Tsunami Early Warning Systems

Michael E. Blackford
IOC-USNWS/International Tsunami Information Center, 737 Bishop Street Suite 2200, Honolulu, Hawaii 96813-3213 USA.
Tel. : +1-808-532-64 23, Fax: + 1-808-532-55 76,
E-mail: michael.blackford@noaa.gov

While a relationship between local, felt earthquakes and tsunamis was realised in ancient times in many societies, it was much later in human history that it was realised that destructive tsunamis can propagate great distances, far beyond the range of human perceptibility of the causative earthquake. In societies that kept written records of such disastrous events there would appear the occasional tsunami that could not be associated with a local earthquake or even with an earthquake that occurred in another province with which the society had contact. The cause of these rogue waves was the source of much speculation among the ancient thinkers and philosophers learning of, or actually experiencing, such waves. By the middle of the nineteenth century, however, most of the world’s coastlines had been explored by societies that kept written records, or had been colonised by such societies. This brought on a vast increase in the communications among advanced societies situated on the coastlines of the world. The realisation that an earthquake on one side of a great sea basin, such as the Pacific Ocean, could cause a destructive tsunami on the opposite side of the basin was becoming firmly established.

In the latter half of the nineteenth century modern seismology had its birth with the development of seismographs that were capable of detecting the seismic waves of earthquakes located at great distances from the instrument, even on the opposite side of the earth. Various types of seismic waves spread out from the earthquake source at speeds of a few to several kilometres per second. The first few decades of modern seismology was involved with the identification and characterisation of the basic types of seismic waves observed on the seismograms from these early instruments. Three important types of waves were identified. These were the primary, or P-waves, the secondary, or S-waves, and the surface waves. Measuring the differences between the arrival times of these various waves allows one to the determine distance between the earthquake and the recording seismograph. This had important implications for the development of techniques to forecast the arrival of tsunamis from distant sources.

During the first half of the twentieth century researchers in Hawaii, which was a relatively frequent recipient of teletsunamis, or tsunamis from distant sources, began to make forecasts of the arrival time of the teletsunamis using the seismic wave data from their newly installed seismographs. They would determine the distance of the earthquake from the seismograph and, if that distance happened to be the same as the distance to a known source of historical teletsunamis, they would forecast a tsunami for Hawaii. These early forecasts were often in error because the earthquake in a tsunami-producing area this time did not generate a tsunami. Another source of error was that the earthquake was actually located in an area that does not produce tsunamis, which was at the same distance as one that does. The early forecasters were also not obligated to make these forecasts, so they missed tsunamis that were generated and travelled to Hawaii during the night while they were sleeping. Later in the first half of the century electromagnetic seismographs, with their photographic recorders, because of the greater sensitivity, began to replace to the earlier mechanical seismographs with visible smoked paper drum recorders. This made it even more difficult to forecast tsunamis because these new seismographs had to be located in darkrooms and their seismograms were typically changed only once each day.

In 1946 an earthquake occurred off Unimak Island at the eastern end of the Aleutian Islands in Alaska and about four and one-half hours later a devastating tsunami struck Hawaii. There was vast destruction at several places in the Islands and nearly 200 lives were lost. Shortly after the occurrence of this disaster the U.S. Government’s Coast & Geodetic Survey was criticised for not warning the populace of the impending tsunami. The Survey operated a seismograph on Oahu at its Honolulu Magnetic Observatory but it was not prepared to issue warnings for such events. The Survey subsequently developed an alarm system to notify the staff of a distant earthquake at any time of the day and a stable electronic amplifier that would allow visible recording of the earthquakes at sensitivities similar to the photographic recorders. They also developed a tsunami travel time chart that would allow them to quickly determine the arrival time of a tsunami in Hawaii from an earthquake source anywhere in the Pacific. Of most importance is that they developed a communications plan that involved receiving timely seismic data from other Coast & Geodetic Observatories that would allow them to more precisely locate the alarm earthquake. Of equal importance, they established a network of tide station observers who would send them timely data on the character of the tsunami as it passed by their stations. This allowed them, in many cases, to ascertain the severity of the tsunami before it was due in Hawaii and to cancel watches before actual evacuations took place. This warning system was formally established fifty years ago in August 1948.

The system grew rapidly in the years after 1948, expanding the numbers of both the seismic stations and the tide stations reporting to the Honolulu Observatory. The area of responsibility also expanded to include not only Hawaii but also the other States of the U. S. bordering the Pacific. The fifties and early sixties were times of several teletsunamis that were destructive in the United States as well as elsewhere in the Pacific. The Honolulu Observatory successfully issued warnings for all the teletsunamis that were destructive in the United States. There were also several false warnings issued during these years, mainly due to difficulties in interpreting the mareographic observations sent in from the tide stations.

In 1960 a very great earthquake occurred in southern Chile that generated a tsunami which was very destructive at many different locations throughout the Pacific including Japan where about 200 lives were lost. The waves did much damage in Hawaii and in spite of a warning over 60 lives were lost. After this disaster that affected many nations in the Pacific, the Honolulu Observatory began to receive requests from many of these nations to receive copies of the messages being sent to the participants in the U.S. tsunami warning system. Discussions began in 1965 and finally in 1968 the Tsunami Warning System in the Pacific was established with the Honolulu Observatory as its operational centre. Shortly thereafter, because the tsunami warning responsibilities of the Observatory began to overshadow the Observatory’s magnetic observation responsibilities to a very great degree, its name was changed to the Pacific Tsunami Warning Center (PTWC). Since the establishment of the Tsunami Warning System in the Pacific, however, there has not been a Pacific-wide tsunami that matches the destructiveness of those in the fifties and sixties.

During the seventies the PTWC acquired its first in-house minicomputer and the process of earthquake location and message generation became somewhat automated. The time between the alarm and the time of message dissemination was cut in half to about one hour. There were still problems interpreting the mareographic information and this resulted in many anxious episodes of cancelling warnings just before evacuations were to have occurred. In the eighties the PTWC finally began to receive mareographic data via geostationary satellite from many stations near the tsunami producing zones of the Pacific. This data, together with improved procedures for determining the estimated size of earthquakes, adopted in the nineties, has allowed the PTWC to make objective assessments of the severity of the tsunami potential more quickly and to cancel warnings more quickly as well.

The success of the Tsunami Warning System in the Pacific in its role as a provider of assurance that a particular earthquake is not going to generate a Pacific-wide tsunami has been very good over the thirty years of its existence. This is true for participants in the system who are some distance from the tsunami-generating earthquake. For those closer to the earthquake this is not necessarily so. The system places those who are within three hours tsunami travel time of the earthquake in a warning. Often those who are more than two hours travel time from the tsunami source experience no destructive tsunami. So, in a sense, the warning is false. On the other hand, because of the nature of the data available to the PTWC, it usually cannot issue a warning until one half of an hour to one hour after the earthquake. This is often way too late for those affected by the local tsunami. The answer to this problem is establishment of regional tsunami warning centres, the paramount goal of the Tsunami Warning System in the Pacific at the beginning of the next millennia.