 |
Inside this Issue |
|||
| Global Model of Indian Ocean Tsunami | ||||
Dr. Zygmunt Kowalik,
Institute of Marine Science/UAF School of Fisheries and Ocean Sciences
|
||||
Story by Lorien Nettleton
The December 26, 2004 Indian Ocean tsunami shocked and riveted a worldwide audience. As it rang like a bell on every continent the tsunami was sensed on every shoreline from Chile to Iceland in the form of increased wave-height. With the swells of the sea came swells of information, buoyed by instant worldwide communication that surfaced in the wake of the single worst tsunami in history. In the Pacific Ocean, there has been significant study and prediction of tsunami for years. The Pacific Warning Center, in Ewa Beach, Hawaii, monitors seismic and sea-level activity for the entire Pacific region, and most nations that border the Pacific Ocean maintain seismic, tide, communication and dissemination facilities. The West Coast and Alaska Tsunami Warning Center, in Palmer, Alaska, monitors seismic events that may generate tsunami that threaten North American shorelines. But following the tragic events of 2004, it became evident that adequate tsunami prediction and warning mechanisms were not in place for other world coastlines. In response to the revelation of inadequate tsunami preparation on a world scale, the United Nations proposed creating an Indian Ocean Tsunami Warning System as an initial step towards the International Early Warning Program.
Dr. Zygmunt Kowalik, Professor of Physical Oceanography at the Institute of Marine Science of the UAF School of Fisheries and Ocean Sciences, is one of many researchers able to use those new data in his ongoing studies. For the last twenty years, Kowalik has investigated tsunami physics around Alaska, applying computational modeling techniques to predict tsunami generation, propagation and run-up. Current research, in collaboration with William Knight and Paul Whitmore of the West Coast and Alaska Tsunami Warning Center (WC/ATWC) and the National Oceanic and Atmospheric Administration (NOAA), Juan Horrillo from the UAF Institute of Marine Science and Tom Logan from the Arctic Region Supercomputing Center (ARSC), has led Kowalik to create the first global tsunami model to display and study the patterns of propagation of a tsunami as it travels to every shoreline on Earth. To test the accuracy of the tsunami simulation, the model was applied to the Indian Ocean tsunami of December 26, 2004. A grid point was assigned to the World Ocean from 80 south latitude to 69 north latitude with a spatial resolution of one minute. This resulted in a massive 220 million grid points. Using Whitmore’s source function that showed the sea floor’s tectonic activity and the subsequent vertical displacement of water, Kowalik implemented an energy flux to investigate energy transfer from the tsunami source across the Indian Ocean to the Atlantic and Pacific Oceans. The sheer size of the computational domain of a global model necessitated the development of a parallel version of the code to run on Klondike, ARSC’s Cray X1™ supercomputer. Tom Logan re-wrote Kowalik’s code into Co-Array Fortran for the study. The process enabled a task of several weeks to be completed in less than ten hours on 60 multi-streaming processors, or just under half the machine’s maximum capability. The global distribution of maximum amplitude – the waves carrying the most energy – was tracked and then visualized, allowing the researchers to see how well the model predicted the propagation patterns of the maximum amplitude, and what the estimated amplitude for each run-up would be. As the tsunami had already occurred, the researchers were able to compare the model results to recorded tidal information in order to evaluate the model’s performance. The computational data revealed a number of interesting features of how the world’s oceans reacted to the dramatic shift of the India/Burma plates. The primary force of the tsunami spread both east and west, with high-energy wavelengths spreading past South Africa towards the Atlantic Ocean. Although source directivity mostly pushed the wave energy towards southern Africa, a strong signal was also directed toward Antarctica, where the energy was redirected along the South Pacific Ridge towards southern Central America. Through the virtue of the model’s visualization, constructed with the help of Roger Edberg, Visualization Specialist at ARSC, Kowalik was able to see the ways in which the slower signals tended to carry more energy and follow the shallower areas of ocean, typically along ridges (Fig. 1). Additionally, the researchers noticed that the tsunami travel time, an important parameter for tsunami warnings, does not follow the conventional wisdom that the first wave will be the biggest wave, as higher amplitude waves took longer to reach each shoreline, but carried greater power, while the initial waves traveled very rapidly, but did not have a significant amplitude. As the method presently used in tsunami warning systems for calculating tsunami travel time predicts the arrival of the first wave, Kowalik’s finding shows that predictions of tsunami arrival time may be inaccurate. The researchers were also able to observe a focusing of the wave’s energy at narrow passages, such as that between Australia and Antarctica, as well as between Africa and South America. The funneling of the waves through such narrow passages concentrated the propagating energy and redefined it back from chaotic motion into articulated waves (Fig. 2).
Making Modeling Available WorldwideIn order to better distribute tsunami modeling for a wide range of research and early warning inquiries, a collaborative effort between several tsunami modeling centers has been initiated to share predictive abilities. A website portal that will allow registered users to run models to make predictions on future tsunami impact is being developed by ARSC, the Geophysical Institute and the Institute of Marine Science at the University of Alaska Fairbanks, the Northwest Alliance for Computational Science and Engineering at Oregon State University and the School of Civil and Environmental Engineering at Cornell University. Once implemented, the portal will allow researchers to input location-specific data through a graphical front-end and run one of two pre-compiled numerical tsunami models on any region, with definable parameters provided through the database. Once the researcher submits their run, the job files will be sent to ARSC where they will be processed on the appropriate computing platform, and the results will be sent back to the researcher and incorporated into the portal’s database.Challenges RemainAs is often the case with predictive modeling, there is always opportunity to improve the model. There are some discrepancies between Kowalik’s modeled tsunami and the observed event in terms of amplitude and times. Due to the lack of high-resolution bathymetry — or ocean-floor topography — for much of the Indian Ocean, the model was run on a one-minute grid, or two-kilometer resolution. The researchers are confident that a higher resolution grid would allow for a closer alignment between the modeled amplitudes and times and the observed data. When it comes to predicting which seaside structures are at risk of being destroyed by various tsunamis, a two-kilometer resolution will do little to inform researchers about the specific destructive risks; mapping a neighborhood’s susceptibility to total destruction would require a grid of a few meters to achieve the results. More accurate models are dependant on physical information such as high-resolution bathymetry, but in many cases, security restraints prevent ocean-floor information from being accessible. This global tsunami model marks the beginning of furthering the shared study of a massively destructive and powerful force of nature. Researchers will continue working towards perfecting a comprehensive tsunami model, which, in addition to its existing capabilities, will introduce physical information and processes that occur in the real ocean as the information becomes available. In the meantime, the West Coast and Alaska Tsunami Warning Center will continue contributing to predictive scenarios to inform people in coastal areas of Alaska of tsunami risks using the higher resolution models available for the region – all of which will require vast computing and storage resources such as those found at ARSC. For more information, visit the UAF School of Fisheries and Ocean Sciences web site: www.sfos.uaf.edu/directory/faculty/kowalik. |
||||
| Challenges Index | ||||
