Tsunami modeling and early warning systems

  • Department of Civil Engineering,
    University of Ottawa,
    Ottawa, Canada
    Email: smurty@hotmail.com

    Natural hazards can be broadly grouped into three types: (A) Permanent, (B) Evanescent and (C) Episodic. Hazards that fit into the permanent category are tides, wind waves, coastal erosion and climate change. Tides are always present in the oceans. Wind waves are also always occurring in the oceans, even though their amplitudes may vary depending upon the strength of the weather systems. Climate change has been happening continuously for the past 280 million years, when the atmosphere first evolved almost in its present form. For permanent hazards such as tides and wind waves, we do not speak of early warning systems, rather routine prediction is the norm. Evanescent hazards are slow and gradual and have no clearly identifiable beginning and no clearly defined ending. Examples are Drought and sea level rise.
    For these types of hazards, it is difficult to set up early warning systems. Rather, we can monitor various parameters to infer what is happening. On the other hand, episodic hazards have a clear beginning and clear ending. Examples are: Earthquakes, tsunamis, volcanic eruptions, land slides, submarine landslides, tropical cyclones ( hurricanes ), extra-tropical cyclones (winter storms), storm surges, freak waves, tornadoes, snow avalanches, river floods etc. The prediction of earthquakes has not progressed to a stage, where any reliable prediction can be made in time and geographically at this time. For most of the other types of episodic hazards, early warning systems are in existence for some time now with varying degrees of success.
    Tsunamis occur mainly from large and shallow under-ocean earthquakes. However, tsunami generation is also possible from under-water volcanic eruptions, large scale submarine landslides, nuclear explosions and large scale chemical explosions conducted in the water and even from asteroid strikes on the ocean. Majority of tsunamis occur in the Pacific Ocean, followed by the Indian Ocean, because these two oceans contain converging tectonic plates. On the otherhand, the Atlantic Ocean has mostly diverging tectonic plates, which give rise to ocean floor spreading. The Few earthquakes and tsunamis in the Atlantic Ocean are mostly at the margins. All the nations around the Pacific rim have colloboratively developped an international tsunami warning system based in Honolulu and Palmer, Alaska, which is in existence since 1948 after the disastrous Aleutian earthquake tsunami of 1946. This system is administered by the IOC( Inter-Governmental Oceanographic Commission ) of UNESCO in Paris. After the very devastating tsunami in the Indian Ocean on 26th December 2004, early warning systems have been put in place for the Indian and Atlantic Oceans also. Quarter wave resonance amplification is among the most common phenomenon in tsunami dynamics. Helmholtz resonance can also contribute to significant tsunami activity in harbors, In the 2004 Indian Ocean tsunami, the sea levels around the Andaman and Nicobar Islands stayed quite high for several days after the main tsunami has passed. This is due to energy trapping around islands in terms of OFC ( Oscillations of the first class ) and OSC ( Oscillations of the second class )
    A typical tsunami warning system consists of a set of seismographs to register the earthquake, ocean bottom pressure sensors to record the deep water signature of the tsunami, a series of tide gauges to record the tsunami, software in terms of numerical models, a 24 by 7 warning center ( such as INCOIS in Hyderabad ) and real time communication systems in place. It should be noted that at present, until the earthquake happens, there is no tsunami prediction. In Japan, sometimes there is only few minutes of elapsed time interval between the occurrence of the earthquake and the arrival of the tsunami on the nearby coastline. In such instances, the earthquake itself serves as the tsunami warning. Also at present there is no organized systematic worldwide tsunami prediction from under-ocean volcanic eruptions, subamarine landslides. Whereas tsunamis from major volcanic eruptions could be large ( as in the 1883 Krakatoa eruption ) tsunamis from land slides will be quite local.
    If we define success as never missing the prediction of a tsunami in real time, then the warning centers are generally 100% successful. However, when it comes to accurate prediction of the coastal inundation ( the horizontal extent, number of waves, which wave is the highest, the strong currents etc ) their success is somewhat muted. Generally speaking, while the warning centers identify the tsunami and provide ETA’s ( expected times of arrival ) at various coastal locations, they leave it to individual nations to deal with the coastal inundation aspects.
    Tsunami travel times depend only upon the ocean depths ( the speed of travel is proportional to the square root of the water depth multiplied by the acceleration due to gravity ). Hence travel times can be pre-computed and observations show that they are generally accurate to within plus or minus one minute for each hour of travel.
    The Pacific Ocean being much larger, it takes about a day for the tsunami to travel from say Chile to Japan. On the otherhand in the much smaller Indian Ocean, with in a few hours ( much less than 24 ) the tsunami arrives at most coastal locations. Because of this difference, in the pacific ocean, reflected waves from far off coastlines do not contribute much to the tsunami in progress, whereas in the Indian Ocean, the contribution from the reflected waves must be factored in. In the Pacific Ocean, the highest wave is among the 3 rd to 5th waves usually, while in the Indian Ocean, it is the 2nd wave that is generally the highest.
    There are some fundamental mathematical problems in the modeling of coastal inundation from events such as storm surges and tsunamis. These problems stem from the multiple-connected nature (from a hydrodynamic sense) of the coastal oceans due to the presence of islands, rivers, estuaries, back waters, lagoons, sand bars etc. The problem is ill-posed from a mathematical point of view, because either there are too many boundary conditions or too few, but never the correct number required. Finally while coastal inundation can be computed very well in one dimension (perpendicular to the coast), the attempts at two-dimensional computation are not completely successful, because of the presence of human-made structures on the coast, as well as vegetation, sand dunes, hills etc.
    The two most commonly used numerical models for tsunamis at present are the TUNAMI N2 and the MOST ( Method of splitting tsunamis ). While these models compute coastal inundation in terms of wet and dry cells, the details of wave breaking are missing in these two models. The next generation models include the SPH ( Smoothed Particle Hydrodynamics ) which can simulate wave breaking with reasonable success. In the 1960′s and 1970′s numerical models used finite-difference( f-d ) approach with rectangular grids. Starting in the 1980′s finite-element ( f-e ) models with irregular triangular grids provided much better resolution of the coastal geometry. An ideal model will be an SPH with fractal grids.
    As far as Kerala is concerned, the two relevant tsunami-genic sources are the Sumatra area and the Makran subduction zone in the Arabian Sea, off the coasts of Iran and Pakistan. Tsunamis from the Andaman& Nicobar regions would have little effect on the Kerala coast. Tsunamis from the Sumatra region can diffract south of Sri Lanka and propagate northward thus affecting the Kerala coast.
    Some unsolved problems include: accurate computation of coastal inundation and a proper model to account for the initial withdrawl ( recession ) of the ocean in certain locations and in some events ( but not everywhere and not in all the events ).
    While the seal level rise of less than one meter over a century as predicted by the IPCC is important and should be dealt with, if we can design coastal structures that can withstand the much greater impact from storm surges and tsunamis ( for example, in the 2004 Indian Ocean tsunami, near Banda Ache in Indonesia, at one location, the sea level rose by about 50 meters in just a few minutes ), then we have automatically protected ourselves from sea level rise