Abstract. Since tsunamis are relatively rare they are often overlooked as a natural hazard until a disastrous one occurs. The occurrence of tsunamis at Hawaii in 1946, and Alaska in 1964, resulted in warning systems being established. A history of 56 reported tsunamis in the Caribbean since 1530 has been compiled showing similar fatality levels in the last 150 years in the Caribbean as in Hawaii, Alaska, and the west coast of the United States combined. Many historical reports of tsunamis in the Caribbean are confused with reports of hurricanes. An unknown number of these reports may represent observations of storm surges. However, there is evidence that earthquakes and tsunamis may occur preferentially during hurricane seasons as well. Given the great increase in the coastal population in recent decades, the potential for a disastrous tsunami is significant. In the past 500 years, the Caribbean has experienced teletsunamis, tectonic tsunamis, landslide tsunamis, and volcanic tsunamis. There is a growing awareness of the hazard in the area as evident in the participation of 130 people from many of the islands' universities and Civil Defense organizations and scientists from the United States and Russia in a tsunami workshop on June 11-13, 1997, in Mayaguez, Puerto Rico. Recommendations from that workshop included establishment of a warning system before the next tsunami disaster occurs. Such a warning system should be based on the unique tsunami hazard of each area as seen in a comprehensive history of past events. Education of the public to recognize tsunami indicators, such as felt earthquakes while near the coast or sudden recession of the sea coast should be considered as a natural potential tsunami warning. The time available before tsunami arrival in the Caribbean may not be long enough, in many cases, for a communicated warning.
Key words: tsunami, Caribbean, tsunami history, mitigation, natural hazards
Tsunamis are not the major natural hazard in the Caribbean but they have the potential to produce a major disaster which can be mitigated with proper preparation. This history has reports on 56 tsunamis affecting the Caribbean in historic times. These have affected 22 countries and administrative areas including northern South America and Central America; these are listed in Table 1. The record for the last hundred years lists 20 tsunamis or about one every 5 years. It contains information about tsunamis from several types of sources. Source mechanisms include tsunamis from remote sources (teletsunamis), landslide tsunamis, volcanic tsunamis, and tectonic tsunamis associated with trenches and with transform faults. Figure 1 shows that the locations affected by tsunamis include nearly the whole of the Caribbean region. This history was compiled mostly from published sources available internationally. Additional events and effects would likely be found by searching locally available sources.
Since destructive tsunamis are relatively rare it is easy to overlook the hazard until a tsunamirelated disaster occurs. The Pacific tsunami warning system was established largely as a response to the destructive tsunamis in Hawaii (1946) and Alaska (1964). The state of tsunami preparedness in the Caribbean today is similar to that in the Pacific prior to the establishment of the Pacific tsunami warning system. A repeat of the 1755 Lisbon earthquake and tsunami or one from a similar location could produce a tsunami with waves as high as 6 meters in the Caribbean and with a travel time to the Caribbean of 7 to 8 hours. Without a warning system little or nothing could be done to mitigate the disaster. An axiom in tsunami studies is that tsunamis in the future can be expected to be at least as large as those that have occurred in the past. Given the increase in the Caribbean, but may not have been looked for on marigrams.
The map of effects in the Caribbean shows gaps in reporting. Was there really no wave observed at San Juan, Puerto Rico, which is on the Atlantic side, the Virgin Islands, which are near the maximum effect areas, and Hispaniola although waves were reportedly seen at Cuba to the northwest? The reported values probably were made by ships in the area on returning to Europe, not all of which became known to the scientific community. Curiously, none of the Caribbean tsunamis have generated damaging effects at teletsunarni distances but some have been recorded on the East coast of the United States.
B. Landslide Tsunamis. Tsunamis generated by landslides are usually, but not always triggered by earthquakes. They can have devastating local effects but the effects are limited to a small area. The 1958 tsunami in Lituya Bay, Alaska, sent a surge of water across the bay and washed trees away to a height of 1720 feet. A true tsunami went down the bay with heights of 30 meters, causing 2 fatalities. It was not observed outside of the bay. As the source is near shore, the warning time for landslide tsunamis may be only a few minutes. Education of the public to seek high ground immediately if they feel an earthquake or notice a withdrawal of the sea is probably the only effective mitigation measure. Local landslide tsunamis in Alaska have frequently generated waves of 30 meters or more and the landslide tsunamis generated by the great 1964 Prince William Sound earthquake, more than 20 separate events, caused 80 percent of the fatalities due to that tsunami. In the Caribbean it appears that most of the tsunamis local to Jamaica are due to landslides, since the major offshore faults are strike slip in that region with little or no vertical displacement. The 1692 Port Royal earthquake and tsunami caused the destruction of the city and about 2000 fatalities. Most of these were probably due to the tsunami given the subsidence of the city and the large waves. That event occurred over 300 years ago and we do not know what the return period may be but in tectonic regions similar to the Caribbean, return time for large earthquakes is on the order of 200 to 500 years. Hence, a repeat of this occurrence is possible at any time.
C. Volcanic Tsunamis. Volcanoes can create tsunamis in a number of ways including explosions, collapse of the craters, and landslides such as the minor 1690 event on Nevis or St. Kitts. The Krakatau, Indonesia, eruption in 1883, caused 30,000 regional fatalities and produced air waves which were recorded as tsunamis in Hawaii, California, Alaska, England, Japan, and Australia Recently a reference was found in the US. Monthly Weather Review of August 1883, that stated that at St. Thomas "A tidal wave occurred here on August 27th. The water receded from the shore three times." The reporter did not mention the correlation with the Krakatau explosion. The Caribbean is nearly at the antipodes of Krakatau and, if the wave was directly observable at St. Thomas, it probably was observable throughout the islands. This kind of wave is, of course, not hazardous but of scientific interest. There was a hurricane north of the island at this time which may have contributed to this observation.
There are reports of submarine debris in the vicinity of Dominica indicating a large scale volcanic deposit dating to about 28,000 years before present (Sigurdsson, et al., 1980). This explosive eruption produced major pyroclastic flows that advanced into the Granada basin, west of Dominica and flowed along the ocean floor, and probably along the ocean surface for a distance of up to 100 km from the source. There is presently concern about the currently active volcanoes of Soufriere and Kick-'em-Jenny (Sigurdsson, personal communication). A tsunami has not been reported with the recent activity, but this means not a destructive or a reported observed tsunami as there are apparently no marigraphs in the area or they were not checked for minor tsunamis. Marigrams could give useful information on the travel times and the relative amplitudes to be expected. Volcanism in the region has recently tended to occur in approximately 10- to 11-year cycles. For example, Kick-'em-Jenny has shown eruptive activity in 1943, 1953, 1965-1966, 1977, 1987, and 1997. This periodicity is similar to that found for world-wide volcanic activity in the statistical study of Stothers(1989). While tsunamis are not quite so regular, there may be periods of time when tsunamis are more likely than during others.
D. Tectonic Tsunamis. Tectonic tsunamis are caused by the thrusting of the oceanic plate into the trench or uplift or drop of a portion of the sea bottom. This type of tsunami would be expected to be important in the Caribbean where there is an active subduction zone at the eastern margin where the Atlantic plate subducts under the Leeward and Windward Islands. Throughout the northern boundary there are numerous strike-slip faults. In Puerto Rico, on October 11, 1918, a magnitude 7.5 earthquake in the Mona Passage created a tsunami which killed at least 34 people and destroyed 300 huts. Current research suggests that the fatalities may have been as high as 100 (Prof. Aurelio Mercado, personal communication). At Point Agujereada several hundred coconut palms were uprooted and eight people drowned. The wave was 5.5 to 6 meters high. At Mayaguez water entered the lower floors of buildings near the water front and destroyed a few native huts (Figs. 4 and 5, photo of houses destroyed). Figure 6 shows the distribution of heights from this event. The source was probably due to thrusting into the trench near the Brownson deep (Ried and Taber, 1919). Figure 7a shows the cross-section of earthquake hypocenters in the region near Puerto Rico, as determined from the Puerto Rico seismic network (1986-1997). While diffuse, the distribution shows the slab subducting steeply at an angle between 60 and 80 degrees. The tsunami efficiency of subduction at this angle is small (Whiteside, et al., in preparation). Figure 7b shows a similar cross-section along the Windward-Leeward Island arc showing a similar angle of descent for the Atlantic plate in this region. A magnitude 7.5 earthquake in the Virgin Island Trough in 1867 killed at least 16 people on St. Thomas and Santa Cruz Islands. The 2.4 meter waves over topped the dock which today may have hundreds of cruise ship passengers shopping on them. Figure 8 shows an artist' s sketch of the La Plata Mail Steamer in Charlotte Amalie harbor from a contemporary magazine. The smaller boat was a coal barge and was destroyed along with most of the crew. The United States was in the process of getting local approval for the purchase of the islands but the damage to two of its war ships and death of four crewmen interrupted the action for 50 years. Figure 9 shows the tsunami travel times from the source through the Caribbean Basin. Waves as high as 3 meters were reported from as far away as Grenada with a warning time of 95 minutes. Maps such as these can be readily made for all ports giving the travel times for any event in the Caribbean to each port.
E. Tsunami Effects. Tsunamis cause damage in a number of ways. While large breaking waves are rare, the force of the waves can destroy buildings, piers, bridges, and other structures. Even relatively small waves can cause strong currents which in San Francisco and Los Angeles have caused millions of dollars in damage, principally by fishing boats and yachts breaking free and colliding with each other and harbor structures. Damage can also be caused through battering by water-carried debris such as logs, boats, autos, etc. The retreating waves can scour the support for bridges, piers, breakwaters, etc., and cause their failure. Chemical spills and fires caused by ruptured storage tanks are also common. Waves can travel long distances up rivers as bores. It is important to include search and rescue in emergency plans as often rescues at sea are possible. The history of the 1867 tsunami~at St. Thomas has many accounts of people being rescued by ships which survived the tsunami.
Reports of some events are apparently incomplete such as the 1831 event which had effects observed in Trinidad and St. Christopher, but not by locations in between. Similarly, the 1842 tsunami was reported as destructive waves at Haiti, and 3.1 m waves in the Virgin Islands, but while the earthquake did damage in P'uerto Rico, no tsunami is reported from there. It was also reported from Grenada, Bequia Island, and Guadeloupe.
F. Tsunami History. The preparation of a thorough history of tsunami occurrences and effects is important in understanding the local nature of the hazard and designing the most effective plan for mitigation. In Alaska, following the 1964 disaster a warning system was set up modeled on the successful Pacific Tsunami Warning System which is designed primarily for teletsunamis. Alaska does not have a history of destructive teletsunamis. Communicated warning systems are very important for teletsunamis for otherwise a disastrous wave can arise without warning. Locally generated tectonic tsunamis such as Alaska has along its Aleutian Trench are accompanied by a natural warning in the strong earthquake shaking. People need to be able to recognize it as a tsunami warning. The evacuation time may be twenty to forty minutes depending on the distance to the trench. In the Prince William Sound area the principal hazard is from landslide tsunamis associated with the glacial terrain. Here, the warning time may be only a few minutes. People should evacuate coastal areas immediately on feeling an earthquake or noticing unusual movement of the water. Education may be the most effective defense for these tsunamis To improve the histories all available information must be sought. Where are marigraphs being operated and where have the been operated in the past? Where are the records stored? What newspapers have been published in the past and when? What sources of historical data are available such as historical societies, museums, libraries, university history departments, Government and church archives, etc.? Given the long and still continuing colonial history, data may be available from European archives, ship logs, diaries, etc. This can best be sought by dedicated local people working together with neighboring islands.
In the Caribbean, there is often confusion in historical reports between a true tsunami and local effects caused by hurricanes. We believe the tsunami catalog is probably contaminated with a number of reported tsunamis which were actually sea waves caused by hurricanes. Figure 10a shows the number of tsunamis reported per month in the Caribbean. During the hurricane season (August through November) 30 tsunamis have been reported, however during the non-hurricane season only 25 have been reported. If hurricanes and reported tsunamis were independent and the expected number of tsunamis were determined from those occurring during non-hurricane season, approximately three tsunamis would be expected per month, a total of about 36 for the catalog. The difference of expected tsunamis during the hurricane season (12) vs. the actual number observed (30) suggests that 2.5 times as many tsunamis than expected are being reported from August through November. The excess number of reported tsunamis (18) could represent incorrect reports or a triggering effect on tsunamigenic earthquakes from hurricanes.
Volcanic eruptions also occur preferentially during hurricane season. Figure 10b, shows the distribution of reported volcanic activity in the Caribbean (Simkin, et al., 1981; Smithsonian Institution, 1981-1997). As with tsunami distribution there is increased eruptive activity during hurricane season, especially during the months of September and October. This suggests triggering effects from the atmospheric changes associated with the low pressure from the hurricane. Volcanoes are known to show periodic tidal behavior. For example, Soufriere Hills has shown a well-defined 12-hour eruptive behavior during the current eruption (Smithsonian Institution, 1997) and eruptions at Kilauea have been tied to tidal periods (Wood, 1917).
In Guam typhoons appear to have a triggering effect on earthquakes (Whiteside, et al., in preparation). Figure 11 shows a similar increase in events within 4 days of hurricanes in the Caribbean as was found in Guam with typhoons. In both cases all earthquakes within 10 days and 1000 km were included. Figure 12 illustrates this triggering effect. In October 1990, Hurricane Klaus sideswiped the West Indies arc for four days with barometric pressures dropping to 985 Mb. There are no other reported instances of a hurricane moving along the island arc for this many days. Hurricanes with similar paths, but faster speed occurred in 1979 - Hurricane Frederic, 1966 -Hurricane Faith, and 1960 - Hurricane Donna (Neumann, et al., 1988). Each of these was followed by earthquakes near Mb 5.0 within 200 hen of 18N 63W within 10 days. With Hurricane Klaus, two earthquakes (Mb 4.5, 5.0) occurred within 12 hours of the maximum storm and within 50 km of the eye on Oct. 7. As Klaus moved to the west, additional earthquakes occurred in conjunction with the hurricane on Oct7, at 15:00 UT, Oct 10, Oct 11, and Oct 9 in westward progression. Along the storm path earthquakes of magnitude 5.9 and 5.4 occurred later on Oct 18 and Oct 25. This example illustrates the triggering of earthquakes by a specific hurricane, but many other examples show this is not uncommon in the Caribbean (Dunbar and Whiteside, 1994). It therefore seems likely that the excess of tsunamis during hurricane season over expected numbers is not completely due to missreporting, but may be in part due to tsunamigenic earthquakes triggered by passing hurricanes. The mechanism of this triggering is probably related to atmospheric loading or winds (Whiteside, et al., in preparation). A tsunami warning system interconnected with the current hurricane warning system would probably be adequate for the region.
1. The history is adequate to show that the Caribbean has the potential for disastrous tsunamis. While it is easier to establish a mitigation program after a major disaster the Caribbean has the opportunity to prepare for and mitigate the coming disaster in advance.
2. Tsunami hazard mitigation needs a warning system, emergency response planning, education of officials and the public, coordinated among the numerous political divisions in the area, and perhaps land use planning, engineering, and insurance programs. The involvement of the UNESCO Intergovernmental Oceanographic Commission and its IOCARIBE is very important, for planning, coordination, and other assistance.
3. The warning and mitigation systems should be based on a thorough understanding of the nature of the hazard including past experiences, tectonics, modeling, and other forms of research for each region.
4. The tsunami hazard for the Caribbean is probably greater at this time than it has been in the past due to economic development and increases in population and tourism. There may be specific factors such as hurricanes or volcanic periodicities which increase the probabilities of tsunamis at particular times.
TOP OF PAGE