1 Introduction

The highly destructive and deadly tsunamis originating off Sumatra (2004), Chile (2010) and Japan (2011) once again emphasise the serious threat that major seismically generated tsunamis pose to coastal communities within the world oceans. The 2004 Sumatra tsunami killed about 230,000 people in 14 countries bordering the Indian Ocean and is likely the deadliest event of its kind in recorded history. The 2011 Tohoku (Great East Japan) tsunami killed almost 20,000 people, destroyed the Fukushima Daiichi nuclear power plant and caused considerable destruction along the coast of Japan. The 2010 Chile tsunami caused hundreds of deaths and severe damage on the Pacific coast of South America.

In contrast to most other natural hazards, such as earthquakes, volcanic explosions, floods, landslides, and storm surges, which have mainly a regional character, tsunamis generated by the strongest megathrust earthquakes (Mw ≥ 8.5–9.0) are global-scale events that can have devastating effects thousands of km from the source area (Gusiakov 2014). Impact from the Sumatra tsunami of 26 December 2004, induced by an earthquake with an estimated magnitude Mw 9.1–9.3, extended all the way to South Africa, roughly 8000 km from the earthquake epicentre (Rabinovich and Thomson 2007). Global tsunami propagation models (Titov et al. 2005; Kowalik et al. 2007) and observational data confirm the world-wide reach of this tsunami, including far distant regions of the North Pacific and North Atlantic, approximately 26,000 km from the source (Rabinovich et al. 2006). Similarly, the 2011 Tohoku tsunami, produced by a Mw 9.0 earthquake, propagated throughout the entire Pacific Ocean and even penetrated into the Atlantic Ocean, while the 2010 Chile earthquake (Mw 8.8) generated a tsunami, with > 2 m wave heights, on the coasts of California, Japan and the Kuril Islands.

The three major events listed above occurred during the “digital era” and were clearly recorded by hundreds of coastal and open-ocean high-quality gauges. The precise measurement of tsunamis at widely separated locations has enabled scientists to examine the events in detail and has renewed interest in major tsunamis that occurred prior to the establishment of permanent networks of digital tsunami gauges. Tsunami waves recorded by analogue (“pen-and-paper”) coastal tide gauges during the pre-digital era were frequently “chopped off” or “blurred” by instrumental or environmental noise. Using the original records, it was difficult to recognise and isolate the tsunami signal that was superimposed on tides and other longwave processes. Nevertheless, even such sparse information on major events is valuable, especially for the estimation of the tsunami risk and long-term tsunami forecast for coastal areas (cf. Leonard et al. 2014).

Five great trans-oceanic tsunamis were generated by earthquakes in the Pacific Ocean during the pre-digital era of the 20th century: the 1946 Aleutian Islands tsunami generated by a Mw 8.6 earthquake; the 1952 Kamchatka tsunami (Mw 9.0); the 1957 Andreanof Islands (Aleutian) tsunami (Mw 8.6); the 1960 Great Chile (Valdivia) tsunami (Mw 9.5); and the 1964 Alaska (Prince William Sound) tsunami (Mw 9.2) (Abe 1979; Satake and Atwater 2007; Gusiakov 2014). The source areas of the great tsunamis were located along the main subduction zones of the Pacific (Fig. 1). These particular tsunamis played key roles in the establishment of the present International Tsunami Warning System (cf. Kong et al. 2015) and created the foundation for modern tsunami science. All five tsunamis were clearly recorded on the coast of British Columbia (BC). Moreover, prior to the 2011 Tohoku tsunami, these five events were the strongest seismically generated tsunamis ever measured on the BC coast (Fig. 2). The 1964 Alaska tsunami was the most hazardous tsunami ever to impact the west coast of Canada, causing multi-million-dollar damage at Port Alberni and Prince Rupert (White 1966; Murty and Boilard 1970; Murty 1992; Stephenson et al. 2007).

Fig. 1
figure 1

Epicentres (orange stars) of the five great earthquakes of the 20th century (1946 Aleutian Islands, 1952 Kamchatka, 1957 Andreanof Islands., 1960 Chile and 1964 Alaska) and the two of the 21st century (2010 Chile and 2011 Tohoku) in the Pacific Ocean that produced major trans-oceanic tsunamis. The size of the star is proportional to the earthquake magnitude (Mw). The solid blue lines are inverted isochrones of the tsunami travel time (in h) from various parts in the Pacific Ocean to Tofino, British Columbia (denoted by the white circle in the black box). The black box encloses the coast of British Columbia, the main study region

Full size image
Fig. 2
figure 2

Maximum wave heights of tsunamis recorded at a Prince Rupert, b Tofino and c Victoria during the period 1910–2018. Up to 1998, measurements were made by analogue tide gauges; since 1998, digital tide gauges have been used. Red bars indicate tsunami wave heights for the five great tsunamis of the 20th century (1946, 1952, 1957, 1960 and 1964) and two of the 21st century (2010 and 2011); all the others are indicated by brown bars

Full size image

Large segments of the British Columbia coastline are susceptible to tsunamis generated within the Pacific Ocean. Destructive tsunamis have occurred on this coast in the past and will occur again in the future (cf. Clague 2001; Clague et al. 2003). Estimation of the potential tsunami risk to the British Columbia coast is vitally important, especially for areas being considered for new construction or major renovations (Leonard et al. 2014). A thorough examination of the 1946, 1952, 1957, 1960 and 1964 tsunamis is, therefore, highly warranted, but for some reason, has never been done before. In fact, there are only two short reports describing the effects of the 1960 Chile (Wigen 1960) and 1964 Alaska (Wigen and White 1964) tsunamis on the BC coast and very limited information on the other three events (Wigen 1983; Stephenson et al. 2007). One possible reason for this omission is the considerable preliminary effort required to digitise the analogue traces of these tsunamis in the tide gauge records. This work was undertaken as part of the present study.

The purpose of this paper is to provide estimates of the main statistical parameters and physical properties of the recorded waves. The information arising from this study can be effectively used to evaluate the risk from major tsunamis arriving at the coast of British Columbia and, more generally, at the Pacific coast of North America. This information will also help to identify the potential sources of tsunami waves that are the greatest threat to the BC coast and to provide background information in support of the future numerical modelling of tsunami waves for this region.

2 Data

Sea level measurements on the coast of British Columbia have been performed by the Canadian Hydrographic Service (CHS) for more than 110 years. The first tide gauge stations to operate on a permanent basis in British Columbia began in 1906–1909 at Victoria (Juan de Fuca Strait, the southeastern coast of Vancouver Island), Tofino (oceanic, southwestern coast of Vancouver Island), Prince Rupert (the head of Dixon Entrance, the northern mainland coast), Caulfeild/Point Atkinson (the Strait of Georgia, near the city of West Vancouver) and Vancouver (within Burrard Inlet). These five stations (Fig. 3) have the longest records for the Pacific coast of Canada and, in general, among the longest records in North America. Prince Rupert, Tofino and Victoria (Fig. 2) have the longest span of recorded tsunami events on the west coast of Canada. An additional tide gauge was installed at Alert Bay (the northeastern coast of Vancouver Island) in 1948 and in Fulford Harbour (southern coast of Salt Spring Island, the Strait of Georgia) at the end of November 1952 (3 weeks after the 1952 Kamchatka tsunami). The available records from these particular permanent instruments (Fig. 3) were used to examine the five great tsunamis. One more permanent tide gauge was installed in 1961 in Bella Bella and was one of the instruments that most clearly recorded the 1964 tsunami.

Fig. 3
figure 3

Locations of CHS permanent and temporary tide gauges operating on the coast of British Columbia before 1960 (white circles). Five gauges—Tofino, Victoria, Caulfeild, Vancouver and Prince Rupert—were working during three events: 1946, 1952 and 1957. The Tofino and Victoria tide gauges recorded all three tsunamis, the Prince Rupert tide gauge two of them (1952 and 1957), while Caulfeild (Point Atkinson) and Vancouver, located inside the Strait of Georgia and sheltered from incoming waves, did not record any of these tsunamis. The Alert Bay tide gauge on the northeastern coast of Vancouver Island was in operation during the 1952 and 1957 tsunamis and clearly recorded both events. The Kitimat temporary gauge detected the 1952 tsunami; the Fulford Harbour and Bella Coola instruments recorded the 1957 tsunami. The Queen Charlotte Islands are now officially known as “Haida Gwaii”

Full size image

In addition to the permanent CHS network of tide gauges on the coast of British Columbia, there were also temporary gauge deployments, whose purpose was to measure the tidal regime at the respective site. Some of these measurements were conducted during the tsunami events. For example, the 1952 Kamchatka tsunami was recorded by a temporary gauge at Kitimat, while the 1957 Aleutian Islands tsunami was recorded by a temporary gauge at Bella Coola. A large number of temporary tide gauges were operating on the coast of British Columbia in May 1960, when the 1960 Chilean tsunami occurred, specifically McKenney Islands, Griffin Pass, Klemtu, Copper Island, Cape St. James and Barkley Sound. At the time of this event, the CHS was conducting hydrographic surveys in the “Seymour-Belize Inlet system” on the central BC coast. Eight temporary tide gauges were deployed in this region and five of them—Johnson Point, Mereworth Sound, Belize Inlet, Nugent Sound and Seymour Inlet—clearly recorded tsunami waves from the event. The three other gauges—Bamford Lagoon, Frederick Sound and Allison Sound—were too well sheltered from the incoming waves to record the tsunami (Wigen 1960). As a result, a total of 17 tide gauges on the BC coast recorded the 1960 tsunami (Wigen 1960; Stephenson et al. 2007)Footnote 1 and we were able to use these records for our analysis of the event. Likewise, several temporary tide gauges were in operation during the 1964 Alaska tsunami: Tasu Sound, Port Alberni, Ballenas Islands and Port Moody. All of these gauges, as well as all permanent gauges, both those on the outer coast (Prince Rupert, Bella Bella, Alert Bay, Tofino and Victoria) and in the Strait of Georgia (Fulford Harbour, Point Atkinson and Vancouver), recorded the 1964 tsunami. Moreover, several additional records of this tsunami are available from other sources. The record from Ocean Falls on the central mainland coast was obtained through the courtesy of Crown Zellerbach Ltd.; the gauge was located at the power house about 300 m upstream from the river mouth. Four records from the Fraser River delta (North Arm, Steveston, New Westminster and Deas Island) and one record from Pitt Lake (a freshwater tidal lake, ~60 km inland from the sea) were provided by the Water Resources Branch, Department of Northern Affairs and National Resources. Thus, there are a total number of 18 records available for this event (Wigen and White 1964).Footnote 2

Most records of the five great tsunamis on the BC coast are from float-actuated gauges, employing a stilling well with an inlet hole to suppress high-frequency oscillations (Wigen and White 1964). However, the records of the 1964 tsunami at Tasu Sound and Ballenas Islands were from pressure gauges (Wigen and White 1964). The main problem with all these instruments was their analogue (“pen-and-paper”) recording system. Tides are strong on the coast of British Columbia and, in some locations, such as Prince Rupert, exceed 6 m (Thomson 1981). Such large tides make the detection of relatively weak tsunamis from paper records almost impossible. Moreover, this makes problematic the reliable direct estimation of tsunami parameters even for great tsunamis, especially for those that have long-period signals. Sydney Wigen, who collected and examined all available data from the 1960 and 1964 events (Wigen 1960; Wigen and White 1964) and the Tofino records for the entire observational period of 1909–1980 (Wigen 1983), concluded that the minimum tsunami wave height recognisable in these records is about 6 cm.

To estimate the wave parameters of the five major historical events, we digitised all available tsunami records and resampled them to a 1-min time step. The astronomical tides were calculated for all sites and subtracted from the original records; the resulting “residual records” were statistically analysed. The estimated wave parameters for these historical tsunamis were found to be slightly different from those presented by Wigen (1960), Wigen and White (1964) and Stephenson et al. (2007), which were based on direct examination of the paper records.

3 Observations

Tide gauges provide the primary records used in our examination of the five major tsunamis on the coast of British Columbia. However, we also make use of additional information, including that from tsunami catalogues, witness reports, local newspapers and other sources.

3.1 The Aleutian Tsunami of 1 April 1946

The 1946 Aleutian tsunami was generated by a Mw =8.6 earthquake originating 150 km from Unimak Island in the eastern Aleutian Islands. The earthquake generated one of the largest trans-Pacific tsunamis ever instrumentally recorded, with maximum runup heights on the coasts of Alaska and the Aleutian Islands of up to 40 m (Lander 1996). Within 48 min of the earthquake, the tsunami struck Scotch Cap, Alaska and completely destroyed a newly built US Coast Guard Lighthouse, surging over the coastal cliff to a height of 35 m above mean sea level (Lander 1996). Then, 4.5 h after the main shock, destructive waves reached the coasts of the Hawaiian Islands. The waves, which were totally unexpected by the island residents, reached a maximum runup of 17 m at Hilo on the Big Island and 10 and 11 m, on the islands of Maui and Oahu, respectively. The tsunami was also recorded along the entire west coast of the United States and crossed the Pacific Ocean to produce 18–20 m runup at locations in the Marquesas Islands, French Polynesia (Okal et al. 2002) and extended southeastwards, damaging boats in Chile. Altogether, the tsunami killed 168 people, most of them in Hawaii (159), but also at the Scotch Cap lighthouse (5), Marquesas (2), Peru (1), and California (1) (Lander and Lockridge 1989). The property damage in the Hawaiian Islands amounted to $25 million (1946 dollars) (Weeks and Studds 1953). A distinguishing feature of the 1946 tsunami was its pronounced directivity, consisting of a narrow energy beam much like a “spotlight” radiating energy towards the Hawaiian and Marquesas islands, and onward to Antarctica (Fuchs 1982; Gusiakov 2014).

The tsunami was unusually powerful for the magnitude of the earthquake. Due to the large discrepancy between the tsunami magnitude, Mt = 9.3 (Abe 1979), and the surface wave magnitude, Ms = 7.4, the 1946 earthquake was identified as a “tsunami earthquake” (Kanamori 1972; Abe 1973), i.e., as an earthquake generating a tsunami that is much larger than it should be based on its magnitude. The severe destruction and large number of casualties in the Hawaiian Islands prompted the United States to create, in Honolulu, Hawaii, the Seismic Sea Wave Warning System (SSWWS), which, following the 1960 Great Chile tsunami, became the Pacific Tsunami Warning Center (PTWC) (Kong et al. 2015).

The 1946 tsunami was recorded at a number of stations in the Pacific Ocean (Green 1946; Sheppard et al. 1950). Of the five CHS tide gauges in operation at the time of the event (Prince Rupert, Tofino, Victoria, Caufeild and Vancouver; see Fig. 3), the tsunami signal was evident at two: Tofino and Victoria (cf. Stephenson et al. 2007). These two records were digitised and the resulting records de-tided. Significant tsunami waves were recorded at both stations (Fig. 4). The Prince Rupert record was also digitised and de-tided, but no tsunami was detected in the record. The question then is: Why would the 1946 tsunami reach Tofino and Victoria, but not Prince Rupert? This is especially intriguing considering that the four other great tsunamis (1952, 1957, 1960 and 1964) were clearly recorded at this station. The most probable explanation is the narrow focus of the tsunami “beam”, with little energy propagating eastward toward Prince Rupert. The 1946 tsunami was also not recorded at two stations, Caulfeild (Point Atkinson) and Vancouver, located deep inside the Strait of Georgia, and therefore well sheltered from arriving tsunami waves; neither the 1952 nor the 1957 tsunamis were observed at these stations (Sects. 3.2 and 3.3). Therefore, our analysis focuses mainly on the Tofino and Victoria records (Fig. 4). The main parameters of the observed waves are shown in Table 1.

Fig. 4
figure 4

The 1 April 1946 Aleutian tsunami recorded at Tofino and Victoria. The solid vertical red line labelled “E” denotes the time of the earthquake

Full size image
Table 1 Parameters of the Aleutian tsunami of 1 April 1946 recorded by CHS analogue tide gauges on the coast of British Columbia (main shock, Mw = 8.6 at 12:29 UTC)
Full size table

The tsunami waves arrived at Tofino at 16:48 UTC, precisely 4 h and 19 min after the earthquake. The first positive wave (wave crest) had a well-defined amplitude of 18 cm. Remarkably, it was not until more than four hours after the first wave arrival that the maximum wave, with an amplitude of 34 cm, arrived at the Tofino tide gauge (Fig. 4). Two wave periods prevail in this record: 18 and 4.5 min. Exactly one hour after the first wave arrival at Tofino, the first wave arrived at the Victoria tide gauge. The arrival time of 5 h 19 min after the main earthquake shock was, in good agreement with the theoretical travel time and with earlier estimates of arrival times (cf. Stephenson et al. 2007). Based on a paper trace record, Sheppard et al. (1950) estimated the maximum tsunami wave height at Victoria to be 20 cm, while according to our estimates, the maximum wave height was 27 cm. Our estimate is approximately half that of the maximum wave height at Tofino (55 cm). The observed tsunami waves were fairly monochromatic, with a very consistent period of ~22 min (Fig. 4).

To examine temporal variations of the observed 1946 Aleutian tsunami at Tofino and Victoria in the frequency domain, we used a multiple-filter method, which is similar to wavelet analysis (cf. Thomson and Emery 2014; Rabinovich et al. 2006, 2013). The method shows changes of tsunami waves as a function of frequency, f, and time, t, termed “f–t diagrams”. The plots for the 1946 event (Fig. 5) reveal distinct characteristics of tsunami waves at these two stations. Specifically, the arrival times are well-defined and mutually consistent (taking into account the one hour time delay of Victoria relative to Tofino). At Tofino, a broad frequency band of 0.7–20 cph (periods of 85–3 min) of tsunami wave oscillations was induced by the incoming waves, with oscillations with peak periods of 18 and 4.5 min clearly dominant. It is noteworthy that the lower frequency waves (18 min) prevail during the first 12 h following the tsunami arrival at this station, while the higher frequency (4.5 min) dominated afterwards. This abrupt change in the character of the Tofino tsunami oscillations is evident in the record (Fig. 4).

Fig. 5
figure 5

Frequency–time plots (f–t diagrams) for the 1 April 1946 Aleutian tsunami records at Tofino and Victoria. The solid vertical red line labelled “E” denotes the time of the earthquake; the dashed white line indicates the tsunami arrival

Full size image

During the 1946 event the tsunami waves at Victoria were substantially different than those at Tofino. As illustrated by the f–t diagram for Victoria (Fig. 5), tsunami oscillations at this station were characterised by a narrow and persistent frequency band of enhanced energy with a peak period of 22 min, compared to 18 min for Tofino. There is also another amplified frequency band with a period of ~55 min at Victoria, which is not noticeable in the record but obvious in the corresponding f–t diagram. The same two spectral peaks for Victoria were obtained by Rabinovich and Stephenson (2004) based on analysis of 9-day background oscillations at this site and by Rabinovich et al. (2018) based on examination of several tsunamis detected at this site. The first period (22 min) is likely associated with the fundamental eigen period of Victoria’s Inner Harbour, while the second period (55 min) appears to be related to the resonant properties of the outer basin and Juan de Fuca Strait (Rabinovich et al. 2018).

3.2 The Kamchatka Tsunami of 4 November 1952

The 1952 tsunami was generated by a Mw = 9.0 earthquake originating near the southern end of the Kamchatka Peninsula, Russia. In the near-field region, the tsunami attained wave heights > 18 m and killed several thousand people in Severo-Kurilsk (Paramushir Island, Northern Kuril Islands) and southern Kamchatka, making it the most devastating tsunami in Russian history (Gusiakov 2014). Six hours after the initial earthquake, 6–8 m waves struck the Hawaiian Islands, reaching a maximum height of 9.1 m on the easternmost coast of Oahu Island (Lander and Lockridge 1989). However, no fatalities were reported from the far-field areas affected by this tsunami and the property damage in the Hawaiian Islands, estimated at $0.8–1.0 million (in 1952 dollars), was significantly smaller than in 1946. The fact that no lives were lost and that the damage was relatively small, is mainly attributed to the seismic sea wave warning system organised on the Hawaiian Islands after the 1946 catastrophic event (Weeks and Studds 1953).

The 1952 tsunami was recorded by many coastal tide gauges around the Pacific Ocean, including gauges in Peru, Chile and numerous islands in the Pacific. As noted by Weeks and Studds (1953), “This <1952> tsunami was probably recorded by more tide gauges than any other before”. The authors of this report also indicate that “Even a few stations up tidal tributaries from the sea recorded it”. As an example, they referred to Kitimat, BC, located 80 km up Douglas Channel (Fig. 3).

The 1952 Kamchatka tsunami was recorded at several locations along the coast of British Columbia. According to Weeks and Studds (1953), Wigen (1983) and Stephenson et al. (2007), the maximum wave height at Tofino was 58 cm, at Victoria and Alert Bay ~40 cm and at Kitimat noticeable, but small. We digitised and re-examined all five records. The de-tided and high-pass filtered tsunami records are shown in Fig. 6 and the estimated parameters of the measured tsunami waves presented in Table 2.

Fig. 6
figure 6

The 4 November 1952 Kamchatka tsunami recorded at Prince Rupert, Kitimat, Alert Bay, Tofino and Victoria on the coast of British Columbia. The solid vertical red line labelled “E” denotes the time of the earthquake

Full size image
Table 2 Parameters of the Kamchatka tsunami of 4 November 1952 recorded by five analogue CHS tide gauges on the coast of British Columbia (main shock, Mw = 9.0 at 16:58 UTC)
Full size table

Well-defined tsunami waves, including the first wave arrival, were observed at all five stations, including Kitimat. The tsunami first arrived at Tofino at 00:30 UTC on 5 November 1952 (7 h 32 min after the earthquake), 13 min later at Prince Rupert (00:43 UTC), then at 01:17–01:19 UTC at Kitimat and Alert Bay, and finally, at 01:40 UTC at Victoria. The maximum observed wave heights were approximately 77 cm (Tofino), 39 cm (Victoria), 35 cm (Alert Bay), 24 cm (Prince Rupert) and 17 cm (Kitimat). Results from the digitised records (see Table 2) can be compared with those previously estimated by Weeks and Studds (1953) and Wigen (1983) directly from the analogue tide gauge records. The arrival times for Tofino, Prince Rupert, Alert Bay and Victoria were indicated to be: 00:42, 02:20, 01:18 and 01:40 UTC. Thus, some of the earlier estimates were quite precise, while others were not. In particular, the Kitimat record could not be examined for the earlier study, while for Prince Rupert the only estimated parameter was the arrival time and the error in this parameter was 1 h 37 min. The main reason for the latter discrepancy is that the tsunami signal at Prince Rupert has a very low-frequency (Fig. 6) and is difficult to discern in the original analogue record.

In general, the 1952 tsunami parameters at the five stations, estimated from the digitised and de-tided tide gauge records (Table 2), significantly correct the results of earlier studies. In particular, the 77-cm tsunami wave height measured at Tofino was higher than any other tsunami earlier recorded on the coast of British Columbia (Fig. 2). Remarkably, this wave occurred 15.1 h after the first tsunami arrival (Table 2, Fig. 6), consistent with the very long (several days) “ringing” and slow energy decay of the tsunami waves observed at all BC stations (Fig. 6). This property was noted by Weeks and Studds (1953), who examined the analogue records at many stations in the Pacific Ocean and found that the typical duration of the 1952 tsunami oscillations was from 3.5 to 5.5 days. At most of the BC stations, as well as at many other stations (Weeks and Studds 1953), maximum waves were observed several hours after the first wave arrival.

Although the maximum wave height recorded at Victoria, ~39 cm, was about one half that at Tofino, it was higher than any tsunami recorded at this site prior to 1952 (Fig. 2). Tsunami waves were noticed by shipping in Victoria Harbour, but there was no damage (Stephenson et al. 2007). The District Engineer (CHS, Victoria) reported “This wave was of somewhat greater amplitude on our coast than the one of 1946 and continued to oscillate for a considerably longer time…” (Weeks and Studds 1953).

The 1952 event was the first tsunami ever recorded or reported for Alert Bay and Kitimat. The fact that this wave was so clearly recorded at Kitimat is especially remarkable considering that the station is located at the head of Kitimat Arm, the innermost inlet at the end of the long and narrow Douglas Channel, roughly 80 km from Hecate Strait (Fig. 3). The tsunami parameters, which were unrecognisable in the original paper record (cf. Weeks and Studds 1953; Stephenson et al. 2007), were easily determined from the digitised and de-tided record (Table 2, Fig. 6).

Tsunami waves were not detected at Prince Rupert during the 1946 Aleutian event but were clearly evident in the 1952 Kamchatka tsunami record (Fig. 6). The waves measured at Prince Rupert had much longer periods than those at Tofino and Victoria (Fig. 6). Wave periods at Kitimat and Alert Bay were intermediate between these sites in that both low and high-frequency waves prevailed. These particular features of the tsunami oscillations are visible in the f–t plots of the records shown in Fig. 7.

Fig. 7
figure 7

Frequency-time plots (f–t diagrams) for the 4 November 1952 Kamchatka tsunami records at five tide gauge stations on the coast of British Columbia. The solid vertical red line labelled “E” denotes the time of the earthquake; the dashed white line indicates the tsunami arrival

Full size image

The f–t diagram for Prince Rupert is characterised by a dominant frequency band with a peak period of ~110 min. There are no high-frequency oscillations with f > 0.8 cph (i.e., periods T < 75 min). As observed in previous studies (cf. Rabinovich and Stephenson 2004; Rabinovich et al. 2018), tsunamis generate relatively high wave energy at low frequencies (with peak frequencies at 0.5–0.6 cph; periods of 100–120 min) but little wave energy at high-frequencies. It appears that Dixon Entrance and the shelf adjacent to Prince Rupert act as an effective low-pass filter that strongly suppresses high-frequency waves, incoming to this region (this effect for that region was discussed in detail by Rabinovich et al. 2018).

The f–t diagrams at Kitimat and Alert Bay are similar. Low-frequency motions are predominant at both sites (with a peak period of about 100 min at Kitimat and 70 min at Alert Bay), but as are relatively high-frequency oscillations with periods of 25–30 min (Fig. 7). No oscillations with periods shorter than 15 min were observed at either Kitimat or Alert Bay, indicating that these oscillations are strongly suppressed by the long channels and straits of the region.

The f–t diagrams for the 1952 tsunami at Tofino and Victoria are similar, as are the f–t diagrams for the 1946 tsunami (Fig. 5). At both stations, arriving tsunami waves induce broad-band (0.5–5 cph) oscillations with two frequency bands of enhanced wave energy: (1) a predominant, shorter-period band with periods of ~22 min at Tofino and 25 min at Victoria; and (2) a secondary, longer-period band with periods of 70 min at Tofino and 55–60 min at Victoria. The latter bands are clearly evident in the f–t diagrams, but not distinguishable in the tsunami records themselves (Fig. 6).

3.3 The Andreanof Islands (Aleutian) Tsunami of 9 March 1957

The tsunami of 9 March 1957 was generated by a Mw =8.6 earthquake originating south of the Andreanof Islands group, in the central part of the Aleutian Islands (Fig. 1). Based on the aftershock distribution, Johnson et al. (1994) estimated the source area of the earthquake as one of the longest of any earthquakes instrumentally recorded, stretching almost 1200 km along the Aleutian Trench. The earthquake generated a major tsunami that was recorded by tide gauges throughout the entire Pacific Ocean (Salsman 1959). The tsunami wave heights on the coasts of some of the unpopulated Aleutian Islands were estimated to be more than 20 m high, with a maximum observed wave height of 22.8 m at Umnak Island (Lander 1996; https://www.drgeorgepc.com/Tsunami1957.html). The north coast of Kauai Island (Hawaii) was hit by waves up to 16 m high. However, the maximum recorded wave height was only 3.6 m—Kahului on the Island of Maui in Hawaii. The Hawaiian Islands suffered the most, with infrastructure damage of approximately $5 million (in 1957 US dollars). Timely tsunami warnings prevented any direct fatalities in the far-field areas affected (Lander and Lockridge 1989).

The effects of the 1957 event were poorly documented for the coast of British Columbia. An analysis of the analogue records by Wigen (1983) indicated that the maximum wave height at Tofino was 52 cm, but he did not examine other records, probably because the tsunami signal was weak and could not be readily extracted from tidal analogue records. For the present study, we digitised all available paper records of this event, except for Port San Juan (Port Renfrew), which was too noisy. We found that the tsunami was recorded at six stations: at five permanent stations—Prince Rupert, Alert Bay, Tofino, Victoria and Fulford Harbour—and at a temporary station at Bella Coola (see Fig. 3 for the location of these stations). The de-tided and high-pass filtered tsunami records for these six stations are shown in Fig. 8 and the estimated parameters of the observed tsunami waves are given in Table 3.

Fig. 8
figure 8

The 9 March 1957 Andreanof Islands tsunami recorded at six stations on the coast of British Columbia. The solid vertical red line labelled “E” denotes the time of the earthquake; the arrows indicate the tsunami arrival

Full size image
Table 3 Parameters of the Andreanof Islands tsunami of 9 March 1957 recorded by six analogue CHS tide gauges on the coast of British Columbia (main shock, Mw = 8.0 at 14:23 UTC)
Full size table

As with the 1952 tsunami event, Tofino was the first station on the BC coast to record the 1957 tsunami. The leading wave arrived at Tofino at 18:46 UTC on 9 March 1957 (4 h 23 min after the earthquake) and, 21 min later, at Prince Rupert (19:07 UTC). Subsequent recorded arrival times were 19:23 UTC at Alert Bay, 19:44 UTC at Bella Coola, 20:05 UTC at Victoria and lastly 20:50 UTC at Fulford Harbour. Once again, the maximum wave height among all stations was recorded at Tofino: ~48 cm. The observed wave heights at other stations were significantly smaller, ranging from 11 cm at Fulford Harbour to 26 cm at Victoria (Table 3). Perhaps the most remarkable feature of the measured waves on the BC coast is that the 1957 tsunami was recorded at Bella Coola. This station is located at the head of North Bentinck Arm, which is the remotest part of the long and narrow Burke Channel. The distance from the open ocean (Queen Charlotte Sound) to this station is ~110 km (Fig. 3). Nevertheless, waves with a maximum height of 15 cm were recorded at this station less than one hour after they arrived at Tofino. A notable property of the 1957 tsunami is the consistency of the visible period of 22–25 min observed at all stations, except Prince Rupert, where this period is 110 min (Table 3), in agreement to what was observed at this station during the 1946 and 1952 events. Additional details of the recorded tsunami oscillations at various stations are seen in their f–t diagrams (Fig. 9).

Fig. 9
figure 9

Frequency-time plots (f–t diagrams) for the 9 March 1957 Andreanof Islands tsunami records at six tide gauge stations on the coast of British Columbia. The solid vertical red line labelled “E” denotes the time of the earthquake; the dashed white line indicates the tsunami arrival

Full size image

The specific attribute of all f–t diagrams, except Prince Rupert (i.e., for Bella Coola, Alert Bay, Tofino, Victoria and Fulford Harbour), is their marked similarity.

  1. (1)

    The main energy of observed tsunami waves spans the frequency band of 1.0–5.0 cph (i.e., periods of 60–12 min).

  2. (2)

    The dominant frequency at the five stations is almost the same: ~ 2.5 cph (T ~ 24 min).

  3. (3)

    The maximum tsunami waves were associated, not with the leading waves, but with a train of waves that arrived 6–8 h later. This aspect of the incoming tsunami waves is visible in the records themselves (Fig. 8).

A particular feature of the Alert Bay and Victoria f–t diagrams is the presence of a secondary frequency band of amplified-energy at frequencies 0.9–1.1 cph (T ~ 55–65 min). For these two stations, approximately the same amplified frequency bands (~ 2.5 cph and ~ 0.9–1.1 cph) were also evident in the f–t diagrams for the 1952 tsunami (Fig. 7); however, for Alert Bay in 1952, low-frequency oscillations (~ 0.9 cph) prevailed, while higher-frequency oscillations (~ 2.5 cph) were secondary.

The f–t diagram for the 1957 tsunami at Prince Rupert is almost identical to that for the 1952 tsunami (Fig. 7). Specifically, there is one well-defined frequency band with a peak period of ~110 min and no oscillations at frequencies higher than 0.85–0.9 cph (i.e., at periods shorter than 65–70 min).

3.4 The Chilean Tsunami of 22 May 1960

The Chilean tsunami of 22 May 1960 was generated by the Mw = 9.5 Great Chilean (Valdivia) Earthquake, the strongest earthquake ever instrumentally recorded on land or in the ocean. The main shock of the earthquake occurred at 19:11 UTC; the earthquake lasted for about 10 min and the rupture length was approximately 900 km (Satake and Atwater 2007). The tsunami from this event was the largest of the 20th century. In southern Chile, about 1655 people were killed, 3000 injured and two million displaced; 61 people lost their lives on the coasts of the Hawaiian Islands, 142 in Japan, 32 in the Philippines and 2 on the west coast of the USA (Kong et al. 2015). The high degree of destruction and loss of life in a number of the Pacific countries located far from the source area was the impetus for international cooperation in tsunami research and mitigation, and resulted in the establishment of the International Tsunami Warning System in the Pacific (ITSU) and the Pacific Tsunami Warning Center (PTWC) in Honolulu, Hawaii, based on the former Seismic Sea Wave Warning System (SSWWS) (Kong et al. 2015).

The 1960 Chilean tsunami propagated throughout the entire Pacific Ocean and even penetrated into the Atlantic and Indian oceans (Van Dorn 1984). The maximum observed far-field wave 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. The tsunami was recorded by an unprecedented number of about 250 tide gauges, including far-field sites at the Aleutian Islands and in the Sea of Okhotsk, located 16–18 thousand km from the source area (Berkman and Symons 1960; Takahashi and Hatori 1961). Strong oscillations with trough-to-crest wave heights of more than 3–5 m were recorded at a number of stations on the coast of Alaska (Lander 1996), at sites relatively close to Prince Rupert. At many tide gauge stations around the Pacific, extreme positive or negative tsunami wave heights were “chopped off”; the instruments were simply not built for such strong oscillations (Berkman and Symons 1960).

Altogether, there were 21 CHS tide gauges in operation on the coast of British Columbia in May 1960. Of the 7 permanent and 14 temporary gauges, 17 of them (6 permanent and 11 temporary; Fig. 10) recorded the 1960 Chilean tsunami (Wigen 1960). The exceptions were the permanent tide gauge at Vancouver and three sheltered temporary gauges installed in the “Seymour-Belize Inlet system” (see Sect. 2 for details). There were also numerous eye-witness reports of strong oscillations along the BC coast. In particular, at Shields Bay located on the west coast of Graham Island, the Queen Charlotte Islands (the bay is indicated in Fig. 10), loggers reported that the water poured in and out of the bay with a violence resembling Seymour Narrows (where maximum tidal currents are 14–16 knots); from photos made in the bay during these disturbances it appeared that the maximum oscillation heights were more than seven feet (2.1 m). One of the most striking features of the 1960 tsunami was the transmission of waves through fast running narrow passages, like Nakwakto Rapids, connecting the Seymour Inlet area to the ocean (see the inset in Fig. 10). No damage or loss of lives caused by the tsunami was reported, but an Indian village on the west coast of Vancouver Island was partly inundated and moored log booms were torn adrift along this coast (Wigen 1960).

Fig. 10
figure 10

Location of CHS tide gauges on the coast of British Columbia that recorded the 1960 Chile tsunami. The white circles denote permanent CHS stations, the yellow circles are temporary stations. The inset shows five stations deployed in narrow mainland inlets and channels of the central BC coast (Seymour–Belize inlet system) as part of the CHS hydrographic study of this region; “QC Str.” denotes Queen Charlotte Strait. The Queen Charlotte Islands are now officially known as “Haida Gwaii”

Full size image

From an observational perspective, the entire situation with the 1960 tsunami tide gauge measurements was completely different from that for the 1946, 1952 and 1957 events. While tsunami waves for the previous events had been recorded at only a few BC stations, the 1960 tsunami was observed at 17 stations, many of them located in regions where tsunami waves had never been observed before. Moreover, the 1960 tsunami was much stronger than any of the previous events recorded on the BC coast (cf. Fig. 2). All 17 existing records were examined by Wigen (1960) using the original paper records, in which tsunami oscillations were superimposed on tides. Thus, we could compare the main characteristics of the observed tsunami waves estimated from digitised and de-tided records with those estimated by Wigen (1960).

One-day records of the 1960 tsunami at the 17 BC tide gauges were digitised and examined in the same manner as for the 1946, 1952 and 1957 events; the evaluated statistical parameters of this tsunami are given in Table 4, the de-tided and high-pass filtered records are shown in Figs. 11 and 12. The arrival times and maximum tsunami wave heights from Wigen (1960) are presented in the respective columns in Table 4 in brackets.Footnote 3

Table 4 Parameters of the Great Chile tsunami of 22 May 1960 recorded by CHS analogue tide gauges on the coast of British Columbia (main shock, Mw = 9.5 at 19:11 UTC)
Full size table
Fig. 11
figure 11

The 22 May 1960 Great Chile Tsunami recorded at nine stations located on the northern mainland coast of British Columbia and on the outer coast of Vancouver Island. The arrows indicate the tsunami arrival. The negative denivelation (“trough wave”) preceding the arrival of the frontal tsunami wave is related to the elasticity of the Earth’s core deflecting under the weight of the propagating tsunami wave (see details in Watada et al. 2014; Eblé et al. 2015)

Full size image
Fig. 12
figure 12

The same as in Fig. 11, but for a five stations located in inlets and channels of the Belize-Seymour Inlet system on the mainland coast of British Columbia, and b a group of southern BC stations consisting of Victoria (Juan de Fuca Strait) and Fulford Harbour and Caulfeild (the Strait of Georgia)

Full size image

Comparison of Wigen’s (1960) estimates with those from our study indicates that in most cases they agree quite well. The corrections in Wigen’s arrival times are small: from 0 min (Alert Bay) to 8 min (Griffin Pass). There is only one exception—Belize Inlet—where the arrival time was corrected by 25 min. Similarly, maximum wave heights differed mainly by a few cm, except at the Barkley Sound site, which was corrected by 27 cm (Table 4). The problem, however, was that Wigen (1960) could not identify the tsunami arrival times at four stations (Prince Rupert, Seymour Inlet, Fulford Harbour and Caulfeild) and maximum wave heights at two stations (Prince Rupert and Caulfeild). In the newly digitised and de-tided records, the tsunami arrivals and maximum wave heights at all stations were clearly observed and these parameters were estimated (Table 4).

Cape St. James, located at the southernmost end of the Queen Charlotte Islands (Fig. 10), was the first station on the coast of British Columbia to detect the 1960 Chile tsunami. The wave arrived at Cape St. James at 12:17 UTC, corresponding to a travel time from the source area near southern Chile of 17 h and 6 min, in good agreement with the theoretical travel time (Fig. 1). Roughly 6–7 min later, the tsunami arrived at Tofino and Barkley Sound, and then at other stations on the outer coast (Table 4). At 12:57 UTC (34 min later than at Tofino), the tsunami arrived at Victoria. The wave then travelled through the Strait of Georgia to Fulford Harbour, arriving at 13:31 UTC, and then to Caulfeild, arriving at 14:02 UTC. The 1960 tsunami also entered Queen Charlotte Sound and propagated northward across the wide and relatively shallow waters of Hecate Strait (Fig. 10), eventually reaching Copper Island (12:42 UTC), McKenney Islands (13:02 UTC), Klemtu (13:13 UTC) and Griffin Passage (13:28 UTC) (Table 4). The wave then reached Prince Rupert at 14:45 UTC. It is not known, without high-resolution numerical modelling, if the first tsunami wave propagated from the south through Hecate Strait or from the west through Dixon Entrance (Fig. 10). The results of numerical simulation of the 2010 Chilean tsunami (see Fig. 5 in Rabinovich et al. 2013), which had the source approximately at the same area as the 1960 tsunami, showed the tsunami waves reached Prince Rupert almost simultaneously by both routes.

Another branch of the 1960 tsunami in Queen Charlotte Sound went around the northern coast of Vancouver Island and at 13:45 UTC was recorded at Alert Bay, on the northeastern side of the island. The tsunami waves also reached the mainland coast and penetrated into the narrow fjords and channels, typical of this coast. After passing through Nakwakto Rapids (see the inset in Fig. 10), the waves entered the Seymour–Belize Inlet System, where they were measured at five stations: Johnson Point, Nugent Sound, Seymour Inlet, Belize Inlet and Mereworth Sound. The last arrival time was at Mereworth Sound at 16:44 UTC, i.e., 21 h 33 min after the main earthquake shock and 4 h 27 min after arrival at Cape St. James, which, as noted earlier, was the first station on the BC coast to record the 1960 tsunami. Thus, in approximately 4.5 h, the 1960 Chilean tsunami spread along the entire coast of British Columbia, even propagating into the innermost inlets, fjords and channels of the mainland coast.

At two stations, Tofino and Barkley Sound, the maximum observed tsunami wave heights were 132 and 131 cm, respectively. Once again, the highest tsunami wave was recorded at Tofino. At six other stations, including Victoria (73 cm), it was more than 40 cm (Table 4). For most stations, the 1960 tsunami was the first tsunami ever recorded, while for Tofino, Victoria and Alert Bay, this was the highest recorded tsunami. An important feature of the 1960 tsunami, which was observed not only in British Columbia but throughout the entire Pacific Ocean, was very long ringing (> 4–5 days) and slow energy decay (cf. Berkman and Symons 1960; Takahashi and Hatori 1961; Miller et al. 1962). The maximum tsunami amplitudes were commonly observed within several hours after the arrival of the first wave. For example, at Victoria, the first waves were weak and had a period of T ~ 55 min; however, roughly four hours after the first wave arrival, a train of prominent waves with much shorter periods (T ~ 22 min) arrived. The later wave trains appear to be due to wave dispersion during the long oceanic trajectory of the tsunami. The maximum trough-to-crest height of 73 cm at Victoria was associated with this train of waves, with the maximum wave arrival at 19:48 UTC (Table 4), almost 7 h after the leading wave.

Pronounced sea level oscillations in several harbours were accompanied by strong currents. This once again demonstrates the prolonged threat of tsunami waves and associated tsunami-induced currents for coastal communities and infrastructure. The typical visual periods of the recorded waves were 20–30 min, but at some BC coastal stations, such as Prince Rupert and Caulfeild, much longer periods were observed (Table 4).

Based on the characteristics of the 1960 tsunami waves, evident in the f–t diagrams for BC coastal sites (Figs. 13 and 14), we can divide all stations into three distinct groups:

  1. (1)

    Low-frequency (Prince Rupert, Caulfeild and Seymour Inlet). The tsunami waves recorded at Prince Rupert and Caulfeild are characterised by prolonged monochromatic oscillations with a distinct long period of ~100 and 180 min, respectively. At Seymour Inlet, long-period oscillations (~120 min) are also strongly predominant, but they continue for only a few hours; weak multi-frequency oscillations arrive later.

  2. (2)

    Moderate-frequency (Belize Inlet, Mereworth Sound and Fulford Harbour). Tsunami oscillations at all three stations were monochromatic, with one predominant period, corresponding to 22 min at Belize and Mereworth and 30 min at Fulford.

  3. (3)

    Polychromatic (other 11 stations). At all other stations, at least two and up to three, frequencies prevail with typical periods between 10 and 90 min.

Fig. 13
figure 13

Frequency-time plots (f–t diagrams) for the 22 May 1960 Great Chile tsunami records at nine stations located on the northern mainland coast of British Columbia and on the outer coast of Vancouver Island. The solid vertical red line labelled “E” denotes the time of the earthquake; the dashed white line indicates the tsunami arrival

Full size image
Fig. 14
figure 14

The same as in Fig. 13 but for eight stations located in Juan de Fuca Strait, the Strait of Georgia and in inlets and channels of the west mainland coast

Full size image

An important feature of the f–t diagrams for Tofino, Victoria, Prince Rupert and Alert Bay (i.e., for those stations that measured other great tsunamis) is the similarity of the plots at each station for different events. In fact, it is well known that the main spectral peaks for different tsunamis at the same site are very similar, while the spectral peaks, from the same tsunami event, even at neighbouring sites are substantially distinct (cf. Honda et al. 1908). This property was considered in detail by Rabinovich (1997) (see also Fig. 8 and the corresponding discussion in Zaytsev et al. 2017). The similarity of the f-t diagrams at the same site for different events demonstrates the dominant influence of the local topography over the individual characteristics of the tsunami source.

3.5 The 1964 Alaska (“Good Friday”) Tsunami of 28 March 1964

The Alaska earthquake of 28 March 1964 with magnitude Mw = 9.2 produced a catastrophic trans-Pacific tsunami, the second strongest in the 20th century (after the 1960 Chilean tsunami). The maximum water rise was 20 m in the source area, while numerous subaerial and submarine landslides, initiated by the earthquake generated local tsunamis, had runups of up to 70 m (Lander 1996). The earthquake, which was the greatest instrumentally recorded earthquake in the North Pacific Ocean and one of the most powerful earthquakes ever recorded, occurred in the region of Prince William Sound, leading to the widely used name of the “Prince William Sound Earthquake” (Spaeth and Berkman 1967; Johnson et al. 1996). Due to the date (28 March 1964), this earthquake and associated tsunami are also called the “Good Friday Earthquake and Tsunami”. The event produced the strongest tsunami response ever observed on the coast of British Columbia. Because the 1964 Alaska tsunami was the highest and the most important tsunami to impact the coast of British Columbia, we give more attention to this event than to any of the earlier tsunamis.

The 1964 earthquake occurred within the Alaska-Aleutian megathrust zone, where the Pacific Plate subducts under the North American Plate (Fig. 15). This zone has the greatest potential to generate destructive tsunamis and is one of the most seismically active fault zones in the North Pacific [cf. The Great Alaska Earthquake of 1964 (1972)]. The earthquake rupture extended for 600–800 km from Prince William Sound along the Kenai Peninsula to Kodiak Island and included a 200–250 km wide zone (Fig. 15) with an average fault slip of more than 10 m (Suito and Freimueller 2009). The 1964 megathrust Alaska earthquake caused the most destructive tsunami in Alaskan history and, further south, strongly impacted the west coasts of the USA and Canada (Johnson et al. 1996; Myers and Baptista 2001; Suleimani et al. 2013). In addition to the major tectonically-generated tsunami, more than 20 local tsunamis were generated by landslides in coastal areas of Alaska. Of 132 fatalities associated with the 1964 earthquake, 122 were caused by the tsunami (Lander 1996). The tsunami spread over the entire Pacific Ocean and was recorded by a great number of coastal tide gauges (Spaeth and Berkman 1967), including those located on the coast of British Columbia (Wigen and White 1964; herein, WW64 in the following text).

Fig. 15
figure 15

(modified from Suleimani et al. 2013). Tide gauges at Sitka, Juneau and Anchorage are indicated by solid black circles; the red circle denotes the tide gauge at Prince Rupert

Map of south-central and southeastern Alaska with the rupture zones of six major historical earthquakes (shaded); the rupture zone of the 1964 Alaska earthquake is contoured by a thick brown dashed line. Red stars indicate epicentres of the earthquakes; the sizes of the stars are proportional to the earthquake magnitudes

Full size image

The tsunami swept southward from the source area in Prince William Sound along the British Columbia coast, causing about $10 million in damage (1964-dollar values). If such an earthquake and tsunami were to occur today, the cost of the damage would be a factor of 10–100 times greater (Rabinovich et al. 2018). The 1964 tsunami was the strongest tsunami ever observed on the Pacific coast of Canada. All tsunamis detected on the BC coast before or after this event, including the 1960 Chile and 2011 Tohoku tsunamis, were considerably smaller (Fig. 2). The tsunami struck the coast of British Columbia near the time of high tide (Figs. 17 and 18). The cumulative effect of the high tide (which was from 1.5 m on the south coast to almost 3.5 m on the north coast) and the tsunami was one of the main reasons for the devastating impact of this event on the BC coast. The earliest recorded arrival was at 05:32 UTC at Tasu Sound on the west coast of Moresby Island, the Queen Charlotte IslandsFootnote 4 (Fig. 16), the tsunami wave arrived at this station only 1 h 56 min after the main shock (Table 5). The highest wave reported in Canada was in Shields Bay on the west coast of Graham Island (Fig. 16), where the crest wave was ~5.2 m above the spring high water (9.8 m above tidal datum). A logging camp in this remote area of the coast was damaged by the wave. The most severe damage from the event occurred at the twin towns of Alberni and Port Alberni, with maximum tsunami height at Port Alberni reaching almost 8 m (White 1966; Stephenson et al. 2007). The waves at these two towns, located at the head of long and narrow Alberni Inlet, were strongly amplified as they moved up Barkley Sound and into the inlet. The wave heights at Port Alberni were approximately 2.5 times higher than at Tofino and Ucluelet on the open coast of Vancouver Island (Fig. 16) (Clague 2001; Stephenson et al. 2007). As originally assumed by Murty and Boilard (1970), and confirmed by Henry and Murty (1995) and Fine et al. (2008), strong amplification of tsunami waves in Alberni Inlet was caused by resonant matching of the dominant period of incoming waves with the fundamental (Helmholtz) eigen period (~110 min) of this inlet. The major destruction that occurred at Alberni and Port Alberni has been described in a number of reports, papers and books, including WW64, White (1966), Clague (2001) and Stephenson et al. (2007).

Fig. 16
figure 16

Location of tide gauges on the coast of British Columbia that recorded the 1964 Alaska (Prince William Sound) tsunami and certain villages affected by tsunami waves; “HSC” = Hot Springs Cove. The inset shows “inner” stations situated in Juan de Fuca Strait (Victoria), the Strait of Georgia (Fulford Harbour) and eight stations in the area of metro Vancouver. The Queen Charlotte Islands are now officially known as "Haida Gwaii"

Full size image
Table 5 Parameters of the Alaska (Prince William Sound) tsunami of 28 March 1964 recorded by CHS analogue tide gauges on the coast of British Columbia (Main shock, Mw = 9.2 at 03:36 UTC)
Full size table

Tsunami waves also created severe damage at Prince Rupert. The details of the tsunami effects in Prince Rupert are described by local newspapers, in particular by the Prince Rupert Daily News (1964). The vessel Yaloa sank at her Digby Island moorings, 8–9 million feet of logs were cut loose and the float and pierhead at Metlakatla were torn out. Also, there was extensive breakwater damage at Fairway Bay and the submarine cable providing telephone service and the main power supply to Digby Island was put out of action, causing Digby Island airport and homes to go on emergency power (Stephenson et al. 2007; Rabinovich et al. 2018).

In addition to Prince Rupert on the mainland, several sites along the outer coast of Vancouver Island were also affected by the 1964 tsunami (Fig. 16). Tsunami waves sank a fishing boat and damaged log booms in Ucluelet, damaged wharf facilities and the municipal water pipeline in Tofino, destroyed houses in a village in Hot Springs Cove and swept buildings off their foundations in Zeballos. The waves did considerable damage to buildings and a radio-telephone communication system in the small logging community of Amai on the northwest coast of Vancouver Island (Stephenson et al. 2007).

A remarkable feature of the 1964 tsunami on the coast of British Columbia was that all sea level records located by WW64, contained the tsunami signal. According to WW64, the tsunami was recorded by twelve CHS tide gauges on the BC coast, consisting of nine permanent and three temporary gauges (Fig. 16). Six of the permanent stations were the same as the ones that recorded the 1960 Chile tsunami, namely, Prince Rupert, Alert Bay, Tofino, Victoria, Fulford Harbour and Point Atkinson (formerly, Caulfeild). The three additional permanent stations were Bella Bella on the mainland coast of central British Columbia, Port Alberni at the head of Alberni Inlet on the Pacific coast of Vancouver Island and Vancouver in the Strait of Georgia. The three temporary stations were Tasu Sound on the outer coast of Moresby Island, Ballenas Islands in the Strait of Georgia, and Port Moody located on the southeastern side of Burrard Inlet (Fig. 16). Tasu and Ballenas Islands were instrumented with pressure gauges; all the other stations had standard float-actuated gauges deployed in stilling wells (WW64).

An additional 1964 tsunami record was provided by Crown Zellerbabach Ltd. for Ocean Falls on the central BC mainland coast (Fig. 16). This gauge was located in the power house about 300 m from the river mouth. As low tides were not measured at the site, parts of the record were chopped off (Fig. 17). WW64 was also able to obtain five river gauge records from the Fraser River delta region provided by the Water Resources Branch, Department of Northern Affairs and National Resources: North Arm, Steveston, New Westminster, Deas Island and Pitt Lake (see the insert in Fig. 16).

Fig. 17
figure 17

The 28 March 1964 Alaska tsunami, including tides, as recorded at six stations located on the northern mainland coast of British Columbia and on the outer coast of Vancouver Island. For display purposes, the records have been vertically shifted relative to each other. The solid vertical red line labelled “E” denotes the time of the earthquake

Full size image

Taken together, WW64 collected and inspected 18 records of the 1964 Alaska tsunami on the coast of British Columbia. They reviewed, slightly revised and enlarged these records and included all of them into their report, except three records: Ballenas Islands, Port Moody and Deas Island “since they differ only slightly from Point Atkinson, Vancouver and Steveston, respectively”. In the analyses that follow, we used not the original records (which were unavailable) but the 15 records presented in the WW64 report. These records were digitised and examined in the same manner as the records for the four other great events. Figures 17 and 18 show 14 outer and inner coast records with tides: tsunamis superimposed on high tides significantly strengthen the destructive effect, promoting extensive inundation of lowland areas, in particular, at the head of Alberni Inlet and the adjacent delta of the Somass River. Also, Fig. 17 reveals the low-tide gaps in the Ocean Falls record that required careful processing.

Fig. 18
figure 18

The same as in Fig. 17 but for eight stations located in Juan de Fuca Strait (Victoria), in the Strait of Georgia (Fulford Harbour and Point Atkinson) and in the area of Vancouver and the Fraser River delta. The solid vertical red line labelled “E” denotes the time of the earthquake

Full size image

WW64 indicated that all records had been adjusted to Pacific Standard Time (PST = GMT − 8 h); however, an accurate prediction of the tides for the Ocean Falls record shows that the time was in PST +1 h (i.e., in GMT—7 h). This means that all previous timing estimates for this station (WW64; Stephenson et al. 2007) had an error of one hour. Tsunami oscillations are evident and prominent at all stations on the outer coast (Fig. 17), as well as at Victoria and Fulford Harbour (Fig. 18). However, at six stations located near Vancouver and the Fraser River delta, the tsunami signal is more difficult to distinguish because the oscillations are relatively weak, have quite long periods and are strongly masked by tides (Fig. 18).

The one tsunami record not presented in Figs. 17 and 18 is that for Port Alberni. This record (Fig. 19), which had the highest tsunami waves, ever instrumentally measured tsunami waves for the coast of Canada, needs a more detailed discussion. The first tsunami wave arrived at Port Alberni at 07:50 UTC on 28 March (4 h 14 min after the main shock) and damaged the tide gauge, putting it out of use for about 1.5 h (WW64). The gauge then recorded a portion of the second wave between 09:17 and 09:30 UTC and a small portion of the third wave before stopping for several hours. According to WW64, the gauge returned to normal operation at 16:08 UTC (Fig. 19). The first three waves propagated up the Somass River and were recorded by the Water Resources Branch water gauge located ~ 5.5 km up-river from Alberni. Although WW64 did not provide a copy of the water gauge record, they indicated the arrival times and heights of the first three wave crests:

Fig. 19
figure 19

The 1964 Alaska tsunami as recorded at Port Alberni, British Columbia (modified from Fine et al. 2018a, b). a A restored copy of the Port Alberni record (from Wigen and White 1964). The predicted tide is shown by the red line. The dashed lines indicate the record segments restored by Wigen and White (1964) based on land marks and the Somass River water gauge observations. The tide gauge stopped working on March 28 after 12:00 UTC, but resumed operation four hours later. The highest recorded sea level, about 4 m above high tide at 10:00 UTC, was due to the second incoming tsunami wave, while the corresponding maximum (peak-to-trough) wave height in Port Alberni was substantially higher than 6 m and occurred during the third, or later wave. The actual sequence numbers of the observed waves and maximum wave height of the largest wave are uncertain because of the 4-h gap in the data. However, based on beach marks, the maximum height was roughly 7.7 m. Oscillations in Alberni Inlet persisted for at least 48 h, with wave periods from 1.7 to 2 h. b The spectrum of the tsunami record at Port Alberni shown in a for the time period between 16:00 of 28 March and 18:00 UTC of 29 March. Periods (in min) of the two main spectral peaks are indicated

Full size image

  • 1st wave: 24 cm at ~10:16 UTC;

  • 2nd wave: 80 cm at ~11:53 UTC;

  • 3rd wave: 30 cm at ~13:30 UTC.

The time differences between the 1st, 2nd and 3rd wave crests were the same, 97 min. WW64 assumed that the same height and time intervals took place at Port Alberni as at the Somass River gauge. In retrospect, we suspect that more precise estimates could have been obtained from the de-tided records, but this had not been done at the time and now the Somass River record of the 1964 tsunami is unavailable. Therefore, all existing estimates of the 1964 Alaska tsunami parameters from this region are derived from the WW64 report, which, in addition to the Somass River gauge, used water marks on Port Alberni buildings and eyewitness evidence at the time of the event.

As indicated by the partly restored Port Alberni tide gauge record (Fig. 19), the maximum absolute water level (tsunami + high tide) was associated with the second tsunami wave crest that reached about 27 feet (8.2 m) above chart datum (Fig. 19). However, the highest trough-to-crest height of the tsunami wave alone (without inclusion of the tide) of ~7.7 m was due to the 3rd wave. Fine et al. (2008) digitised the continuous part of the Port Alberni record (after 16:08 UTC) and, from spectral analysis of this record, found a prominent spectral peak with a period of 110 min.Footnote 5 The computed period of the fundamental (Helmholtz) mode of Alberni Inlet was ~112 min, in close agreement with the observations. These results confirm the resonant nature of the 1964 tsunami oscillations in Alberni Inlet proposed earlier by Murty and Boilard (1970) and Henry and Murty (1995), who tried to explain the strong amplification of tsunami waves at Port Alberni.

The fourteen tide gauge records of the 1964 tsunami, shown in Figs. 17 and 18, were de-tided (Figs. 20 and 21) and used to estimate the principal tsunami parameters (Table 5). The 1964 Alaska tsunami was a unique event in that the maximum recorded trough-to-crest wave heights at eight stations were >1 m and at five stations >2 m (Table 5). According to the results in WW64 and those from the numerical modelling of Fine et al. (2018a, b), wave heights along the entire outer coast of British Columbia were close to, or higher, than 2 m. As indicated above, tsunami waves first arrived at Tasu Sound and then to the other five “northern” stations (Fig. 20). According to our estimates, the frontal wave of the 1964 tsunami arrived at Prince Rupert at 06:30 UTC, 2 h and 54 min after the main shock (previous estimates by WW64 showed an arrival time of 06:52). This wave had an amplitude of 62 cm (43 cm according to WW64); the second wave was the highest (Fig. 20), with a wave height of 271 cm (the same as was estimated by WW64). These estimates are in good agreement with the numerical results of Fine et al. (2018a). The dominant period of the observed waves at Prince Rupert was about 110 min, which is approximately the same as during the 1952, 1957 and 1960 tsunamis.

Fig. 20
figure 20

De-tided but unfiltered records of the 1964 tsunami from six “northern” stations on the coast of British Columbia that recorded the 1964 Alaska (Prince William) tsunami. The solid vertical red line labelled “E” denotes the time of the earthquake; the arrows indicate the tsunami arrival

Full size image
Fig. 21
figure 21

The same as in Fig. 20, but for the “southern” stations located in a Juan de Fuca Strait (Victoria) and the Strait of Georgia (Fulford Harbour); and b six stations in the area of Vancouver and the Fraser River delta

Full size image

As it rounded the Queen Charlotte Islands, the 1964 tsunami entered Queen Charlotte Sound, arriving at Ocean Falls at 6:46 UTC and then, 3 min later, at Bella Bella. These two stations are located on the central mainland BC coast, relatively close to each other (Fig. 16). The shape and general character of the first few oscillations (which were the largest), look similar at these stations (Fig. 20), however the maximum recorded tsunami wave height at Ocean Falls (3.76 m) was almost two times larger than at Bella Bella (1.90 m). The difference demonstrates the strong influence of local topography on the observed tsunami heights. After entering Queen Charlotte Sound, the tsunami travelled southward along the northern coast of Vancouver Island and at 07:23 UTC (3 h 47 min after the earthquake) reached Alert Bay, where it produced a maximum wave height of 2.22 m (Table 5).

The main branch of the 1964 Alaska tsunami propagated along the west coast of Vancouver Island, arriving at Tofino at 7:01 UTC; 49 min later, the tsunami inundated Alberni and Port Alberni. Had the Tsunami Warning System been in operation in 1964, this time difference could have provided an effective early warning. The maximum tsunami wave height of 7.7 m at Port Alberni was ~ 3.2 times higher than the corresponding height of 2.37 m at Tofino, demonstrating once again the strong tsunami amplification from resonance within Alberni Inlet. It is also important to note that at most stations along the outer coast—Tasu Sound, Ocean Falls, Bella Bella (Fig. 20), and Victoria (Fig. 21a)—the first wave was the strongest, whereas at Port Alberni, it was the third wave: Due to the resonance, the fundamental inlet mode consequently accumulated energy and amplified.

As it propagated southward along the outer coast of Vancouver Island, part of the energy of the 1964 tsunami entered Juan de Fuca Strait. The first wave reached Victoria at 8:08 UTC and then entered the Strait of Georgia, arriving at Fulford Harbour at 8:35 UTC. As the tsunami propagated from the ocean inward along the straits, it began to gradually attenuate, with maximum wave heights falling from 2.37 m at Tofino, to 1.47 m at Victoria and 0.83 m at Fulford Harbour (Table 5, Fig. 21a). In Victoria Harbour, the leading wave was the highest. The wave arrived as a pronounced wave crest of 70 cm, followed by an abrupt trough of 77 cm; the period of this wave was about 100 min. Although the same wave is evident in the Fulford Harbour record, the highest wave was the third wave, not the first, apparently as a result of local harbour resonant effects. In contrast to other major tsunamis (1946, 1952, 1957, 1960, 2010 and 2011), the 1964 tsunami at Victoria and Fulford, as well as at other stations of the British Columbia coast (Figs. 20 and 21), decayed relatively quickly (Rabinovich et al. 2018). The reason for this fast decay appears to be the closeness of the 1964 source region to the observation area. As shown by Rabinovich et al. (2011) for the global 2004 Sumatra tsunami, near-field tsunami oscillations decay much faster than more distant far-field oscillations. The prominent feature of this tsunami is the close similarity of the first 2–3 oscillations along the entire outer coast (Fig. 20) and at Victoria and Fulford Harbour (Fig. 21a). The observed tsunami oscillations at Victoria were in good agreement with the results of a high-resolution numerical model of the 1964 tsunami for Victoria Harbour, constructed by Fine et al. (2018b).

After it entered the Strait of Georgia, the 1964 tsunami reached Point Atkinson (formerly, Caulfeild) at 9:14 UTC on 28 March and then propagated into the numerous narrow inlets on the mainland coast and into the Fraser River (see the inset in Fig. 16 for station locations). Although the tsunami oscillations are barely recognisable in the original tidal records for these areas (Fig. 18), the waves are clearly evident in the residuals (de-tided) (Fig. 21b). The character of these oscillations is significantly different from those observed in other regions of the BC coast in that the wave oscillations at all six stations were very similar and much more regular than at other stations. Short-period oscillations were not observed at these stations, while long-period oscillations were very consistent among the records, with a predominant period T ~ 2 h. The maximum tsunami wave heights at all stations (ignoring Pitt Lake) ranged from 22 cm at Vancouver to 50 cm at the North Arm in the Fraser River (Table 5).

A unique feature of the 1964 event was tsunami oscillations recorded at Pitt Lake. Pitt Lake has a tidal range of ~1 m and is one of the few tidal lakes on the BC coast. The water gauge is located at the southern tip of the lake, about 18 km from the confluence of the Pitt and Fraser rivers, which, in turn, is about 46 km from the Strait of Georgia. To reach the Pitt Lake gauge from the strait, the 1964 tsunami had to propagate ~ 64 km upstream. The first wave arrived at the lake at 11:49 UTC, 2 h 35 min later than at Point Atkinson and 1 h 20 min later than at New Westminster (Table 5; the station locations are shown in the inset of Fig. 16). Despite the small wave height of 9 cm, the tsunami is clearly detectable in the Pitt Lake record (Fig. 21b).

The f–t diagrams in Figs. 22 and 23 provide details of the wave structure and time evolution of the 14 records of the 1964 tsunami shown in Figs. 20 and 21. These diagrams are substantially different from those of the four other tsunamis discussed above: As indicated in f–t diagrams in Figs. 4, 6, 8, 13 and 14, there is considerable individuality among the various stations but a strong similarity at a given station for the different events. In contrast, the 1964 f–t diagrams for all BC stations were alike: the tsunami energy was mainly at low frequencies of ~ 0.4–0.8 cph (periods of 2.5–1.2 h) and decayed abruptly within the first 5–6 h. Higher-frequency, shorter duration oscillations with f ~ 2 cph (T ~ 30 min) are evident at Ocean Falls, Bella Bella, Alert Bay, Tofino, Victoria and Fulford Harbour. At the two northern stations (Prince Rupert and Tasu Sound), as well as at the six stations located in the Fraser River delta region, high-frequency oscillations were totally absent, with the dominant wave period typically around 2 h. Based on these observations, it appears that, in contrast to previous events, the properties of the 1964 oscillations were strongly determined by the source, while for the other events, the influence of the regional/local topography prevailed. We believe that this is because the 1964 tsunami source was located much closer to the BC coast than the 1946, 1952, 1957 or 1960 sources.

Fig. 22
figure 22

Frequency-time plots (f–t diagrams) for the 1964 Alaska tsunami recorded at six “northern” stations on the coast of British Columbia. The solid vertical red line labelled “E” denotes the time of the earthquake; the dashed white line indicates the tsunami arrival

Full size image
Fig. 23
figure 23

The same as in Fig. 22, but for eight stations located in Juan de Fuca Strait, the Strait of Georgia and the Fraser River delta region (see inset in Fig. 16 for station locations)

Full size image

Although the characteristics of the source region have a marked effect on the incoming waves from the 1964 tsunami on the BC coast, the local topography also plays a major role in determining their properties. The extreme waves and pronounced damage in Alberni/Port Alberni were caused by the close match between the dominant period of the arriving waves (~ 2 h according to the f–t diagrams shown in Figs. 22 and 23) and the fundamental period of Alberni Inlet. Witness reports and water gauge measurements in several rivers (Wigen and White 1964; White 1966) indicate significant amplification of tsunami waves entering other inlets and fjords of the BC coast (Fig. 24). Also, certain straits and channels act as a low-pass filter that effectively suppresses high-frequency (HF) oscillations, while allowing low-frequency (LF) waves to pass through. In particular, Dixon Entrance filters out waves with periods shorter than 1–1.5 h, which is why only long-period waves were recorded at this station during the 1964 event, as well as during the 1952, 1957 and 1960 tsunamis. It appears that Tasu Sound plays a similar role by preventing penetration of the HF “tail” of the 1964 tsunami (Fig. 22). Another region that is “blocked” from HF waves is metro Vancouver and the Fraser River delta. This blockage is the reason why the six f-t diagrams for this region (Fig. 23) look almost identical, with dominance of long-period motions and the absence of HF oscillations (periods shorter than 1 h).

Fig. 24
figure 24

Map of Vancouver Island showing maximum tsunami runup along the coast. The white circles denote tide gauges that were working during the 1964 Alaska event. The red bars indicate maximum wave heights estimated from the tide gauge data (Table 5); the orange bars are runup heights collected by Wilson and Tørum (1972) from Wigen and White (1964), White (1966), witness reports and in-river gauges

Full size image

Wilson and Tørum (1972) summarised available tide gauge/river water gauge measurements and witness reports for the 1964 Alaska tsunami along the coast of Vancouver Island. We re-examined this information and expanded the findings based on our analysis of digitised versions of all existing tsunami records. This enabled us to construct a map of observed runup heights for the island (Fig. 24). In addition to providing a general overview of the 1964 tsunami waves on the coast of Vancouver Island, we are able to show that large wave heights of more than 3–5 m were typical along the entire oceanic coast of the island. These strong oscillations were mainly associated with the resonant interaction of the incoming tsunami waves with local topographic features of the region. The results provide emergency response managers with insight into runup heights to be expected from future tsunamis originating from a Mw ~9.2 Alaska subduction zone failure in the northeast Pacific.

4 Discussion

One important question that arises from our analysis of the coastal tide gauge records of the five great tsunamis of the 20th century is: How can we use the information from one “reference” station for real-time tsunami warning and forecasting of tsunami heights and arrival times at other sites? Our analysis of these five events enables us to examine this problem for three selected stations: Tofino, Victoria and Port Alberni.

The Tsunami Warning Station at Tofino is the most important tsunami monitoring station on the west coast of Canada and one of the most important on the Pacific coast of North America. The station has been in operation for 114 years and has detected 75 tsunamis (Fig. 2), forming one of the most complete compilations of instrumentally recorded tsunamis in the World Ocean. For the Canadian Tsunami Warning System, Tofino can be considered the “bell-weather station”. Thus, we can use this station as the “reference (warning) station” and correlate tsunami parameters at TWS Tofino with those measured at Victoria and Port Alberni.

To complete this study, we have used additional data from two recent great tsunamis: The tsunamis of 27 February 2010 Chile (Maule) and 11 March 2011 Tohoku (East Japan). Both tsunamis were recorded along the entire coast of British Columbia (cf. Rabinovich et al. 2013, 2018) and are the strongest tsunamis ever recorded by high-precision digital tide gauges on the coast of Canada. Estimated parameters of these two tsunamis for Tofino, Port Alberni and Victoria are presented in Table 6. The epicentres of all seven great tsunamis are indicated in Fig. 1.

Table 6 Parameters of the Chile (Maule) tsunami of 27 February 2010 and the Tohoku (East Japan) tsunami of 11 March 2011 at three CHS stations on the coast of Vancouver Island
Full size table

All seven great tsunamis considered in this study were recorded at both Tofino and Victoria. Therefore, we can use the resulting parameters to estimate empirical relationships between observed maximum tsunami wave heights and arrival times for the two stations. As shown in Fig. 25, the mean ratio of maximum tsunami wave heights at Victoria versus those at Tofino, HVict/HTof ≈ 0.57 ± 0.03 (N = 7), where N is the number of the events, is well constrained. Similarly, the mean square correlation coefficient, R2 = 0.89, among events is very high (Fig. 25a). As a consequence, the maximum tsunami wave height expected at Victoria will almost invariably be roughly half that for Tofino. The situation with arrival times is more complicated. In this case, the time differences (specifically, the lead time at Tofino relative to Victoria) depends on the source location. The sources of the seven great tsunamis were located in three distinct regions: (1) Aleutian-Alaska; (2) Kamchatka-Japan; and (3) Chile (Fig. 1). Relative to the coast of Vancouver Island, these source regions were located to the (1) northwest (2) west and (3) south. Accordingly, tsunami waves arrived at Tofino and Victoria from three different directions. In Fig. 25b, the three regions are indicated by source-region specific symbols: Maximum time differences arise when the tsunami waves arrive from the northwest, while minimum time differences occur when the waves arrive from the south. The mean values for the three source regions are: (1) 69 min; (2) 62 min; and (3) 34 min. These estimates show that TWS Tofino can be effectively used to forecast tsunami waves at Victoria. For tsunami sources located near the coast of South America, the lead time will be a little more than 30 min, while for sources located near Alaska, the Aleutian Islands, the Kamchatka, Kuril Islands and Japan, it will be ~60–70 min.

Fig. 25
figure 25

source regions: Aleutian-Alaska (1946, 1957 and 1964); Kamchatka-Japan (1952 and 2011); and Chile (1960 and 2011)

a An empirical relationship between the maximum tsunami wave heights at Victoria vs Tofino for the seven great events (1946, 1952, 1957, 1960, 1964, 2010 and 2011). b The differences in tsunami arrival times (“lead time”) between Tofino and Victoria for the three groups of

Full size image

An attempt to generate similar statistics using tsunami values for Port Alberni was only marginally successful. There were only three great tsunamis recorded at Port Alberni (1964, 2010 and 2011) and these led to markedly different height ratios, HPA/HTof, for Port Alberni versus Tofino. Specifically, 3.2 (1964), 2.6 (2010) and 1.1 (2011), making it impossible to provide definite conclusions regarding the expected tsunami heights at Port Alberni based solely on the information from Tofino. However, the estimates of arrival time differences for the three events are much steadier: 49, 34 and 42 min, respectively. This means that for “southern sources” (Chile), the lead time of Tofino over Port Alberni is approximately the same as for Victoria, while for “northwestern” and “western” sources, it is ~ 40–50 min.

5 Conclusions

The five great earthquakes examined in this study—1946 Aleutian Islands (Mw 8.6), 1952 Kamchatka (Mw 9.0), 1957 Andreanof Islands (Mw 8.6), 1960 Chile (Mw 9.5), and 1964 Alaska (Mw 9.2) earthquakes—generated destructive trans-Pacific tsunamis that were recorded by analogue tide gauges along the coast of British Columbia. Analysis of these analogue records is important because the five tsunamis, together with the recent 2010 Chile and 2011 Tohoku tsunamis, were the strongest events ever instrumentally recorded on the west coast of Canada. Moreover, one them—the 1964 Alaska tsunami—was the most destructive tsunami in Canadian history, causing about $10 million in damage (in 1964 dollars). If such a tsunami were to occur today, the cost of the damage would be much greater. The maximum tsunami wave heights ever recorded on the BC coast are associated with this particular event.

Altogether, we found and processed 2 records for 1946, 5 for 1952, 6 for 1957, 17 for 1960 and 15 for 1964. All five tsunamis were recorded at Tofino and Victoria, and four of the tsunamis (except 1946) were also measured at Prince Rupert and Alert Bay. The results of these analyses are essential for addressing three major problems:

  1. (1)

    Cataloguing of tsunami events.

    Based on the new data and tsunami wave estimates, we were able to correct earlier information on the five events for the BC coast presented in the catalogue of Stephenson et al. (2007).

  2. (2)

    Estimation of the potential tsunami risk to the BC coast.

    Precise values of maximum tsunami heights and amplitudes are critical for long-term tsunami forecasting. This information has already been used to evaluate possible tsunami wave heights at Prince Rupert and Victoria (Rabinovich et al. 2018) and can be used for other sites on the BC coast. These estimates also help identify the probable sources of tsunami waves that are a major threat for this coast. Based on our analysis of the five tsunamis, and taking into account the 2010 and 2011 events, we conclude that the Alaska region, and major earthquakes in this region similar to the 1964 event, have highest potential as a source region for major tsunamis on the coast of British Columbia.

  3. (3)

    Verification and calibration of numerical tsunami models.

    The research has enabled us to provide model verification data for reliable modelling of maximum tsunami wave heights and associated currents for specific sites along the coast of British Columbia. In particular, the 1964 tsunami records at Prince Rupert, Victoria and Fulford Harbour were used to verify the 1964 numerical model of Fine et al. (2018a, b).

Each of the five examined great tsunami events has its own individuality and specific properties. At the same time, the waves have many similar features. The five tsunamis had different amplitudes on the BC coast and produced quite different wave heights. However, the general character of the recorded oscillations, their dominant frequencies and evolution in time are mostly alike. It is well known that the main spectral peaks for different tsunami events at the same site are similar, while the spectral peaks from the same tsunami event, but at different sites, are substantially different (cf. Honda et al. 1908; Rabinovich 1997; see also Fig. 8 in Zaytsev et al. 2017). The results of the present study demonstrate that not only spectra, but also f–t diagrams (wavelets), for the same site are similar for different events. This means that tsunami waves for different events not only oscillate at the same frequencies (likely as a result of frequency-specific resonance) but also evolve and decay in comparable ways. It is evident that this similarity is related to the local/regional topography of the sites and that, for far-field tsunamis (like those in 1946, 1952, 1957 and 1960), the topographic admittance function (frequency response) determines the character of the observed waves. Rabinovich and Stephenson (2004) used long time series of background oscillations to estimate the frequency response of the primary BC tide gauge stations. Not unexpectedly, the peaks of these functions for the respective sites match the high-energy frequency bands at the corresponding f–t plots for these four tsunamis. One of the important properties of the four earliest (1946–1960) tsunamis examined in this study, is the long ringing duration of 3–5 days and the slow energy decay. For some sites, it was quite common to find maximum tsunami wave heights occurring several hours after the first wave arrival.

The 1964 Alaska tsunami for the BC coast was distinct in several ways. Specifically, this was the strongest tsunami ever recorded on this coast, with waves at eight stations reaching maximum trough-to-crest wave heights that were higher than 1 m. This included Port Alberni (7.7 m), Ocean Falls (3.76 m), Prince Rupert (2.71 m), Tofino (2.37 m) and Alert Bay (2.22 m). At all of these stations, the maximum wave occurred within the first three waves. Wave heights then rapidly attenuated. In comparison with other events, the “ringing” of the 1964 tsunami was much shorter (< 1.5 days) and the energy decay much faster. While the wave spectra for four of the other tsunamis differed among stations, with each having its own spectral “individuality” (as evident in their f–t diagrams), corresponding diagrams for the 1964 tsunami records were quite similar. It appears that the differences in the f–t diagrams for individual stations for the earlier (1946–1960) tsunamis are determined by the differences in topographic response, while the similarity of the 1964 diagrams is determined by the source characteristics. The distinct properties of the 1964 tsunami waves are explained by the proximity of the 1964 source area to the BC coast. For this coast, the 1964 tsunami can be considered as a “near-field” event; at stations in the Pacific Ocean located at greater distances from the source, the ringing was much longer, the energy decay much slower and the spectral structure markedly different for different stations (cf. Spaeth and Berkman 1967).

An intriguing feature of the five tsunamis is their deep penetration inside the narrow channels, fjords and inlets of the BC coast. For example, the 1952 Kamchatka tsunami was recorded at Kitimat at the head of Kitimat Arm, which itself is a continuation of the long and narrow Douglas Channel. The wave propagated ~ 80 km from Hecate Strait to the Kitimat tide gauge. The 1957 Andreanof Islands tsunami was measured at Bella Coola, located at the head of North Bentinck Arm, the remotest part of Burke Channel, a distance of ~ 110 from the open ocean (Queen Charlotte Sound). The 1960 Chile tsunami was recorded at 17 stations on the BC coast; five of these stations were located well within the Seymour—Belize Inlet System on the central mainland coast of British Columbia. To reach these stations, the tsunami had to go through Nakawakto Rapids, a narrow constriction having the fastest tidal currents in the world ocean (up to 17 knots). The 1960 tsunami also propagated over 100 km through Juan de Fuca Strait and a series of narrow channels in the Gulf Islands and San Juan Islands into the Strait of Georgia.

The 1964 Alaska tsunami was recorded at 18 tide gauge stations. More than half of these stations were located in the Strait of Georgia and the Fraser River. The tsunami waves propagated far upstream in the river and were recorded at Pitt Lake, which is connected with the Strait of Georgia by a 64-km route through the Fraser and Pitt rivers. It took about 2.5 h for the tsunami wave to travel this route, all the time the attenuating waves maintaining their shape and spectral structure. The recorded waves at this station, as well as at other stations in the region of metro Vancouver and the Fraser River delta, had an almost monochromatic character with one strongly dominant period of about two hours.

With respect to tsunami wave penetration into fjords and inlets, we emphasise that, although the channels hinder propagation of the waves, they can also lead to strong resonant amplification when the dominant frequency of the incoming waves coincides with the eigen (natural) frequency of the interconnected basins. The most noteworthy example of this kind for the BC coast, is the resonant amplification of the 1964 tsunami waves in Alberni Inlet, which resulted in severe damage to the towns of Alberni and Port Alberni.