, Hawaii, North and South America, killed 126 people and seriously damaged Hilo (Hawaii). The maximum runup height (35 m) was recorded near the source at Unimak Island lighthouse (Lander, 1996). The earthquake had moderate surface wave magnitude (Ms=7.4), so the generated tsunami was disproportionally large. That is why the earthquake was identified as the “tsunami earthquake”; the event attracted researchers for many years (e.g. Okal et al., 2003; Lopez et al., 2006; Rabinovich et al., 2019).
We used the seismic source based on the inversion of the tsunami waveform provided by (Jonhson and Satake, 1997). The preferable solution (Model B) consisted of three subfaults. Jonhson and Satake sought an optimal distribution of the slips on 3 subfaults on the plane with strike 250°, dip 6° and found out that the trust faults (rake 90°) at 9.8‒9.6 m slip values provided considerable synthetic tsunami wave forms at most of tide gauge location at Alaska, Hawaii and California. The tsunami was generated by both vertical and horizontal displacements of the seabed. Figure 2a shows the results of the modeling of the 1946 Alaska tsunami in the Bering Sea. The main part of the tsunami energy was reflected by the Aleutian Islands and did not penetrate the Bering Sea. The maximum amplitudes of 50‒60 cm occurred in Bristol Bay (Fig. 2a). In the west Bering Sea, the maximum amplitudes of ⁓10‒15 cm were in the Anastasia Bay. In neighboring Machevna Bay the amplitude reached 13 cm. The amplitudes decreased to 1‒3 cm in the Bering Strait. Waves attenuated even more after passing the strait: the amplitudes were only up to 1 cm in Point Hope and even less (⁓1 cm) on the coast of Wrangel Island. The central part of the Bering Sea, namely St. Lawerence Island and St. Matthew Island, was characterized by the wave amplitudes reaching ⁓10 cm.
II. 1952 Kamchatka. The Great Kamchatka earthquake of 4 November 1952 was one of the largest earthquakes in the 20th century; it had the estimated magnitude of \({M}_{w}\) = 9.0 (Johnson & Satake, 1999). The earthquake created a catastrophic tsunami. In the near-field region, the tsunami reached wave heights > 18 m and killed several thousand people in Severo-Kurilsk (Paramushir Island, Northern Kuril Islands) and southern Kamchatka, thus becoming the most devastating tsunami ever occurred in Russia (Gusiakov, 2014). The earthquake (52.75°N, 159.50°E) had a seismic moment, for different sources ranging from 180×1020 N·m to 350×1020 N·m or \({M}_{w}\) = 8.8‒9.0 (Johnson & Satake, 1999).
The tsunami source was taken from Johnson and Satake study (1999). They sought a slip distribution at 12 subfaults using inverse tsunami modeling technique and tide gauge observations in Alaska, Hawaii, Japan, and California. The optimum solution has an average slip of 3.2 m with a maximum of 11.4 m at the shallow water segment. The detailed source parameters are presented by (Johnson & Satake (1999). Using Okada’s (1985) equations for each subfault, we computed the surface uplift distribution and then summarized all results gained into one composite tsunami source that is finally used in our modeling.
According to our modelled results (Fig. 2b), the tsunami waves generated by this earthquake came into the Bering Sea and had maximum amplitudes of 177 cm on the Russian coast at Machevna Bay. On the east (American) coast of the Bering Sea, the maximum wave amplitude was 66 cm in Norton Sound. Northwardly, the waves were becoming weaker: they were only of 10‒18 cm in the Bering Strait region. Then in the Chukchi Sea, the simulated wave amplitudes were 7 cm in Kotzebue Sound, 3 cm at Point Hope and merely 1.8 cm on the coast of Wrangel Island. In the Bering Sea, at St. Matthew Island the simulated wave amplitude was 49 cm.
III. 1957 Andreanof Islands. This powerful tsunamigenic earthquake (51.50°N, 175.63°E) occurred on 9 March 1957 and had a magnitude of \({M}_{w}\) = 8.6. The maximum tsunami wave amplitude was 32 m. It was recorded at Unalaska Island, the Aleutian Islands (Griswold et al., 2019). At Unimak, near Cape Scotch-Cap, the tsunami wave amplitude was about 22 m according to Lander (1996) or 12‒15 m according to Soloviev & Go (1984). The north coast of Kauai Island (Hawaii) was hit by the waves that reached 16 m (Rabinovich et al., 2019). The tsunami spread throughout the Pacific Ocean and was recorded by more than 50 tide gauges on the coast of the Hawaiian Islands, in Japan and in North and South America (NGDR/WDS tsunami database, NOAA 2022).
We used composite tsunami source reconstructed from subfault model of Johnson et al., 1994 and the Okada (1985) equations. Johnson et al. (1994) applied inverse tsunami modeling and non-negative least squares approach to get slip distribution from a set of 13 subfaults. The rupture zone was located along the Aleutian Ridge from the Semisopochnoi Island to Unalaska Island; the subfaults were located along the Aleutian Arc, so dip was 15°, the strike angle changed from 245° in the east to 270° in the west. The rake angle changed from pure dip-slip in the eastern end of the rupture zone to nearly equal components of dip-slip and strike-slip in the west. The resulting slip was from 0 to 8.6 m, which corresponds to a seismic moment of 88×1020 N m, that is Mw=8.6 (for more details, see Johnson et al., 1994).
The results of the 1957 Andreanof Islands tsunami modeling are shown in Fig. 2c. The energy flux of the tsunami was mainly directed towards the south. The highest simulated tsunami wave in the Bering Sea was 75 cm in Anastasia Bay on the Russian coast. The amplitudes were 49 cm and 45 cm at neighboring sites in Machevna Bay and Olyutorsky Gulf, respectively. The simulated wave amplitude was about 4 cm at Point Hope (Alaska) in the Chukchi Sea. The amplitudes of the computed tsunami waves were 3‒6 cm in the Bering Strait; at Wrangel Island, the maximum amplitude was just 1.4 cm.
IV. 1960 Chile. The Great Chilean Earthquake of 22 May 1960 had a momentum magnitude of \({M}_{w}\) = 9.5 and was the strongest earthquake ever instrumentally recorded; its epicenter was located at 38.37°S, 72.807°W, the depth of the hypocenter was 30 km (Kanamori, 1977). The rupture simulated by the variable slip planer model was approximately 900 km long, 130 km wide, with the maximum slip exceeding 40 m. It had the following angles: Strike = 7°, Dip = 20°, Rake = 80° (Barrientos & Ward, 1990; Moreno et al., 2009). The seismic moment release of this model was 2×1023 N·m. Our input of the 1960 tsunami source is based on the results of the recent seismic model of Moreno et al. (2009), who used coseismic coastal data and a finite element technique to reconstruct the seismic source of the Great Chilean 1960 earthquake. This source was used in the tsunami model of Rabinovich et al. (2011).
Figure 3a shows maximum computed Chilean tsunami amplitudes for the entire Pacific Ocean, the Bering and Chukchi seas. The tsunami energy flux was mainly directed northwestwards to the Marquesas, the Hawaiian Islands, the Philippines and Japan. These particular regions were catastrophically affected by the 1960 Chilean tsunami: The maximum observed far-field tsunami heights were 10.5 m on the Hawaiian Islands, 6‒9 m on other Pacific islands and in Japan and 4‒6 m in Russia (Berkman and Symons, 1960; Rabinovich et al., 2019).
According to our modeling results, the maximum tsunami wave amplitudes associated with the 1960 Chilean earthquake on the Russian coast of the Bering Sea were observed in Machevna Bay (up to 192 cm) and Korfa Bay/Olyutorsky Gulf (140 cm). On the east (Alaskan) coast of the sea, the amplitude of the tsunami wave was up to 135 cm in Bristol Bay. On St. Lawrence Island the wave amplitude was 57 cm. At the entrance into the Chukchi Sea, the wave amplitude values varied from 15 to 24 cm. At Point Hope, where the 1960 local habitats mentioned ice cracking and strong noise during the event, we estimated a wave amplitude to be of only 3 cm. In Kotzebue Sound, it was a little more than 9 cm, while at Wrangel Island (the Chukchi Sea) it was 1.4 cm.
V. 1964 Alaska. On 27 March 1964, the Prince William Sound area of Alaska was struck by the largest earthquake ever recorded in North America and, in general, the strongest in the northern hemisphere. The megathrust earthquake with magnitude of \({M}_{w}=\)9.2 generated the most destructive tsunami in Alaskan history. The tsunamis reached heights with 52 m runup at Shoup Bay (Lander, 1996). About 20 local landslide tsunamis were generated that had runups of up to 70 m (Lander 1996). The tectonic tsunami devastated many towns along the Gulf of Alaska, left severe damage in British Columbia, Hawaii, and along the western coast of the USA.
We used updated coseismic deformation model of Suleimani and Freymueller (2020) to simulate the event. This model is based on the results of Suito and Freymueller (2009) who developed a 3-D viscoelastic seismic source model in combination with an afterslip model. They used realistic geometry with a shallow-dipping elastic slab to describe the postseismic deformation initiated by the 1964 earthquake. The model of Suleimani et al. (2013) enabled us to evaluate the contribution of coseismic horizontal displacements into the initial tsunami wavefield. This additional component of the ocean surface uplift is caused by the horizontal displacement motion on the steep ocean bottom slope.
The energy flux of the 1964 Alaskan tsunami was mainly directed southeastward, to the west coasts of Canada (British Columbia) and the USA (Washington and Oregon) and then to the coasts of Peru and Chile (Fig. 3b). The highest simulated 1964 tsunami wave of 75 cm in the Bering Sea was identified in the Olyutorsky Gulf. It is remarkable that on the opposite (eastern) coast of the Bering Sea its amplitude had a comparable value of 65 cm. It appears that a branch of tsunami waves after penetrating the Bering Sea went around the Alaska Peninsula. On the coasts of the Bering Strait, the computed wave amplitudes were 10‒15 cm. On the west coast, in the Kolyuchin Bay, there were 6-cm waves and about the same on the east, in Norton Bay. At Point Hope, the numerically simulated maximum wave amplitude was 2 cm and 1.3 cm at Wrangel Island. In the middle of the Bering Sea, on St. Matthew Island, the wave amplitude was 15 cm.
VI. 1965 Rat Islands. The strong earthquake (51.25 °N, 178.72°E) occurred on 4 February 1965 with \({M}_{w}\) = 8.7. The earthquake generated a trans-Pacific tsunami, which hit the Aleutian Islands. At Amchitka and Attu Islands, the amplitude of the tsunami wave reached 1.5‒2 m (Soloviev & Go, 1974; Lander, 1996). The height of the water rise was estimated to be 9‒10 m, on the southern coast of Shemya Island (Soloviev & Go, 1974).
We used a slip distribution from (Johnson & Satake, 1996), which is based on the inverse tsunami modeling. The total source was considered as the sum of 14 sub-faults with slips ranging from 0 to 6 m. Each of these sub-faults had a spatial scale of 100⋅60 km with the following angles: strike = 290°, rake = 41.4°, dip = 18°. The seismic moment from the surface wave inversions was estimated from 66 to 140⋅1020 N·m.
The main direction of the Rat Islands tsunami energy flux was toward the southwest (Fig. 3c). In the Bering Sea, the maximum computed amplitude of the tsunami wave was 89 cm in Anastasia Bay and 64 cm at the Ozernoy Cape of Kamchatka) On St. Matthew and St. Lawrence Islands located in the open part of the sea, the wave amplitude was about 40 cm. On the Alaska coast, the amplitudes were smaller: 14‒38 cm near Kuskokwim River. In Bering Strait the amplitude was 15 cm. In the Chukchi Sea, the wave amplitude further decreases to 1.6 cm at Wrangel Island and to 5.2 cm at Point Hope.
VII. 2011 Tohoku-Oki. On 11 March 2011, a subduction megathrust earthquake with a moment magnitude \({M}_{w}\) = 9.1 occurred 32 km below the ocean floor on the east coast of Honshu, Japan (Song et al., 2012). This earthquake and the following tsunami killed almost 20,000 people in Japan and caused widespread destruction. The Tohoku tsunami had a maximum runup of ~ 41 m in Japan (Aomori Prefecture); wave amplitudes were up to 2.5 m on the west coast of the USA and up to 2.3 m at Kuril Islands according to the tide gauge and runup field measurements (Mori et al., 2012; NGDC/WDS, 2022).
In this study, we used the finite-fault seismic source model constructed by Hayes (2011), who used 325 sub-faults, 25⋅20 km each, to estimate slip distribution. The low-angle nodal plane (Dip = 10°, Strike = 194.4°) was selected as the preferred fault plane, with dimensions of 650 km along the strike and 260 km across the strike. The seismic moment release of this model was 4.99⋅1022 N·m which is somewhat smaller than the Global CMT solution (5.59⋅1022 N·m). The model was reformulated by Fine et al. (2013) based on Okada (1985) elastic equations and non-hydrostatic transformation to get a tsunami source. This tsunami source was used to model the wave energy decay and transformation in the Pacific Ocean (Fine et al. 2013).
The results of the 2011 Tohoku modeling are shown in Fig. 4a. The Tohoku tsunami energy spread mainly to the southeast, towards the South America (Peru and Chile). According to our simulations, the maximum amplitude of tsunami waves in the Bering Sea was observed in Machevna Bay (97 cm), at Commander Islands (91 cm) and in Olyutorsky Gulf (71 cm). In general, the waves are higher on the Russian coast comparing to the Alaskan coast, where wave amplitudes were below 39 cm. To the north, in the Bering Strait, the wave amplitudes decreased to 8‒15 cm, and then further decrease in the Arctic; in the Chukchi Sea the largest 7cm high wave occurred in Kotzebue Sound, it was weaker at Point Hope (3 cm) and it had the amplitude of only 1.6 cm on the coast of Wrangel Island.
VIII. 2012 Haida Gwaii. A strong Mw=7.8 earthquake at 52.79°N, 132.101°W and a depth of 14 km occurred on 28 October 2012. This earthquake is likely associated with the relative motion across the Queen Charlotte fault system offshore of British Columbia, Canada. The earthquake was the second most powerful instrumentally recorded earthquake in Canadian history; it generated the largest local tsunami ever recorded on the coast of British Columbia. A field survey on the Pacific side of Haida Gwaii found up to 7.6 m above the tidal level at sites sheltered from the direct influence of incoming waves and up to 13 m in a small unsheltered inlet (Leonard & Bednarski, 2014; Fine et al., 2015). It generated waves of more than 1 m on the coast of the Hawaiian Islands (Fine et al., 2015).
We used the tsunami source model constructed by Fine et al. (2015) from the finite fault model of Hayes (USGS) to simulate the respective tsunami waves. According to our computations (Fig. 4b), the maximum waves were just below 14 cm on the west coast and 7‒9 cm on the east coast of the Bering Sea. In the Chukchi Sea, the wave amplitudes were negligibly small, < 1 cm.
IX. 2017 Commander Islands. A strong earthquake (Mw 7.7) occurred on 17 July 2017 (54.44°N, 168.86°E, hypocenter depth 10 km) near the Commander Islands (Bering Island, Russia) as a result of transform faulting on or near the plate boundary between the Pacific and North America plates in the Northwest Pacific Ocean. This earthquake was the largest earthquake instrumentally recorded in the Bering Sea basin since 1900. The earthquake generated a small tsunami; waves with amplitudes of up to 9 cm were observed at Shemia Island, Alaska (NGDR/WDS tsunami database, NOAA 2022). We used the parameters of the nodal plane from USGS: Strike = 305°, Dip = 86°, and Rake = 163° (USGS) to model the event. The size of the fault and the slip were estimated via scaling law of Wells and Coppersmith.
The results of the tsunami modeling are shown in Fig. 4c. At the coastal stations, the largest amplitude was in Olyutorskiy Gulf, 54 cm. At St. Matthew Island, the simulated amplitude was 14.8 cm, located in the direction of the main energy flux. Another local increase was in the neighborhood of the source on the coast of Kamchatka: 24 cm at the Ozernoy Cape. In the Bering Strait, the amplitude decreased to 0.4‒0.5 cm and in the Chukchi Sea, the simulated tsunami waves were only 0.3 cm.
The tsunami waveforms (Fig. 5) differ significantly depending on the site location and parameters of the seismic source. The strongest earthquakes (Kamchatka, Chile, Alaska, Tohoku) generated waves with a noticeable contribution of low-frequency components (Fig. 5). Weaker earthquakes, which sources were located along the Aleutian Ridge, generated higher-frequency sea level oscillations. This is apparently due to the smaller corresponding source sizes and their extension along with the Aleutian Islands; as a result, relatively high-frequency tsunami waves were directed towards the Bering Sea. Various coastal sites have different frequency-selective properties. This significantly influenced the tsunami waveforms in individual gulfs, bays, and inlets. Thus, in Korfa and Bristol bays, the arriving tsunami waves caused significant intensification of their natural (eigen) oscillations with typical periods of 20‒25 minutes for most events. In Bristol Bay and Kresta Bay, tsunami waveforms have a longer prevailing period (Fig. 5). The highest tsunami waves in the Bering Sea were caused by the 1960 Chile earthquake. In Korfa Bay, the trough-to-creast heights of generated sea-level oscillations reached 2 m and during 10 hours had a steady monochromatic character with a period of ~ 1.5 hours (Fig. 5).