Rupturing the sea floor: How the Sumatran Earthquake created such a large tsunami

LA-UR-05-0544
Andrew Newman, EES-9
1/19/05

On December 26, 2004 a magnitude 9.0 subduction mega-thrust earthquake, the largest in the past 40 years, occurred offshore the heavily populated northern region of the Indonesian island of Sumatra.
The earthquake was created by plate tectonics, driving the eastern portion of the Indian plate at an angle into and under the small Burma plate at a rate of about 2 inches per year. As the plates collide their interface locks up building strain energy to be later released in frequent shallow crustal strike-slip earthquakes and occasional catastrophic mega-thrust earthquakes along the interface, like the event of last month.

The earthquake was unique, not just for its size, but for the large and devastating transoceanic tsunami that it produced, which was responsible for the loss of more than 210,000 human lives, causing the most fatalities for a tsunami in human history. It is likely that this earthquake was especially effective at producing the tsunami because the event ruptured the ocean floor, something that is rather uncommon for most mega-thrust earthquakes.

One of the last such events identified as rupturing in a similar way was a magnitude 7.8 earthquake off the west coast of Nicaragua in 1992.
This event is generally described as a classic example of an anomalous "tsunami earthquake" because its efficiency at producing a tsunami much larger than predicted by its initial magnitude. One of the most notable features of such events is their unexpectedly slow rupture, for their size. In the case of the Nicaraguan event, the rupture duration was nearly 10 times longer, ~120 seconds, than expected for an earthquake of its magnitude. Though it is currently not clear that the Sumatran event had such an anomalously slow rupture it does share the unique character of the 1992 "tsunami earthquake" in that it too had ruptured the ocean floor.

What caused the tsunami?
Though submarine landslides, volcanic eruptions and meteor impacts are all capable of generating large tsunamis, the most common cause is from shallow oceanic earthquakes. A tsunami is created when an earthquake shifts the sea floor, creating a displacement in the water column that dissipates in gravity waves. Because the Sumatran earthquake ruptured to the sea floor, it was especially effective at generating a tsunami because it could offset oceanic water waves by several feet, rather than several inches as many large deeper earthquakes would create.

The height of the tsunami waves first shrink as they spread out in all directions, but subsequently raise dramatically as they near shore, up to 100 feet, in order to conserve energy as speed slows due to the shallowing water. By developing techniques that identify whether an earthquake ruptures the ocean floor, we can improve the current state-of-the-art in tsunami warning systems.

Determining the risk of Tsunami Earthquakes
How can we assess in near-real-time, whether a large subduction zone earthquake has ruptured the ocean floor, thus having the potential for effectively generating a larger than expected tsunami? Two approaches are useful here. First, it is beneficial to determine whether the near-trench region of a given subduction zone was previously locked, allowing strain to accumulate before rupturing in a "tsunami earthquake". With this earthquake and recent results from other "tsunami earthquakes", it appears likely that the shallow subduction interface becomes strongly locked late in the seismic cycle, and is only relaxed once it ruptures in a large subduction thrust event. After which, the region becomes weak, slowly strengthening during the interseismic before the next large event. The Middle America Trench west of Nicaragua and Costa Rica, appear to behave in this fashion, with the Nicaraguan portion being weak since it recently ruptured in a large tsunami earthquake, and the northern Costa Rican portion exhibiting strong coupling in the portion of the interface updip of current seismic activity. Though the mechanism for locking in the shallow regions is poorly understood, it appears important to understanding the slow shallow rupture of these "tsunami earthquakes".

By learning more about how this region ruptured in the Sumatran earthquake, and whether it was previously locked, will yield a much improved understanding of the underlying physics of these events. Though answering this question on a global scale difficult at present (no reliable ocean-bottom geodetic sensors currently exist) it would be helpful for assessing the potential for future events. For the Sumatran earthquake I intend to examine available geodetic and seismic data before and after the Sumatran earthquake to determine the likelihood of whether the up-dip region was locked prior to the main event, or continued to rupture due to momentum. Because there is little data available (only a few points in the southern portion of the event), results from this component are expected to be poorly constrained, but should give at least some insight into the state of up-dip portion of the subduction interface there.

How can we improve real-time warning?
The second approach to near-real time assessment of near trench rupture in great earthquakes is to develop an algorithm to detect, in near-real time, the portion of the rupture that slows down while rupturing to the trench. This can be done in real-time with data immediately available from a global network of broadband seismometers. Because slow rupture is not efficient at producing high-frequency compressional body waves I have previously established a method to determine in real-time whether an entire earthquake was deficient in high-frequency energy, thus slow and more likely to create a large tsunami. This methodology correctly identified the three known modern slow "tsunami earthquakes" (Java 1994, Peru 1996, and Nicaragua 1992) by being an order of magnitude more deficient in P-wave energy.

This technique, now utilized by the Pacific Tsunami Warning Center (PTWC), has also identified the Sumatran earthquake as being deficient in P-wave energy as compared to Moment release, such that this event appears comparatively even slower than the 1992 Tsunami event, the slowest previous event. Unfortunately, because of problems with losing part of the P-wave train in the S-arrivals of such a large event, the real-time detection was unreliable for such a large event. Instead, I plan to develop a technique to calculate the temporal distribution of energy to moment ratio throughout the earthquake rupture to determine, eventually in real-time, whether a portion of a rupture appears slow and possibly rupturing near to the trench.

Potential risk of tsunami disasters at home
The west coast of the United States, Hawaii, as well as numerous geostratigically located US and allied military bases are susceptible to tsunamis generated by large subduction zone earthquakes from anywhere in the pacific "Ring-of-fire". Additionally, the Cascadian subduction zone, running from Northern California to southern Canada, has recently been determined to rupture in similar magnitude events every 300 to 500 years, with the last occurring in 1700 A.D.

The real-time identification of tsunamis, such as the one created by the Sumatran Earthquake, are of particular interest to DOE and LANL, because these events have the capability of causing massive losses in human life and properties, thus potentially destabilizing regions. A real-time warning of such events can help to minimize losses, particularly in life.

Tectonic makeup of the Indian Ocean and South East Asia. The Indian Ocean is broken into several tectonic plates, all moving in different directions, and causing earthquakes at their boundaries. In the Northern portion of Sumatra and the Andaman and Nicobar Islands. the Indian plate collides obliqely into the Burma plate at a speed of about 6cm/year.

Earthquakes in the vecinity of the Dec. 26, 2004 Sumatran Earthquake. 1 week of aftershocks, yellow circles and darkly shaded focal mechanisms, likely represent the full rupture area of the main shock. These events seem to show that the main shock ruptured to the trench/ocean floorin the south, and early in the rupture. before migrating northward where it remains deeper. Interestingly, it appears that the near-trench region of the rupture was seismically quiescent in at least 30 years preceding the earthquake (few red older earthquakes).

Cross section of a Subduction zone. As a dense oceanic plate collides into lighter continental crust, it subducting under it. Much of the interface between these two plates locks up and as the oceanic plate continues to push into the continental plate, energy builds up to be released in large earthquakes. For Tsunami earthquakes, the uppermost portion of this interface slips casuing potentially large Tsunamis.

A Sumatran-type Earthquake in the US? Along the Northern California, Oregon, Washington and southwestern canada, the Juan de Fuca palte subducts beneath North America. This interface is approximate in size to that of the Sumatran Earthquake, as shown by the overlain outline of the Sumatran earthquake (laterally flipped and rotated) over the subduction region. The region has been shown to have similar, magnitude 9, earthquakes about every 300 - 500 years in the recent past, with the last such event occuring in 1700 A.D

EES Home

RESOURCES

Informational Flyers