THE 21st CENTURY
from a paper presented at the Second International Tsunami Workshop
on the Technical Aspects of Tsunami Warning Systems, Tsunami
Analysis, Preparedness, Observation and Instrumentation. Novosibirsk,
USSR, 4-5 August 1989.
published in Intergovernmental Oceanographic Commission-UNESCO.
Workshop Report No. 58
Copyright © 1989. All Rights
Tsunamis recur with frequencies
of short, medium, long, or super-long cycles. Most are generated
by earthquakes, although volcanic eruptions, rockfalls, landslides
and other impulsive sources can also generate them. Each region
of the world appears to have its own cycle and pattern in generating
tsunamis. The majority of events are usually small. However some
tsunamis may be highly destructive locally or at great distances
across an entire ocean.
In certain very active
tectonic regions, particularly in regions of subduction, large
earthquakes occur with frequency. In such regions, tsunamis usually
follow similar recurrence patterns as those of the large earthquakes.
However, in certain other tectonic regions where large earthquakes
may be infrequent, tsunami recurrence cycles are irregular, intermittent
and of varying periodicity. Events may occur decades or even
hundreds of years apart. For events that occur hundreds of years
apart, there is no way of establishing periodicity other than
through the interpretation of geological and paleontological
data. Making accurate predictions as to when exactly destructive
tsunamis will occur in the future, is not possible. However,
given the time window of a century - and having the benefit of
historical information and the knowledge of ongoing tectonic
processes on our planet - we can certainly indicate with relative
accuracy the locations where small and great tsunamis can be
expected in the future. Such information may useful for purposes
of tsunami risk assessment, disaster awareness and for overall
The ability to forecast
or predict tsunamis with any degree of accuracy is limited. Long
term tsunami prediction is based primarily on statistical methods
of earthquake prediction in identifuing seismic gaps. Medium-term
prediction is also based on statistical recurrence frequency,
but it is also of limited value. Overall statistic probabilities
are of limited use for forecasting purposes - except for events
that follow yearly or seasonal cycles - as wheather phenomena
do not always occur in seismic-gap regions that have been clearly
identified as such. For example, the 1975 tsunami in the Philippines,
the 1979 tsunami in Colombia, and the 1983 tsunami in the Sea
of Japan, were all local events that occurred unexpectedly. No
one had suspected seismic gaps or build-up of stresses. These
regions of the world were not actively monitored. No tsunami
warnings were issued for these events that could have been of
any value to the threatened population. Thus many lives, were
lost. However a review of historical events and an understanding
of tectonic processes could have eliminated the surprise element
when these events occurred. There had been no destructive tsunamis
in these areas in recent times, but there had been several in
the distant past. History of disasters repeats itself. A warning
could have been issued based on the earthquake parameters alone
or the knowledge of past events.
There are many other
limitations in making prediction as to where and when destructive
tsunamis may strike. However, a real time prediction of a tsunami
can be made on the basis of seismic parameters after an earthquake
has occurred in an area that had been impacted in the past. Such
short-term, real-time predictions can be made and tsunami warnings
can be issued for coastal regions that are about one to one and
a half hour away from the tsunami generating source. However,
it is extremely difficult to warn, in a timely fashion, coastal
areas that are very close to the tsunami generating source. The
tsunami may only take a few minutes to reach the nearest shore.
In recent times, refinements have been made to cut down the time
of evaluating such events and in disseminating local warnings.
Regional Tsunami Warning Centers have been established that can
now issue warnings withing a few minutes after tsunami generation
for their immediate areas of their jurisdiction. However, the
evaluation and the warnings are often based on nothing more than
vague earthquake parameters and the region's known tsunamigenic
The following analysis
is simply an overview of where destructive local regional and
ocean-wide tsunamis can be reasonably expected in the 21st Century
and subsequently. This overview is based on historical information
and known tectonic processes and interactions. No attempt is
made to explain the mechanisms of the interactions.
Difficulties and Limitations
in Determining When and Where Future Tsunamis May Strike?
Although most of the
large tsunamigenic earthquakes occur in geological cycles which
may be considered regular or rhythmical, conditions that are
exogenous or random often cause time variations in their recurrence.
This is what makes their prediction difficult. Because of the
scarcity of historical data for large tsunamigenic earthquakes,
only statistical methods of extremes have been used to obtain
statistical recurrence values. These are of limited value because
the confidence limits are low. Similarly, because the epicenters
of most of the large tsunamigenic earthquakes are in the ocean,
it is often difficult to measure other geophysical and geochemical
precursory phenomena that would give clues as to event recurrence.
However, based on currently available information, each known
tsunamigenic region of the world can be assigned its own recurrence
frequency or recurrence cycle - if studied and documented extensively.
Even though each destructive tsunami has its own unique source
pattern and its own time and impact variables, certain general
principles seem to apply to most of them, which will be simply
referred to here by the author as abiding by the proposed "law
of tsunami disaster cycles".
A Statistical Approach
To Tsunami Event Prediction - Inadequacy of Statistical Methods
and of the Seismic Tsunami Gap Theory.
As mentioned, a statistical
approach - such as the seismic gap theory - may not be valid
for tsunami predictions. In this author's opinion, the 30-year
time interval - which is the criterion for establishing a seismic
gap - may be too short of an interval for predictions of destructive
tsunamis, particularly for certain geographical regions of the
world. For example, numerous large earthquakes have occurred
along the Azores Gibraltar Fault Zone. However, there has been
no recurrence of the great Lisbon earthquake and tsunami of 1755.
What is proposed here
is a new theory which is named the "tsunami gap theory."
The criteria for the development of the tsunami gap designations
will not be uniform, but will differ in accordance to the tectonic
characteristics of each region. According to this approach, the
time path of variables of each future tsunami could be established,
or at least estimated statistically for each potential tsunamigenic
area, by using both historical and geophysical data. The same
historical and geophysical criteria could be projected to form
predictive schemes relating to other seismic tsunami source parameters,
and for expected near and far-field tsunami effects - the latter
further determined by analysis of numerous factors at each terminal
point. For example, for vulnerable locations such as Hilo, Hawaii,
there is more than adequate information on historical tsunamis
and their runups, on bottom bathymetry, on geomorphological configuration
of the coastaline, on land topography and on frictional parameters.
This information is already incorporated in the evaluation process
in predicting the height, terminal velocity and potential runup
of any tsunami from any direction of approach - regardless of
the source region. Such specific tsunami risk analysis would
be required for other critical coastlines vulnerable to tsunamis
of both local and distant origin.
This type of historical,
statistical, geophysical, oceanographic and engineering analysis,
if properly executed, can establish tsunami gap regions and can
lead to better forecasts of future destructive events and their
impact to near or distant shores. The centerpiece of the proposed
approach is named "historical tsunami determinism."
It is much more than collecting and cataloguing historical tsunami
information. The basic premise of the proposed process is that
tsunamis follow certain patterns, although these are not uniform
for all regions of the world. A simple qualitative historical
information database is not sufficient. For each tsunamigenic
region, the database must include accurate quantitative measurements
and cataloguing of all meaningful tsunami disaster parameters,
including seismic parameters, and geophysical precursory phenomena.
Furthermore the database
This may be somewhat
difficult to do for all regions of the world, but it should be
definetely and we may have to establish a new methodology for
doing it. Also it will require the development of an extensive
tsunami data base which should include, not only historical information,
but other extensive data and catalogues as for example - data
on focal mechanisms, power spectra, seismic moments of previous
seismic events on specific broadband seismic signatures. Developing
such historical and geophysical data bases would require a great
deal of effort. However, we could standardize the development
of such data bases, and we could agree conceptually on the methodology
and the development of such historical tsunami determinism, and
we could integrate this knowledge into real-time operational
assessment of the tsunami risk for warning purposes.
Forecasting The Tsunamis of the 21st Century
There should not have
been surprise by the occurrence of the 1975 Philippine tsunami
or the 1979 Colombian tsunami, or the 1983 Sea of Japan tsunami.
There is historical and geologic precedents for all these events,
and their occurrence should have been expected. Similarly, source
and impact regions of future tsunamis could be forecasted for
the remainder of this century and for the 21st Century, by utilizing
these proposed extensive historical and geophysical databases.
Forecasting of future
tsunamis on the medium or longer term may be possible. Just as
with earthquakes, prolonged periods of quiescence of a region
that has been historically tsunamigenic (using the tsunami gap
hypothesis and criteria) may be signaling the eventual occurrence
of another destructive tsunami. Many such areas exist presently
in the Pacific and elsewhere, and the location of these regions
can be identified. If we develop the proper criteria of historic
tsunami determinism, we can also assign a time frame for tsunami
recurrence. This can be done not only for areas of established
seismic gaps but for areas where seismic gaps have not been identified,
or for which a different recurrence time frequency applies.
The data exists, but
its proper analysis has not been undertaken. A total of 482 tsunamis
have been reported in the 20th Century alone, with at least 133
having a runup greater than 1.5 meters. We know that Japan, the
West Coast of South America, Alaska, the Aleutian Islands, Kamchatka,
and the Kuril Islands are potential tsunami generating areas.
We know that these are the boundaries of major tectonic plates.
But what about all the other subplates of the inland seas that
have produced also the big destructive tsunamis and for which
we have not established seismic gaps?
Where in the Pacific
Ocean can we expect the big tsunamis in the 21st Century other
than the areas mentioned? Let us be more specific. There are
many tsunamigenic regions that have shown high density of seismic
energy release and where large future tsunamis can be expected.
For example, one such area is a segment of the Peruvian coastal
region between 8.5° S and 14° S. This is a region of
extremely high seismic energy release and site of large but infrequent
historical tsunamis. Other parts of the South American seismic
belt are tsunami gap regions and these regions in the 17th, 18th
and 19th Centuries produced several destructive tsunamis, destroying
such towns in Chile as Arica, Antofagasta and Valparaiso. There
is also a great potential for another destructive tsunami on
the Pacific side of Colombia, in the vicinity of the State of
Narino. The west coast of Mexico can be expected to experience
larger tsunamis. Large destructive tsunamis can be expected again
in the Moro Gulf in the Philippines, in the Celebes and Sulu
Sea, in the Java Sea and elsewhere in the South West Pacific.
A lot can be said about the potential
of such future tsunamis, but to make reasonable predictions about
the time and place of future events, the criteria of the proposed
historical determinism must be utilized. These criteria have
not been as yet developed and I propose that we develop them.
As a starting point, we need to develop a uniform and standardized
program of tsunami, seismic and geologic data collection. A wealth
of such data already exists but this data is not properly organized,
is not uniformly collected, and of course it is not readily available.
Therefore, standards must be established for the collection of
such data and a tsunami data base must be organized on a regional
basis initially, and shared on a global scale at a later time.
Finally, the methodology of historical tsunami determinism as
I described earlier for real-time operational use must be established.
Links to other
1963-2007 George Pararas-Carayannis / all rights reserved / Information
on this site is for viewing and personal information only - protected
by copyright. Any unauthorized use or reproduction of material
from this site without written permission is prohibited.
Now available from Amazon, Barnes
and Noble and other major bookstores. A signed by the author
copy can be also ordered by contacting directly by email Aston Forbes
Miscellaneous Non-technical Writings