Global Historical Megatsunamis Catalog (GHMCat)

1. Introduction

The comprehensive megatsunami global catalog here presented describes for the first time all known historical megatsunamis that have been documented by historical records or geological evidence (i.e. measurements and field observations).

Some previous studies present lists of "large tsunamis” compiled for specific areas or time intervals [1], and partial lists of landslide-generated "giant waves" or “large waves” have also been published [2,3,4,5,6].

Historical tsunamis are those documented through written records from eyewitnesses or direct observations and measurements. The number of megatsunamis known from historical documentary references or deduced from geological evidence is very limited, the oldest of which can be said to be that caused by the great explosive eruption on the island of Thera or Santorini in Greece around 3600 BP.

Undoubtedly, many megatsunamis must have occurred on Earth in historical times, far beyond the few records available, but they were neither observed nor reported. Also, throughout the earth’s history these processes must have been relatively frequent, although both because of the short period of historical record and the fact that most of their effects have disappeared from the face of the earth, we can only access a few records.

Megatsunamis, or giant waves, are extreme unexpected processes that strike the earth’s cliff shores in any geological or climatic region, as they are not related to any specific conditioning factor, but to the presence of a water body and the occurrence of a geological triggering-process, most often a large subaerial landslide. The term, even without a clear, agreed definition, has been widely used in scientific publications in the last decades, mainly in relation to prehistoric oceanic tsunamis associated with large bolide impacts and giant volcanic island flank collapses [7,8,9,10,11]. It has also been applied to the extensive and destructive seismic tsunamis of 2004 Indonesia and 2011 Japan, and earlier events such as the 1946 Aleutians, 1960 Chile and 1964 Alaska oceanic tsunamis [1].

However, the most agreed and appropriate definition of a megatsunami refers only to the maximum height reached by a wave on land ‒or runup‒, the extent and impact of the tsunami being irrelevant. On this basis, a megatsunami can be defined as a giant wave, with a height of 40 m or more, as established in this study.

Megatsunamis are always local events, originating from a nearby source, which effects are confined to a small area, as the height of the waves rapidly decreases as they move away from their source. Opposite to “conventional” or “normal” tsunamis resulting from offshore earthquakes, megatsunamis reach very large heights when hitting the coast, with maximum runups observed or measured of several hundred meters on the shores of confined marine or inland bodies of water, such as lakes or reservoirs.

Most of historical megatsunamis have been generated by large subaerial landslides, and only a few cases are related to violent volcanic eruptions. Landslide-generated tsunamis are the largest due to the kinetic energy of huge rock masses falling from considerable heights and violently entering bays or lakes; landslides are sometimes triggered by large earthquakes, and often occur in confined bodies of water, bays or lakes. Megatsunamis caused by subaerial landslides have been recorded since the 18th century in Norway and Japan.

Among the problems associated with historical information, are those derived from the inaccuracy and incompleteness of the data mainly about old or geographically remote events, and errors in the interpretation of old descriptions and conversion of old units of measurement for wave height.

From 1674 to 2023, a total of 37 megatsunamis (i.e. those with wave heights > 40 m) have been included in the Global Historical Megatsunami Catalog (GHMCat), after a thorough investigation of the available documentation −more than 280 publications between 1888 and 2023 have been consulted−. For some of these, extreme heights in excess of 100 m have been reported.

Although only a few records from the 19th century have been properly characterized, during the 20th century the number of recorded events has increased significantly, and in the first 20 years of the 21st century alone the same number of cases has been recorded as for the 20th century. It is clear that the record of large tsunamis increases as scientific knowledge develops along with industrial development from the second half of the 19th century and especially in the 20th century.

To the list of cataloged events, a new undocumented megatsunami has been added from the review of old Alaskan explorers’ documents from the late 18th century: a megatsunami prior to 1786 in Lituya Bay [12]. However, it has not been included in the GHMCat as its year of occurrence is unknown, and we can only establish its upper limit.

The GHMCat presented here brings together all the historical events recorded so far, and provides for the first time the whole list of events and information on the causes and significant effects, damages or casualties, when available.

Data on maximum recorded wave heights, their causes and effects are important for understanding the frequency and distribution of the phenomenon, its generation and potential risks and harmful effects, and can contribute to prevention and mitigations tasks.

2. What is a megatsunami?

2.1. Definition and background

A megatsunami is a giant wave. Although there is no formal definition of megatsunami, there is a general consensus based on the maximum height reached by the wave when it hits the coast, as this seems to be the more useful and used parameter to differentiate between megatsunami and “normal” (seismic) tsunami waves. The former can therefore be defined as a tsunami, or a wave, that reaches or exceeds a certain height when hitting the coast; however, there is no specific wave height or threshold above which the term "megatsunami" can be applied.

The term megatsunami was first used in scientific literature around 1990 [13] in connection with the giant waves (“mega-tsunami”) caused by the impact of a huge asteroid in the Gulf of Mexico millions of years ago, at the Cretaceous-Paleogene boundary, which left characteristic coarse-grained deposits throughout the region. Previously, Bourgeois et al. [14] referred to wave heights in the range of 50 to 100 m for this paleotsunami, using the term "very large tsunami".

Another early reference appears in Paskoff [15], in reference to a possible "mega-tsunami" to explain the presence of large boulders embedded in shelly beach deposits on an elevated marine terrace up to 40 m a.s.l. on the Chilean coast; tentatively, the cause of the tsunami was associated with an earthquake of exceptional magnitude during the Plio-Quaternary. Previously, the term "giant wave" was used to explain the origin of high marine coarse deposits on the island of Lanai, Hawaii [16,17]: according to their elevation, the waves must have reached between 190 and 375 m high, and probably were originated by a huge submarine landslide. The same term “giant wave” was used earlier [2] to describe the waves of a tsunami generated by a large rockslide in 1958 at Lituya Bay, Alaska.

With regard to the threshold height value reached by the waves to be classified as a megatsunami, few publications provide specific values. Some authors have subjectively provided minimum wave height values of 40 or 50 m a.s.l., or even 100 m; others, from several tens to hundreds of meters high.

Alexander and Neall [18] point out that megatsunamis are above 40 m, and up to hundreds of meters, while “normal” tsunamis have heights of up to about 10 m; according to Krehl [19], mega-tsunamis are defined in the literature as waves over 100 m high, even over 300 m, although the author does not include any bibliographical reference. These and other references are compiled by Goff et al. [20] in a text focused on the definition of the term "mega-tsunami", finally proposing a definition based solely on initial wave height/amplitude at source exceeding 100 m/50 m respectively; thus, only including those extreme events on geological timescales generated by large bolide impacts, extremely violent volcanic activity or giant landslides, and possibly an extreme submarine earthquake not yet historically documented.

In the same way, Naranjo et al. [21] define the term megatsunami as a tsunami with an initial wave amplitude or height of several tens or hundreds of meters, well above a "normal" tsunami; most have been caused by major impacts, such as large-scale landslides, devastating volcanic eruptions or meteorite strikes, while normal tsunamis are caused by tectonic activity that violently displaces the seabed.

The authors do not agree with those definitions based on the initial wave height at source, as a tsunami means a wave that reach or hit the coast, and inevitably the height it can reach is influenced by near-offshore bathymetry and onshore topography and geomorphology (e.g. cliff coasts versus flat coasts). Thus, the most objective parameter to define a megatsunami would be the maximum height reached by the waves at any point of the affected coast. The term most commonly used to refer to this parameter is vertical runup or just runup. The maximum horizontal distance on land (inundation or horizontal runup) is not considered in the definition of megatsunami. Both values depend largely on the elevation and orography of the coast.

According to UNESCO [22], and other publications [23], runup is the difference between the elevation of maximum tsunami penetration inland and the sea level, and can be only measured where there is clear evidence of the inundation limit on the shore. This definition is clearly aimed at “normal” seismic tsunamis on flat coasts, where the maximum wave height on land occurs at the maximum inundation distance. However, especially in the case of megatsunamis, there could be waves higher than those corresponding to the maximum inundation limit, i.e. when the wave hits a local rocky cliff, splashing and reaching an elevation higher than that at the point of maximum inundation distance.

In order to clarify the term and its meaning, the most useful definition for runup is the maximum height the wave reaches onshore above a reference level, usually mean sea level. Runup can be observed or measured after the tsunami. At present, rigorous measurements of the largest tsunami waves are carried out through post-tsunami field surveys, looking at the marks left by the waves on the coast. Maximum runup height can be inferred from the vertical limit of vegetation washed away by the wave, marks on wooded hillsides, damage to trees or debris deposited on hillsides or on trees.

According to Goff et al. [20], the term megatsunami has been arbitrarily used in a variety of settings, to mean not only large waves at the source of the event or on land, but also geographically extensive tsunamis (such as the 2004 and 2011 ocean-scale tsunamis) with or without giant waves. Gusiakov [1] refers to “catastrophic trans-oceanic mega-tsunamis affecting an entire oceanic basin…”, listing seven occurred within the last 77 years, resulting from large subduction earthquakes, with maximum runup heights between 23 m and 67 m (one of them caused by the largest earthquake in recorded history, the 1960 Chile earthquake, with maximum wave height between 12 and >20 m, depending on the source).

Both the geographical extent (even on an oceanic scale) and the impact or destruction caused by the waves in coastal areas are irrelevant to the definition of a megatsunami, based solely on the height of the waves on land. In fact, megatsunamis are always local events, originating from a nearby source, which effects are confined to a small area; maximum runups are focused along a limited length of coast −not exceeding some tens of kilometers at most, against the hundreds of kilometers that a tectonically-induced tsunami can reach [5]− and decreases rapidly with distance. Near-field runup can be many times higher than that produced by seismic tsunamis, and can reach hundreds of meters in exceptional circumstances.

Literature references including lists of megatsunamis or large tsunamis are very scarce. A compilation of “large tsunamis” collected from the two existing global historical databases (GHTDs) for the period 1900-2019 is presented by Gusiakov [1], based on the annual maximum runups observed or measured (Figure 1), of which 17 events exceed or equal 40 m in height; according to their sources most of them have been landslide-generated. A list of data on localized “giant waves” generated by subaerial landslides is presented by Miller [2], and partial lists have also been published [3,4,5,6].

2.2. Megatsunami wave height threshold

After a comprehensive review of the literature and available data on the largest historical tsunamis, a megatsunami is considered here when the waves reach or exceed 40 m in height on land with respect to the mean water level (sea, lake, river); the height may correspond to any type of recorded observation: maximum height reached by the wave on land (runup height), marks left by the violent impact of the wave, or others. This lower limit is considered here as the most representative, as it differentiates an elite group of about thirty cases in the GHDBs originating from geological processes other than earthquakes, thus maintaining the exceptionality to which the term itself implicitly refers. The 40 m threshold corresponds to the lowest of those proposed for megatsunamis by different authors. If a lower limit (i.e. 30 m) were considered, the group would include many other cases mostly related to earthquakes; approximately 1.5% of all historically tsunamis of any origin documented in the GHTDs are assigned a runup ≥30 m, and 1% ≥40 m, none of them directly caused by an earthquake. Some threshold above, e.g. 100 m, would only include an elite group of 6 or 7 events of those included in the GHTDs.

A remark should be made concerning what is considered the maximum height reached by the water: this data is always punctual, and is not representative of the medium, even the maximum, height of the waves reaching the coast, and can sometimes correspond to the surge or splashes of the waves or seiches that are generated by the movement of the water in confined bodies of water. E.g. for the 1958 Lituya Bay megatsunami, a maximum runup of 525 m was measured in situ at the highest point where the surge of water reached after hit a steep slope and splash to very high elevations, this being the value assigned to this megatsunami. However, some authors differentiate between this local runup and the maximum wave height reflected on the slopes more extensively, ~150 m, or the maximum wave amplitude, ~50 m [23].

Measurements on cliffs or steep coastlines may give high local runup values, very different from the maximum wave heights reached in the surrounding areas. In the case of the Indonesian tsunami in 2004, the maximum runup of 50-51 m measured west of Banda Aceh, northern Sumatra [24,25] corresponds to the height of the vegetation washed away by waves on the cliff walls of a small isthmus; this value is considered to be the highest ever measured for a tsunami directly attributed to an earthquake (actually for a non-landslide tsunami), although given its very local nature it cannot be considered representative of the tsunami; in fact, some authors [24,26] give maximum wave heights of about 35 m for the same area, the most affected by large waves, differentiating between maximum waves heights and maximum point runup height due to the particular orography of the coast.

2.3. Causes of megatsunamis

Megatsunamis are caused by the impact of large-scale landslides, gigantic explosive volcanic eruptions and large asteroids on the sea or confined bodies of water, while “normal” tsunamis ‒seismic sea waves‒ are caused by tectonic activity that violently displaces vertically the sea floor and the water column above. In contrast to large seismically-generated tsunamis, megatsunamis are local processes, as the height of the waves rapidly decreases as they move away from their source; according to Lander [23] “Large breaking waves … are rare and almost unheard of outside of the generating area.”

Because the probability of occurrence of these large-scale geological processes is low, that is also the probability of occurrence of megatsunamis around the world. Prehistoric volcanic-island-flank landslides, the largest ever to have occurred on Earth (with a volume of several hundred km³), originated megatsunamis up to several hundred meters high, according to geological evidence from deposits left by the waves on the coasts [7,8,9]. No volcanic island megalandslides or asteroid impact have occurred in historical times.

Most megatsunamis are caused by subaerial landslides and rock avalanches on the slopes of confined bays and fjords and, to a lesser extent, on lakes, reservoirs and rivers.

The exact origin of megastunamis associated with large earthquakes has been (, and remains in some cases, controversial. While for some researchers they are generated by large submarine landslides triggered by the earthquake, for others it is the rupture mechanism of the fault itself the origin of the giant waves; these megathrust earthquakes occur at convergent plate boundaries, and are called “slow tsunami earthquake” or “tsunami earthquake” because of their potential to generate large tsunamis. In the first case, evidence of landslides can only be obtained by marine geological and geophysical surveys, as has been the case for several historical megatsunamis initially attributed to earthquakes [23,27,28].

First evidence of a large earthquake-triggered underwater landslide was during the 1929 Grand Banks earthquake (M 7.2), North Atlantic, after numerous undersea telegraph cables were broken as a result of major mass movements downslope. In the more recent 1998 Papua New Guinea earthquake (M 7.1) a submarine landslide was inferred from a variety of evidence [29]. Both underwater mass movement generated waves of up to 13-15 m in height on the nearby coasts, disproportionate to the magnitude of the earthquakes, i.e. earthquakes were too small to generate the recorded tsunami waves. Before the 1998 event, it was questionable as to whether a submarine landslide could cause large tsunamis.

Local megatsunamis generated by submarine landslides far exceed the height of the main −more extensive− tectonic tsunami, and reach the coast within a few minutes after the earthquake; both superimpose on the shores near the origin of the earthquake.

3. Methods

The compilation for the megatsunami catalog is initially based on the information on maximum runup heights collected in the Global Historical Tsunami Databases (GHTDs), where, in a first step, the events that meet the requirement imposed on the minimum wave height reached on land were selected.

From this initial list each event was researched for its available references including earliest. The information on megatsunamis has subsequently been cross-checked with data from the documentary sources. Some of the most outstanding events are well documented and described in papers and reports, while others are poorly documented, in some cases on a single publication. Also, a systematic search for “new” events not included in the GHTDs was undertaken. Additional contemporary documentary sources were sought, principally scientific papers and technical reports published or on line.

3.1. Data collection

Initial data were selected from the two existing online Global Historical Tsunami Databases (GHTDs):

• NCEI/WDS Global Historical Tsunami Database [30] supported by the National Geophysical Data Center of the National Oceanic and Atmospheric Administration (NOAA), USA.
• TL/ICMMG Global Historical Tsunami Database [31] supported by the Tsunami Laboratory, Institute of Computational Mathematics and Mathematical Geophysics of Siberian Division of Russian Academy of Sciences, Russia.

Both have been compiled from historical tsunami data mainly from earlier regional catalogs and studies, local catalogs and other old and contemporary scientific sources, individual event reports and studies, and unpublished works and documents, all with different spatial and temporal scope. The two databases are generally consistent in terms of data format and content (about 100 events from each database are not recorded in the other), with some differences in the estimates of maximum water heights (runups), confidence level or degree of validity assigned, or attribution of tsunami sources or causes, mainly for older historical events.

The American and Russian databases collect available and documented information on both the geological processes generating the tsunamis (tsunamigenic events), mostly earthquakes, and on the maximum wave heights observed or measured in all locations where data are available. The detail and veracity of the data depends on the sources and, above all, the date of occurrence. The period of record covers the last 4 centuries. Since the beginning of the 20th century, there has been a notable increase in the number of records and observations, which has been accentuated since the 1960s and especially since the beginning of the 21st century, coinciding with the great tsunami in Indonesia in 2004, from which scientific research and publications have intensified.

The NCEI/WDS database contains information on over 2800 tsunami events (of which 1269 include measured or estimated runup or wave height data, 706 of these with heights >1 m) in the Atlantic, Indian, and Pacific Oceans, and the Mediterranean and Caribbean Seas, from 2000 BC to the present (2023). The TL/ICMMG database includes information on 2700 events (of which 1128 include some wave height or runup data and, 619 with heights >1 m) occurred in the World Ocean, for the same period of time. Both computerized, easily accessible databases can be kept constantly updated.

Most of the tsunamis have been generated by submarine earthquakes (~80%), especially those prior to 1500; the rest are associated with landslides (~9%), volcanic eruptions (4%-5%) and meteorological effects (2%-3%). Tsunamis caused by landslides have been recorded since the 17th century, increasing in number by the 20th century, especially from the 1960s onwards.

Of all historically documented tsunamis of any origin, approximately 1.5% have reached or exceeded 30 m wave height, and only 1% have reached or exceeded 40 m, none of them directly generated by an earthquake.

The main uncertainties when dealing with historical information come from the temporal and spatial biases in their recording, and the validity of the data. The limitations and problems of existing tsunami catalogs and databases are described by Gusiakov [32]; among the problems are those derived from the inaccuracy and incompleteness of the data mainly about old or geographically remote events, and the errors in interpretation of old descriptions. One way of indicating these uncertainties in tsunami databases is the assignment of a validity index, generally ranging from 0 or -1 (false or erroneous) to 4 (definite).

3.2. Literature review

The literature review was carried out with a dual purpose: on the one hand, to verify megatsunami data, mainly runup and source, collected from the GHTDs databases; and on the other hand, to search for megatsunami events not included in the databases.

For these purposes, references cited in the GHDBs have been reviewed, as well as any other information on a specific megatsunami event, even searching, where possible, for its earliest references. Numerous regional and local catalogs, scientific papers, published case studies and reports, books and any other reliable source of information, including available on-line old catalogs and documents with original data, and personal consultations with expert researchers. More than 250 publications between 1888 and 2023 have been reviewed, some of which include references to 17th and 18th century written reports, the oldest in 1675 [33] and ~1750 [34]. Old records are often unclear and incomplete, mostly written long after the event by those who were not eyewitnesses. Occasionally, the dates of older events may not correspond exactly in all bibliographic sources, as there have been some adaptations of the calendars used in ancient sources.

More than a hundred regional or local tsunami catalogs are available since the mid-20th century, and some even earlier. A detailed description of these information sources and their contents and evolution, since the appearance of the first catalogs in the 1930s and 1940s, as well as the development of the two global historical databases since the 1980s and 1990s, is provided by Gusiakov [32].

The website of the Tsunami Laboratory of the ICMMG of Russia include a list with more than 120 published historical tsunami catalogs (http://tsun.sscc.ru/tsu_catalogs.htm). Wiegel [35,36,37] has published a comprehensive list of tsunami information sources worldwide, including all available catalogs up to the time of his publications. More recently, Cheng et al. [38] have published a detailed comparative study of the content of existing global databases.

The first historical tsunami catalog was compiled by N.H. Heck in 1934 [32] who summarized the tsunami data from previous earthquake catalogs. Initially published in French, it was published in English in 1947 [39], undoubtedly linked to the 1946 Aleutian tsunami that caused damage and losses of life at Hilo, Hawaii and further damage on shores of the Pacific Ocean. It contains 270 waves, classified under the terms of "seismic sea wave," "sea wave," "tidal wave," and "tsunami"; wave height data are not included.

Listed below are some of the most comprehensive and representative catalogs of historical tsunamis data and additional information, that include some of the megatsunamis referenced in this study:

• Iida, K., Cox, D.C. & Pararas-Carayannis, G. 1967. Preliminary catalog of tsunamis occurring in the Pacific Ocean [40].
• Soloviev, S.L. & Go, Ch.N. 1974. A catalogue of tsunamis on the western shore of the Pacific Ocean (173-1968) [41].
• Soloviev, S.L. & Go, Ch.N. 1975. A catalogue of tsunamis on the eastern shore of the Pacific Ocean (1513-1968) [42].
• Iida, K. 1984. Catalog of tsunamis in Japan and its neighboring countries [43].
• Lander, J.F. 1996. Tsunamis affecting Alaska 1737-1996 [23].
• Harris, R. & Major, J. 2016. Waves of destruction in the East Indies: The Wichmann catalogue of earthquakes and tsunami in the Indonesian region from 1538 to 1877 [44].

3.3. Data verification

The selected events (all with runup ≥40 m) from the GHTDs were researched in order to obtain as much information as possible about the sources and validity of the data, looking for historical evidence to verify and check the maximum runup provided by the databases, as well as the causes of the events. The main goal has been to check the information with reliable historical data or evidence. In general, for megatsunamis recorded over the last century, the data on the maximum height reached by the waves on land are reliable and supported by evidence. The review process led to rule out some of the initially selected megatsunamis due to lack of evidence or significant uncertainties or doubts, as it was not possible to verify the data, mainly older historical events before the 19th century; but it also led to identify some events that initially did not meet megatsunami criteria here imposed, as well as other megatsunamis occurred in the last decades not included in either GHTDs.

4. Results

The final global catalog of megatsunamis consists of those verified megatsunamis from the group initially collected from the two GHTDs (all with a wave height ≥40 m, regardless of the origin assigned) and those events, not included in the GTDBs, selected after the documentary research carried out. As a result, a total of 37 historical megatsunamis have been compiled (excluding Santorini).

In addition to the maximum wave heights, the exhaustive review has also made it possible to define their causes, especially in the oldest cases for which new documentary and field research has been carried out.

4.1. Megatsunamis included in the GTDBs

The NCEI/WDS and the TL/ICMMG GHTDs include 29 and 27 megatsunami events −i.e. with reported maximum water height ≥40 m−, respectively, for the period 1674-2023, of which 25 coincide in the two databases. The oldest event documented is from 1674 (excluding Santorini, 3,600 BP): a huge wave ~100 m high, the largest ever documented in Indonesia, generated after a violent earthquake, affecting the north coast of the island of Ambon; the earthquake and tsunami caused more than 2,300 casualties. All of the events are classified as definite −the highest level of validity of the actual tsunami occurrence− in the GHTDs, although there are sometimes large differences in the maximum wave height or runup reached by the waves.

The volcanic tsunami of the island of Thera or Santorini, in Greece, around 1600 BC, is included in both databases, although only the NCEI/WDS gives a maximum wave height of 90 m. The historical evidence of a large tsunami caused by the explosion and collapse of the island are indisputable. Wave heights have been estimated at between 40 and 90 m according to geological evidence, although could have exceeded 250 m in height [45,46]. This event has not been included in the global catalog of megatsunamis presented here.

Data on these megatsunamis are presented in Table 1, as of 16 October 2023; the latest recorded event is from 2020. A total of 32 megatsunamis have been compiled. Few megatsunamis (6 events) have reached or exceeded 100 m in heigh. The apparent accuracy of some of the older runup figures is due to the precision used for unit conversions to the metric system, especially for old events.

Most of the megatsunamis in the American database, 22 of the 29 total events, are assigned a landslide origin (6 of them triggered by earthquakes and 3 by volcanic eruptions), 3 are attributed to volcanic eruptions and 4 directly to earthquakes. In the Russian database, 19 are attributed to landslides (5 of them triggered by earthquakes), 2 to volcanic eruptions and 6 to earthquakes.

4.2. Definitive megatsunamis

The review of documentary sources on the megatsunami events initially listed from the GHTDs (31 events not including Santorini) has led to discarding 5 of them due to reasonable doubts, uncertainties or contradictions in the data provided for the maximum height, or runup, reached by the waves. On the other hand, the extensive literature review has also led to the inclusion of some events not listed initially as megatsunamis, and some other significant events, mainly from the mid-1980s, not recorded at all in the GHTDs, thus increasing the catalog of megatsunamis significantly.

As a result, a total of 37 historical megatsunamis have been finally compiled, 26 of which are included in the two GTDBs, while 11 are new events. These are listed by year in Table 2, including their maximum runup height and cause. Figure 2 shows the distribution of megatsunamis by year and runup height.

4.2.1. Verification of old megatsunamis from the GHTDs

The following events have been discarded: 1737 (Russia), 1741 and 1771 (Japan), 1788 and 1880 (Alaska); all but the last occurred during the 18th century. For three of these ancient events, only one of the global databases provides wave heights >40 m, based on oral references in old documents (see Table 1): the earthquake-triggered tsunami of 1737 in the Kuril-Kamchatka region; the tsunami of 1741, caused by a large volcanic landslide on the island of Oshima-Oshima; and the 1880 tsunami by a seismic landslide in Alaska. In all cases, there is a large difference between the wave heights provided by the two GHTDs.

The maximum runup heights, between 63 and 100 m, for 18th century tsunamis attributed to earthquakes (1737, 1771, 1788) are not very reliable, as they come from ancient observations not confirmed by subsequent geological investigations or historical evidence. Besides, such heights have never been observed for tsunamis caused by even very high magnitude earthquakes. Some recent scientific papers review these data, considerably lowering the wave heights, in some cases well below those recorded in the global databases.

The maximum runup of the October 1737 event provided by the Russian global database was based on reported heights of driftwood and trees washed over a cliff, not definitively linked to the tsunami (Lander, 96). Some of the available documentary sources report very different data on the maximum height reached by the waves: 63 m according to written testimonies of the time following an expedition in 1738 [34,47], and 8-10 m according to a study of field data and written sources on this tsunami [48]. No definitive conclusions can be drawn for the 1737 earthquake-triggered tsunami, which also affected Alaska with a maximum wave height of 12-15 m. According to Lander (96), the maximum waves in Kamchatka reached 30 m.

The April 1771 tsunami, known as the Great Meiwa Tsunami ‒with 12,000 deaths‒, reached maximum wave heights of up to 85 m according to some available documentation. Recent publications indicate a more reliable maximum value of around 30-35 m (30 m [49]; 32 m [50]; 30 m [51]; 30 m [52], in any case <35 m; ≤35 m [53]), discarding older data from historical record considered over-estimated, such as the maximum height of 85 m in the southern coast of Ishigaki Island from a historical record just after the disaster [43,54]. Soloviev and Go [41] attribute these differences to confusion in the conversion of units of measure of length at the time.

The source of the 1771 tsunami has been attributed to a M 7.4 earthquake, and alternatively to an earthquake-triggered landslide, based on submarine geological evidence [51]. According to Matsumoto et al. [54], since no historical record of tremor was identified that corresponds to the Great Meiwa Tsunami, it might be reasonable to consider a large-scale underwater landslide as the origin of the tsunami.

The 6 August 1788 tsunami, the earliest recorded tsunami in the Aleutian-Alaskan trench region, Gulf of Alaska, is attributed to a M 8 earthquake in both GHTDs. The maximum runup of 88 m at Unga Island or Sanak Island is very doubtful and not documented [23], coming from oral testimonies of the period, published several decades later [55]. According to [55] and [23], the maximum height reached could be 30 m or “few tens of meters” on Unga and Sanak Islands, and could not be attributed solely to displacement of the seabed by tectonic failure; in fact, old reports cite that the earthquake caused landslides from the mountains and shores [23].

Due to the above considerations, particularly since the maximum wave heights cannot be verified in the absence of reliable historical records, the seismic tsunamis discussed above cannot be considered megatsunamis.

The 26 October 1880 Alaskan tsunami reached a maximum wave height of 60 m and 1.8 m in Sitka, according to the Russian and American GHDBs respectively, the former without citing the source of the data. Both include other tsunamis associated with the same earthquake (estimated M 6.3) with runup not exceeding 2 m; also, the events are classified as "probable", not definitive. According to [23], based on information from newspapers of the time, the cause of the unusual waves was “very local submarine landslide-generated tsunamis", with maximum height of 1.8 m, although the author also makes the following comment: "huge wave ran into bay", without specifying further. According to [42], a “large tidal wave appeared” after the earthquake. This contradictory and inconclusive data, together with the fact that the supposed 60 m wave is located in a bay more than 200 km from where the other smaller tsunamis were recorded, seems to indicate that the runup height is erroneous.

The 1741 tsunami was caused by a landslide on the north flank of the small Oshima-Oshima volcanic island in the Sea of Japan, SW of the island of Hokkaido. Historical records refer to the existence of volcanic activity, but do not record the occurrence of any earthquakes. The landslide, of about 2.4 km³, one of the largest landslides in history, affected the subaerial and submarine flank of the island. The 90 m maximum wave height for the tsunami is not supported by evidence; it appears in the NOAA database and originally comes from the Catalog of tsunamis in Japan and its neighbouring countries [43] where only an oral reference is given. According to reliable historical documents the maximum runup was 13 m, while data providing higher heights are oral and unverified [56]. According to the Russian database and other review catalogs and literature sources, the maximum wave height reached about 9-10 m off the coast of Hokkaido [40,41,56]. Therefore, this event cannot be described as a megatsunami.

After ruling out the 1741 tsunami, there are only 2 documented historical cases of megatsunamis from volcanic island flank landslides: on Mount Mayuyama, 1792, Japan, and very recently on the island of Anak Krakatau, Indonesia, 2018.

The most recent large volcanic island flank landslide, although not causing a megatsunami, occurred on Ritter Island (Papua New Guinea) in 1888, with a volume of ~2.4 km³ similar to Oshima-Oshima, both being the only two large-volume (>1 km³) flank landslides historically recorded on volcanic islands, ranking among the largest historical on Earth. The Ritter landslide generated a tsunami with maximum runup of 15 m along the nearby coastline [57].

4.2.2. New documented megatsunamis not included in the GHTDs

Eleven “new” megatsunamis have been added to the elite group of megatsunamis after a comprehensive review of the corresponding documentary sources (Table 3):

• 7 are not recorded in the GHTDs: 2 in 1936 in Norway; 1985, 2003 and 2018 in China; 2007 in Mexico; and 2020 in Canada.
• 4 are recorded in the GHTDs with runup values <40 m: ~38 m for the 1756 Norwegian, 1896 Japanese and 2007 Canadian events; and runup of 9 and 30 m for the 1946 Canadian tsunami. The first 3 are the only events in the GHTDs with runup >35 and <40m.

The events are listed in Table 3, with the final runup values assigned after review of the corresponding documentary sources, described in the GHMCat. All but two have runup heights above 40 m. For the 2007 tsunami (Canada), the maximum runup height of 38 m is definitive from a post-tsunami field survey and, as it is the only one with runup <40 m, it has been considered for inclusion in the global catalog of megatsunamis. The same criterion has been applied for the “new” event in 2003, in China, with a 39-m runup.

4.3. A new event in Lituya Bay prior to 1786

Lituya Bay, on the northeast shore of the Gulf of Alaska, was officially discovered by La Pérouse in 1786, during his expedition around the world. Cenotaph Island, a large wooded mound at the center of the bay, is named for the 21 members of the La Perouse expedition who drowned after three small boats capsized in rough seas at the shallow entrance to the bay. A monument was erected in their memory on the island.

The bay is currently well known for having experienced the highest wave runup ever recorded (525 m), and for being one of the two places on earth where the largest number of megatsunamis have been recorded since the mid-19th century: at least four megatsunamis during a little more than a century (1853, 1899, 1936 and 1958) have rushed out from the head of the bay; all except the 1899 event are among the ten largest historical megatsunamis in the world, with more than 100 m. In all cases, the megatsunamis destroyed the forested slopes leaving clear markers of the corresponding runup heights reached. Another tsunami, with runup of 20-30 m, occurred around 1874, as deduced from the trimline identified in a photograph of 1894 of the north shore of the bay and observations of the stage of vegetation growth below and above the tsunami trimlines [2]. These old marks left by the waves on the vegetation of the slopes, were largely erased in the 1936 megatsunami, and almost completely in the 1958 event.

The review of historical documents for this study has led to infer the occurrence of another megatsunami prior to 1786 in Lituya Bay [12]: an engraving of the bay made in the summer of that year by the expedition of the frenchman La Pérouse, its official discoverer, clearly shows the effects of huge waves on the slopes of the bay, with the vegetation completely washed away up to a certain height, several tens of meters above sea level (Figure 3) [58]. In the image below there is even another lower trimline, clearly visible on the right-hand slope, which could represent a lower tsunami; and, what is definite evidence of the occurrence of megatsunamis prior to 1786, a large rockslide on the cliff at the bottom of the bay can be observed, in the same area where the earthquake-triggered rockslide of 1958 occurred.

The details of the drawing undoubtedly indicate that both processes must have occurred not long before the arrival of the expedition. Since the Lituya Bay history begins very briefly in 1786 (La Perouse expedition) or 1788 (Russian exploration and settlement) there are no records available to collate the occurrence of a megatsunami in earlier years. However, it can no longer be maintained that the oldest known megatsunami in Lituya is that of 1853.

The high frequency of megatsunamis in the bay (at least five since the mid-18th century), compared to other similar bays in the area, may be partly due to the presence of a large coastal fault, the Fairweather Fault, with very significant seismic activity: the 1958 megatsunami, and most probably the 1899 megatsunami, were caused by rock and ice avalanches triggered by earthquakes [2,23]. The inferred megatsunami prior to 1786 cannot be associated with a significant earthquake as the history of Alaska can be said to begin in 1784, when the first Russian settlements were established, four years before than a Russian ship entered Lituya Bay to claim the territory; in fact, the earliest earthquake and associated tsunamis ‒in July 1788, two years after the La Pérouse visit‒ in the Aleutian-Alaskan trench region were recorded by a Russian missionary in 1940, describing “very strong earthquake” and “terrible floods” [23,55].

5. Discussion

The analysis of the 37 documented historical megatsunamis, most since the beginning of the 19th century, has made it possible not only to verify and compare each of them, but also to analyze their causes as a whole. One of the most remarkable aspects is that, despite the limitations related to the temporal and spatial biases of the records, the vast majority of megatsunamis, and the highest, have been caused by subaerial large landslides.

It is evident that the record of large local tsunamis has grown as scientific and technological knowledge has advanced, mainly from the second half of the 19th century and particularly in the 20th century.

Regarding the historical record of megatsunamis, it should be taken into account that it begins in the 17th century ‒excluding the Santorini event‒ and until the middle of the 19th century, the number of data is very scarce and the information is incomplete and uncertain. Only 3 megatsunamis are recorded during this period. From 1850 onwards, the number of records increases until the 20th century, with 15 events throughout this century, with a gradual increase in the number as time progresses, until the 21st century, when a surge of cases occurs reaching 15 megatsunamis documented during the last 23 years. As might be expected, significant spatial and temporal biases can be inferred from the number, frequency and distribution of the cases, mainly conditioned by the presence of human settlements and the availability of written historical records.

The megatsunamis record over the last 90 years show an increase in number of events, with some of the most significant occurring in glaciated areas, where half of the events of more than 70 m wave height have occurred (see Table 2). It is especially worthy of remarks that the fjords and bays of Alaska and Norway have experienced more than a third of all the world’s recorded megatsunamis over the past 125 years, mainly during the first half of the 20th century. The reasons of this may include the inherent spatial biases in the global historical record, and that the regional concentrations may likely reflect the high level of landslide research in these regions and the availability of documents [118]. But it cannot be dismissed the role of the geological and climatic conditions in the frequency and number of megatsunamis in these areas, conditions that combine to make them prone to large-rockslides-triggered megatsunamis; in contrast, the example of Japan can be cited, where the historical record of tsunamis spans many hundreds of years and just three megatsunamis have been described, or Russia, with only one case in its far east coast in almost 300 years [47].

The cases of Lituya Bay (Alaska) and Lake Lovatnet (Norway), with several historical megatsunamis (at least three times at each location over a period of a little more than a century) have also experienced numerous tsunamis with smaller, but still significant, wave heights, always originating from rockslides or rock avalanches [2,4]. Also in the areas of Vaigat Strait and Karrat Fjord (central West Greenland), where megatsunamis have occurred recently (2000 and 2017 respectively), there have been several other landslide-generated tsunamis in recent decades [119]. These facts indicate the possibility that megatsunamis could occur again in the same or nearby places. In glaciated areas, the combined effects of steep slopes, fractured rock masses, glacial retreat, high precipitation and freeze-thaw cycles play an important role in the occurrence of large rockslides and rock avalanches.

The phenomenon of global warming, which implies a global increase in temperatures, has a particular impact in the colder high latitudes of both hemispheres, especially in the frozen areas, where rapid deglaciation occurs and glaciers are gradually retreating, with a remarkable influence on slope instability processes. Climate warming is likely driving an increase in the frequency of large landslides [109]. After the retreat of glaciers, steep slopes are exposed to erosion and weathering processes as well as changes in the state of in situ stress, modifying the slopes equilibrium conditions, which can lead to large rockslides and avalanches that, falling suddenly into confined deep water bodies such as fjords and bays, may trigger huge megatsunami waves.

The close relationship between rising temperatures and the increasing number of large waves in confined waters in glacial areas vulnerable to slope instability processes seems evident, and attention should be paid in order to prevent and mitigate possible consequences.

6. Conclusions

From the analysis of the 37 megatsunamis catalogued in the GHMCat since the availability of historical accounts or sources from the 17th century onwards, with the first recorded in 1674 (excluding Santorini, 3600 BP), some conclusions or significant questions have been drawn that can contribute to improve the scientific perspective on these extraordinary natural phenomena, as a useful means of prevention and mitigation of their potentially damaging effects. In this regard, the following points can be highlighted in relation to their wave height, origin and occurrence:

• By wave height: the most frequent runup heights ranged between 40 and 60 m, in 50% of the cases. However, 24% of the events have reached heights above 100 m, all generated by subaerial landslides or rock avalanches in confined bodies of water.
• By origin: All the recorded megatsunamis have been originated by landslides, most of them (>80%) large subaerial landslides, rockslides or rock avalanches ‒in some cases triggered by high-magnitude earthquakes‒, and some by volcanic flank landslides associated to large explosive eruptions. Less frequently, the cause has been large submarine landslides (14%), all triggered by very high magnitude earthquakes (M ≥8.6). Subaerial landslide-generated megatsunamis have the largest recorded runups (up to 525 m), but these are always local events which effects are confined to a small coastal strip.

Although for some authors, the exact origin of the megatsunamis associated with major earthquakes (e.g., 1946, 1964, 2004, 2011) remains controversial, there is geological evidence of large submarine landslides triggered by the earthquakes ‒except for the 2004 event‒, as explained in the corresponding descriptions in the GHMCat.

• By geographical distribution and geological context: historical megatsunamis have been recorded in America (~40%), Asia (~33%) and Europe¹ (~27%), where giant waves have only been recorded in Norway ‒except for two in Italy, anthropogenic, and in Iceland‒. The bays and fjords of Alaska and Norway have experienced the largest number of events, having accounted for 32% of all megatsunamis worldwide in the last 350 years; followed by the shores of Indonesia, Japan, Canada and China, in the latter case with 4 megatsunamis all occurring in rivers or reservoirs. Less frequently, large waves have also been recorded in Greenland and Chile, with two events each (three of them in fjords).

Apart from the large number of tsunamis in fjords and confined bays in glaciated areas (40%), giant waves have also occurred in open sea shores (22%), mountain lakes (19%), reservoirs (11%), and rivers (8%). It is worth noting that in the cases of megatsunamis in reservoirs in China and Italy, the landslides have been influenced by human activity, while in all other cases they have been originated by natural geological processes.

• For Australia and New Zealand there are no documented historical megatsunamis, since the history in these regions can be considered to have started towards the end of the 18th century. But this does not mean that such extreme processes have not occurred in the last hundreds or thousands of years, as some geological evidence shows [120,121,122].

Regarding future trends, the possibility of megatsunamis in glaciated areas needs to be considered. There is some evidence to suggest that megatsunamis may become more frequent, possibly as a result of glacial retreat by global warming. In contrast, there is a very low probability of occurrence of megatsunamis associated with large explosive eruptions or volcanic island flank failures, as is the occurrence of local, near-field megatsunamis generated by large magnitude earthquake-triggered submarine landslides.